feat(admin): add profiling options for debugging high memory/CPU usage (#8923 )

* feat(admin): add profiling options for debugging high memory/CPU usage Add -debug, -debug.port, -cpuprofile, and -memprofile flags to the admin command, matching the profiling support already available in master, volume, and other server commands. This enables investigation of resource usage issues like #8919. * refactor(admin): move profiling flags into AdminOptions struct Move cpuprofile and memprofile flags from global variables into the AdminOptions struct and init() function for consistency with other flags. * fix(debug): bind pprof server to localhost only and document profiling flags StartDebugServer was binding to all interfaces (0.0.0.0), exposing runtime profiling data to the network. Restrict to 127.0.0.1 since this is a development/debugging tool. Also add a "Debugging and Profiling" section to the admin command's help text documenting the new flags.
fix(volume-rust): fix volume balance between Go and Rust servers (#8915 )
2026-10-11 09:05:50 +00:00 · 2026-04-04 10:05:19 -07:00 · 2026-04-04 09:13:23 -07:00 · 2026-04-03 20:01:28 -07:00 · 2026-04-03 19:57:30 -07:00 · 2026-04-03 18:37:54 -07:00
373 changed files with 6726 additions and 47784 deletions
@@ -0,0 +1 @@
+{"sessionId":"d6574c47-eafc-4a94-9dce-f9ffea22b53c","pid":10111,"acquiredAt":1775248373916}
@@ -39,7 +39,80 @@ concurrency:
  cancel-in-progress: false

 jobs:
+
+  # ── Pre-build Rust volume server binaries natively ──────────────────
+  # Cross-compiles for amd64 and arm64 without QEMU, turning a 5-hour
+  # emulated cargo build into ~15 minutes of native compilation.
+  build-rust-binaries:
+    runs-on: ubuntu-22.04
+    strategy:
+      matrix:
+        include:
+          - target: x86_64-unknown-linux-musl
+            arch: amd64
+          - target: aarch64-unknown-linux-musl
+            arch: arm64
+            cross: true
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v6
+
+      - name: Install protobuf compiler
+        run: sudo apt-get update && sudo apt-get install -y protobuf-compiler
+
+      - name: Install Rust toolchain
+        uses: dtolnay/rust-toolchain@stable
+        with:
+          targets: ${{ matrix.target }}
+
+      - name: Install musl tools (amd64)
+        if: ${{ !matrix.cross }}
+        run: sudo apt-get install -y musl-tools
+
+      - name: Install cross-compilation tools (arm64)
+        if: matrix.cross
+        run: |
+          sudo apt-get install -y gcc-aarch64-linux-gnu
+          echo "CARGO_TARGET_AARCH64_UNKNOWN_LINUX_MUSL_LINKER=aarch64-linux-gnu-gcc" >> "$GITHUB_ENV"
+
+      - name: Cache cargo registry and target
+        uses: actions/cache@v5
+        with:
+          path: |
+            ~/.cargo/registry
+            ~/.cargo/git
+            seaweed-volume/target
+          key: rust-docker-${{ matrix.target }}-${{ hashFiles('seaweed-volume/Cargo.lock') }}
+          restore-keys: |
+            rust-docker-${{ matrix.target }}-
+
+      - name: Build large-disk variant
+        env:
+          SEAWEEDFS_COMMIT: ${{ github.sha }}
+        run: |
+          cd seaweed-volume
+          cargo build --release --target ${{ matrix.target }}
+          cp target/${{ matrix.target }}/release/weed-volume ../weed-volume-large-disk-${{ matrix.arch }}
+
+      - name: Build normal variant
+        env:
+          SEAWEEDFS_COMMIT: ${{ github.sha }}
+        run: |
+          cd seaweed-volume
+          cargo build --release --target ${{ matrix.target }} --no-default-features
+          cp target/${{ matrix.target }}/release/weed-volume ../weed-volume-normal-${{ matrix.arch }}
+
+      - name: Upload artifacts
+        uses: actions/upload-artifact@v4
+        with:
+          name: rust-volume-${{ matrix.arch }}
+          path: |
+            weed-volume-large-disk-${{ matrix.arch }}
+            weed-volume-normal-${{ matrix.arch }}
+
+  # ── Build Docker containers ─────────────────────────────────────────
  build:
+    needs: [build-rust-binaries]
    runs-on: ubuntu-latest
    strategy:
      # Build sequentially to avoid rate limits
@@ -52,20 +125,23 @@ jobs:
            dockerfile: ./docker/Dockerfile.go_build
            build_args: ""
            tag_suffix: ""
-            
-          # Large disk - multi-arch  
+            rust_variant: normal
+
+          # Large disk - multi-arch
          - variant: large_disk
            platforms: linux/amd64,linux/arm64,linux/arm/v7,linux/386
            dockerfile: ./docker/Dockerfile.go_build
            build_args: TAGS=5BytesOffset
            tag_suffix: _large_disk
-            
+            rust_variant: large-disk
+
          # Full tags - multi-arch
          - variant: full
            platforms: linux/amd64,linux/arm64
            dockerfile: ./docker/Dockerfile.go_build
            build_args: TAGS=elastic,gocdk,rclone,sqlite,tarantool,tikv,ydb
            tag_suffix: _full
+            rust_variant: normal

          # Large disk + full tags - multi-arch
          - variant: large_disk_full
@@ -73,19 +149,42 @@ jobs:
            dockerfile: ./docker/Dockerfile.go_build
            build_args: TAGS=5BytesOffset,elastic,gocdk,rclone,sqlite,tarantool,tikv,ydb
            tag_suffix: _large_disk_full
-            
+            rust_variant: large-disk
+
          # RocksDB large disk - amd64 only
          - variant: rocksdb
            platforms: linux/amd64
            dockerfile: ./docker/Dockerfile.rocksdb_large
            build_args: ""
            tag_suffix: _large_disk_rocksdb
+            rust_variant: large-disk
            
    steps:
      - name: Checkout
        if: github.event_name != 'workflow_dispatch' || github.event.inputs.variant == 'all' || github.event.inputs.variant == matrix.variant
        uses: actions/checkout@v6
-        
+
+      - name: Download pre-built Rust binaries
+        if: github.event_name != 'workflow_dispatch' || github.event.inputs.variant == 'all' || github.event.inputs.variant == matrix.variant
+        uses: actions/download-artifact@v4
+        with:
+          pattern: rust-volume-*
+          merge-multiple: true
+          path: ./rust-bins
+
+      - name: Place Rust binaries in Docker context
+        if: github.event_name != 'workflow_dispatch' || github.event.inputs.variant == 'all' || github.event.inputs.variant == matrix.variant
+        run: |
+          mkdir -p docker/weed-volume-prebuilt
+          for arch in amd64 arm64; do
+            src="./rust-bins/weed-volume-${{ matrix.rust_variant }}-${arch}"
+            if [ -f "$src" ]; then
+              cp "$src" "docker/weed-volume-prebuilt/weed-volume-${arch}"
+              echo "Placed pre-built Rust binary for ${arch}"
+            fi
+          done
+          ls -la docker/weed-volume-prebuilt/
+
      - name: Free Disk Space
        if: github.event_name != 'workflow_dispatch' || github.event.inputs.variant == 'all' || github.event.inputs.variant == matrix.variant
        run: |
@@ -93,7 +93,7 @@ jobs:
      - name: Build Go weed binary
        run: |
          cd weed
-          go build -o weed .
+          go build -tags 5BytesOffset -o weed .
          chmod +x weed
          ./weed version

@@ -169,7 +169,7 @@ jobs:
      - name: Build Go weed binary
        run: |
          cd weed
-          go build -o weed .
+          go build -tags 5BytesOffset -o weed .
          chmod +x weed
          ./weed version

@@ -39,8 +39,16 @@ jobs:
      - name: Install cross-compilation tools
        if: matrix.cross
        run: |
-          sudo apt-get install -y gcc-aarch64-linux-gnu
+          sudo dpkg --add-architecture arm64
+          sudo sed -i 's/^deb /deb [arch=amd64] /' /etc/apt/sources.list
+          echo "deb [arch=arm64] http://ports.ubuntu.com/ jammy main restricted universe multiverse" | sudo tee /etc/apt/sources.list.d/arm64.list
+          echo "deb [arch=arm64] http://ports.ubuntu.com/ jammy-updates main restricted universe multiverse" | sudo tee -a /etc/apt/sources.list.d/arm64.list
+          sudo apt-get update
+          sudo apt-get install -y gcc-aarch64-linux-gnu libssl-dev:arm64
          echo "CARGO_TARGET_AARCH64_UNKNOWN_LINUX_GNU_LINKER=aarch64-linux-gnu-gcc" >> "$GITHUB_ENV"
+          echo "OPENSSL_DIR=/usr" >> "$GITHUB_ENV"
+          echo "OPENSSL_INCLUDE_DIR=/usr/include" >> "$GITHUB_ENV"
+          echo "OPENSSL_LIB_DIR=/usr/lib/aarch64-linux-gnu" >> "$GITHUB_ENV"

      - name: Cache cargo registry and target
        uses: actions/cache@v5
@@ -80,6 +88,7 @@ jobs:
          rm weed-volume-normal

      - name: Upload release assets
+        if: startsWith(github.ref, 'refs/tags/')
        uses: softprops/action-gh-release@v2
        with:
          files: |
@@ -88,6 +97,15 @@ jobs:
        env:
          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}

+      - name: Upload artifacts
+        if: ${{ !startsWith(github.ref, 'refs/tags/') }}
+        uses: actions/upload-artifact@v4
+        with:
+          name: rust-volume-${{ matrix.asset_suffix }}
+          path: |
+            weed-volume_large_disk_${{ matrix.asset_suffix }}.tar.gz
+            weed-volume_${{ matrix.asset_suffix }}.tar.gz
+
  build-rust-volume-darwin:
    permissions:
      contents: write
@@ -147,6 +165,7 @@ jobs:
          rm weed-volume-normal

      - name: Upload release assets
+        if: startsWith(github.ref, 'refs/tags/')
        uses: softprops/action-gh-release@v2
        with:
          files: |
@@ -155,6 +174,15 @@ jobs:
        env:
          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}

+      - name: Upload artifacts
+        if: ${{ !startsWith(github.ref, 'refs/tags/') }}
+        uses: actions/upload-artifact@v4
+        with:
+          name: rust-volume-${{ matrix.asset_suffix }}
+          path: |
+            weed-volume_large_disk_${{ matrix.asset_suffix }}.tar.gz
+            weed-volume_${{ matrix.asset_suffix }}.tar.gz
+
  build-rust-volume-windows:
    permissions:
      contents: write
@@ -206,6 +234,7 @@ jobs:
          rm weed-volume-normal.exe

      - name: Upload release assets
+        if: startsWith(github.ref, 'refs/tags/')
        uses: softprops/action-gh-release@v2
        with:
          files: |
@@ -213,3 +242,12 @@ jobs:
            weed-volume_windows_amd64.zip
        env:
          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+
+      - name: Upload artifacts
+        if: ${{ !startsWith(github.ref, 'refs/tags/') }}
+        uses: actions/upload-artifact@v4
+        with:
+          name: rust-volume-windows_amd64
+          path: |
+            weed-volume_large_disk_windows_amd64.zip
+            weed-volume_windows_amd64.zip
@@ -0,0 +1,5 @@
+{
+  "setup": [],
+  "teardown": [],
+  "run": []
+}
@@ -16,15 +16,21 @@ RUN cd /go/src/github.com/seaweedfs/seaweedfs/weed \
  && export LDFLAGS="-X github.com/seaweedfs/seaweedfs/weed/util/version.COMMIT=$(git rev-parse --short HEAD)" \
  && CGO_ENABLED=0 go install -tags "$TAGS" -ldflags "-extldflags -static ${LDFLAGS}"

-# Rust volume server builder. Alpine packages avoid depending on the
-# upstream rust:alpine manifest list, which no longer includes linux/386.
+# Rust volume server: use pre-built binary from CI when available (placed in
+# weed-volume-prebuilt/ by the build-rust-binaries job), otherwise compile
+# from source.  Pre-building avoids a multi-hour QEMU-emulated cargo build
+# for non-native architectures.
 FROM alpine:3.23 as rust_builder
 ARG TARGETARCH
+ARG TAGS
+COPY weed-volume-prebuilt/ /prebuilt/
 COPY --from=builder /go/src/github.com/seaweedfs/seaweedfs/seaweed-volume /build/seaweed-volume
 COPY --from=builder /go/src/github.com/seaweedfs/seaweedfs/weed /build/weed
 WORKDIR /build/seaweed-volume
-ARG TAGS
-RUN if [ "$TARGETARCH" = "amd64" ] || [ "$TARGETARCH" = "arm64" ]; then \
+RUN if [ -f "/prebuilt/weed-volume-${TARGETARCH}" ]; then \
+      echo "Using pre-built Rust binary for ${TARGETARCH}" && \
+      cp "/prebuilt/weed-volume-${TARGETARCH}" /weed-volume; \
+    elif [ "$TARGETARCH" = "amd64" ] || [ "$TARGETARCH" = "arm64" ]; then \
      apk add --no-cache musl-dev openssl-dev protobuf-dev git rust cargo; \
      if [ "$TAGS" = "5BytesOffset" ]; then \
        cargo build --release; \
@@ -51,7 +57,7 @@ COPY --from=builder /go/src/github.com/seaweedfs/seaweedfs/docker/entrypoint.sh

 # Install dependencies and create non-root user
 RUN apk upgrade --no-cache zlib && \
-    apk add --no-cache fuse curl su-exec && \
+    apk add --no-cache fuse curl su-exec libgcc && \
    addgroup -g 1000 seaweed && \
    adduser -D -u 1000 -G seaweed seaweed

@@ -0,0 +1,53 @@
+{
+  "identities": [
+    {
+      "name": "admin",
+      "credentials": [
+        {
+          "accessKey": "AKIAIOSFODNN7EXAMPLE",
+          "secretKey": "wJalrXUtnFEMI/K7MDENG/bPxRfiCYEXAMPLEKEY"
+        }
+      ],
+      "actions": [
+        "Admin",
+        "Read",
+        "List",
+        "Tagging",
+        "Write"
+      ]
+    },
+    {
+      "name": "steward",
+      "credentials": [
+        {
+          "accessKey": "steward-key",
+          "secretKey": "steward-secret"
+        }
+      ],
+      "actions": [
+        "Read",
+        "List",
+        "Write"
+      ]
+    },
+    {
+      "name": "le001",
+      "credentials": [
+        {
+          "accessKey": "le001-key",
+          "secretKey": "le001-secret"
+        }
+      ],
+      "actions": [
+        "Read",
+        "List"
+      ]
+    },
+    {
+      "name": "anonymous",
+      "actions": [
+        "Read"
+      ]
+    }
+  ]
+}
@@ -371,7 +371,7 @@ require (
 	github.com/geoffgarside/ber v1.2.0 // indirect
 	github.com/go-chi/chi/v5 v5.2.5 // indirect
 	github.com/go-darwin/apfs v0.0.0-20211011131704-f84b94dbf348 // indirect
-	github.com/go-jose/go-jose/v4 v4.1.3 // indirect
+	github.com/go-jose/go-jose/v4 v4.1.4 // indirect
 	github.com/go-logr/logr v1.4.3 // indirect
 	github.com/go-logr/stdr v1.2.2 // indirect
 	github.com/go-ole/go-ole v1.3.0 // indirect
@@ -1076,8 +1076,8 @@ github.com/go-git/go-billy/v5 v5.6.2/go.mod h1:rcFC2rAsp/erv7CMz9GczHcuD0D32fWzH
 github.com/go-gl/glfw v0.0.0-20190409004039-e6da0acd62b1/go.mod h1:vR7hzQXu2zJy9AVAgeJqvqgH9Q5CA+iKCZ2gyEVpxRU=
 github.com/go-gl/glfw/v3.3/glfw v0.0.0-20191125211704-12ad95a8df72/go.mod h1:tQ2UAYgL5IevRw8kRxooKSPJfGvJ9fJQFa0TUsXzTg8=
 github.com/go-gl/glfw/v3.3/glfw v0.0.0-20200222043503-6f7a984d4dc4/go.mod h1:tQ2UAYgL5IevRw8kRxooKSPJfGvJ9fJQFa0TUsXzTg8=
-github.com/go-jose/go-jose/v4 v4.1.3 h1:CVLmWDhDVRa6Mi/IgCgaopNosCaHz7zrMeF9MlZRkrs=
-github.com/go-jose/go-jose/v4 v4.1.3/go.mod h1:x4oUasVrzR7071A4TnHLGSPpNOm2a21K9Kf04k1rs08=
+github.com/go-jose/go-jose/v4 v4.1.4 h1:moDMcTHmvE6Groj34emNPLs/qtYXRVcd6S7NHbHz3kA=
+github.com/go-jose/go-jose/v4 v4.1.4/go.mod h1:x4oUasVrzR7071A4TnHLGSPpNOm2a21K9Kf04k1rs08=
 github.com/go-kit/kit v0.8.0/go.mod h1:xBxKIO96dXMWWy0MnWVtmwkA9/13aqxPnvrjFYMA2as=
 github.com/go-kit/kit v0.9.0/go.mod h1:xBxKIO96dXMWWy0MnWVtmwkA9/13aqxPnvrjFYMA2as=
 github.com/go-kit/log v0.1.0/go.mod h1:zbhenjAZHb184qTLMA9ZjW7ThYL0H2mk7Q6pNt4vbaY=
@@ -1,6 +1,6 @@
 apiVersion: v1
 description: SeaweedFS
 name: seaweedfs
-appVersion: "4.17"
+appVersion: "4.18"
 # Dev note: Trigger a helm chart release by `git tag -a helm-<version>`
-version: 4.17.0
+version: 4.18.0
@@ -2561,6 +2561,15 @@ version = "0.2.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "7c87def4c32ab89d880effc9e097653c8da5d6ef28e6b539d313baaacfbafcbe"

+[[package]]
+name = "openssl-src"
+version = "300.5.5+3.5.5"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "3f1787d533e03597a7934fd0a765f0d28e94ecc5fb7789f8053b1e699a56f709"
+dependencies = [
+ "cc",
+]
+
 [[package]]
 name = "openssl-sys"
 version = "0.9.111"
@@ -2569,6 +2578,7 @@ checksum = "82cab2d520aa75e3c58898289429321eb788c3106963d0dc886ec7a5f4adc321"
 dependencies = [
 "cc",
 "libc",
+ "openssl-src",
 "pkg-config",
 "vcpkg",
 ]
@@ -4654,6 +4664,7 @@ dependencies = [
 "memmap2",
 "mime_guess",
 "multer",
+ "openssl",
 "parking_lot 0.12.5",
 "pprof",
 "prometheus",
@@ -20,7 +20,7 @@ default = ["5bytes"]
 [dependencies]
 # Async runtime
 tokio = { version = "1", features = ["full"] }
-tokio-stream = "0.1"
+tokio-stream = { version = "0.1", features = ["net"] }
 tokio-io-timeout = "1"

 # gRPC + protobuf
@@ -65,8 +65,6 @@ reed-solomon-erasure = "6"
 # Logging
 tracing = "0.1"
 tracing-subscriber = { version = "0.3", features = ["env-filter"] }
-pprof = { version = "0.15", features = ["prost-codec"] }
-
 # Config
 toml = "0.8"
 serde = { version = "1", features = ["derive"] }
@@ -127,6 +125,10 @@ aws-sdk-s3 = { version = "1.125.0", default-features = false, features = ["sigv4
 aws-credential-types = "1"
 aws-types = "1"

+# pprof is Unix-only (requires libc/nix APIs not available on Windows)
+[target.'cfg(unix)'.dependencies]
+pprof = { version = "0.15", features = ["prost-codec"] }
+
 [dev-dependencies]
 tempfile = "3"

@@ -10,9 +10,11 @@ use seaweed_volume::security::tls::{
    GrpcClientAuthPolicy, TlsPolicy,
 };
 use seaweed_volume::security::{Guard, SigningKey};
+#[cfg(unix)]
 use seaweed_volume::server::debug::build_debug_router;
 use seaweed_volume::server::grpc_client::load_outgoing_grpc_tls;
 use seaweed_volume::server::grpc_server::VolumeGrpcService;
+#[cfg(unix)]
 use seaweed_volume::server::profiling::CpuProfileSession;
 use seaweed_volume::server::request_id::GrpcRequestIdLayer;
 use seaweed_volume::server::volume_server::{
@@ -24,6 +26,11 @@ use seaweed_volume::storage::types::DiskType;

 use tokio_rustls::TlsAcceptor;

+#[cfg(unix)]
+type CpuProfileParam = Option<CpuProfileSession>;
+#[cfg(not(unix))]
+type CpuProfileParam = Option<()>;
+
 const GRPC_MAX_MESSAGE_SIZE: usize = 1 << 30;
 const GRPC_KEEPALIVE_INTERVAL: std::time::Duration = std::time::Duration::from_secs(60);
 const GRPC_KEEPALIVE_TIMEOUT: std::time::Duration = std::time::Duration::from_secs(20);
@@ -42,6 +49,7 @@ fn main() {

    let config = config::parse_cli();
    seaweed_volume::server::server_stats::init_process_start();
+    #[cfg(unix)]
    let cpu_profile = match CpuProfileSession::start(&config) {
        Ok(session) => session,
        Err(e) => {
@@ -49,6 +57,8 @@ fn main() {
            std::process::exit(1);
        }
    };
+    #[cfg(not(unix))]
+    let cpu_profile: Option<()> = None;
    info!(
        "SeaweedFS Volume Server (Rust) v{}",
        seaweed_volume::version::full_version()
@@ -257,7 +267,7 @@ where

 async fn run(
    config: VolumeServerConfig,
-    cpu_profile: Option<CpuProfileSession>,
+    #[allow(unused_variables)] cpu_profile: CpuProfileParam,
 ) -> Result<(), Box<dyn std::error::Error>> {
    // Initialize the store
    let mut store = Store::new(config.index_type);
@@ -431,10 +441,12 @@ async fn run(
    }

    // Build HTTP routers
+    #[allow(unused_mut)]
    let mut admin_router = seaweed_volume::server::volume_server::build_admin_router_with_ui(
        state.clone(),
        config.ui_enabled,
    );
+    #[cfg(unix)]
    if config.pprof {
        admin_router = admin_router.merge(build_debug_router());
    }
@@ -588,31 +600,35 @@ async fn run(
        })
    };

+    // Bind the gRPC listener before spawning to propagate bind errors at startup.
+    let grpc_listener = tokio::net::TcpListener::bind(&grpc_addr)
+        .await
+        .unwrap_or_else(|e| panic!("Failed to bind gRPC to {}: {}", grpc_addr, e));
+    let grpc_local_addr = grpc_listener
+        .local_addr()
+        .unwrap_or_else(|e| panic!("Failed to get gRPC local addr: {}", e));
+
    let grpc_handle = {
        let grpc_state = state.clone();
-        let grpc_addr = grpc_addr.clone();
        let grpc_tls_acceptor = grpc_tls_acceptor.clone();
        let mut shutdown_rx = shutdown_tx.subscribe();
+        let shutdown_tx_grpc = shutdown_tx.clone();
        tokio::spawn(async move {
-            let addr = grpc_addr.parse().expect("Invalid gRPC address");
            let grpc_service = VolumeGrpcService {
                state: grpc_state.clone(),
            };
-            if let Some(tls_acceptor) = grpc_tls_acceptor {
-                let listener = tokio::net::TcpListener::bind(&grpc_addr)
-                    .await
-                    .unwrap_or_else(|e| panic!("Failed to bind gRPC to {}: {}", grpc_addr, e));
-                let incoming = grpc_tls_incoming(listener, tls_acceptor);
-                let reflection_v1 = tonic_reflection::server::Builder::configure()
-                    .register_encoded_file_descriptor_set(seaweed_volume::pb::FILE_DESCRIPTOR_SET)
-                    .build_v1()
-                    .expect("Failed to build gRPC reflection v1 service");
-                let reflection_v1alpha = tonic_reflection::server::Builder::configure()
-                    .register_encoded_file_descriptor_set(seaweed_volume::pb::FILE_DESCRIPTOR_SET)
-                    .build_v1alpha()
-                    .expect("Failed to build gRPC reflection v1alpha service");
-                info!("gRPC server listening on {} (TLS enabled)", addr);
-                if let Err(e) = build_grpc_server_builder()
+            let reflection_v1 = tonic_reflection::server::Builder::configure()
+                .register_encoded_file_descriptor_set(seaweed_volume::pb::FILE_DESCRIPTOR_SET)
+                .build_v1()
+                .expect("Failed to build gRPC reflection v1 service");
+            let reflection_v1alpha = tonic_reflection::server::Builder::configure()
+                .register_encoded_file_descriptor_set(seaweed_volume::pb::FILE_DESCRIPTOR_SET)
+                .build_v1alpha()
+                .expect("Failed to build gRPC reflection v1alpha service");
+            let result = if let Some(tls_acceptor) = grpc_tls_acceptor {
+                let incoming = grpc_tls_incoming(grpc_listener, tls_acceptor);
+                info!("gRPC server listening on {} (TLS enabled)", grpc_local_addr);
+                build_grpc_server_builder()
                    .layer(GrpcRequestIdLayer)
                    .add_service(reflection_v1)
                    .add_service(reflection_v1alpha)
@@ -621,32 +637,25 @@ async fn run(
                        let _ = shutdown_rx.recv().await;
                    })
                    .await
-                {
-                    error!("gRPC server error: {}", e);
-                }
            } else {
-                let reflection_v1 = tonic_reflection::server::Builder::configure()
-                    .register_encoded_file_descriptor_set(seaweed_volume::pb::FILE_DESCRIPTOR_SET)
-                    .build_v1()
-                    .expect("Failed to build gRPC reflection v1 service");
-                let reflection_v1alpha = tonic_reflection::server::Builder::configure()
-                    .register_encoded_file_descriptor_set(seaweed_volume::pb::FILE_DESCRIPTOR_SET)
-                    .build_v1alpha()
-                    .expect("Failed to build gRPC reflection v1alpha service");
-                info!("gRPC server listening on {}", addr);
-                if let Err(e) = build_grpc_server_builder()
+                let incoming =
+                    tokio_stream::wrappers::TcpListenerStream::new(grpc_listener);
+                info!("gRPC server listening on {}", grpc_local_addr);
+                build_grpc_server_builder()
                    .layer(GrpcRequestIdLayer)
                    .add_service(reflection_v1)
                    .add_service(reflection_v1alpha)
                    .add_service(build_volume_grpc_service(grpc_service))
-                    .serve_with_shutdown(addr, async move {
+                    .serve_with_incoming_shutdown(incoming, async move {
                        let _ = shutdown_rx.recv().await;
                    })
                    .await
-                {
-                    error!("gRPC server error: {}", e);
-                }
+            };
+            if let Err(ref e) = result {
+                error!("gRPC server error: {}", e);
+                let _ = shutdown_tx_grpc.send(());
            }
+            result
        })
    };

@@ -721,6 +730,7 @@ async fn run(
        None
    };

+    #[cfg(unix)]
    let debug_handle = if config.debug {
        let debug_addr = format!("0.0.0.0:{}", config.debug_port);
        info!("Debug pprof server listening on {}", debug_addr);
@@ -742,6 +752,8 @@ async fn run(
    } else {
        None
    };
+    #[cfg(not(unix))]
+    let debug_handle: Option<tokio::task::JoinHandle<()>> = None;

    let metrics_push_handle = {
        let push_state = state.clone();
@@ -752,9 +764,40 @@ async fn run(
        }))
    };

-    // Wait for all servers
+    // Wait for servers. Use select! with &mut so the losing handle is not
+    // dropped, then await it explicitly afterward.
+    let mut server_err: Option<String> = None;
+    let mut http_handle = http_handle;
+    let mut grpc_handle = grpc_handle;
+    let grpc_finished_first = tokio::select! {
+        _ = &mut http_handle => false,
+        _ = &mut grpc_handle => true,
+    };
+    // Inspect the gRPC result (already resolved if it finished first,
+    // otherwise await it now).
+    let grpc_result = if grpc_finished_first {
+        grpc_handle.await
+    } else {
+        // HTTP finished first; gRPC is still running. Await it.
+        grpc_handle.await
+    };
+    match grpc_result {
+        Ok(Ok(())) => {}
+        Ok(Err(e)) => {
+            let msg = format!("gRPC server exited with error: {}", e);
+            error!("{}", msg);
+            server_err = Some(msg);
+            // serve error already sent shutdown inside the task
+        }
+        Err(e) => {
+            let msg = format!("gRPC task panicked: {}", e);
+            error!("{}", msg);
+            server_err = Some(msg);
+            let _ = shutdown_tx.send(());
+        }
+    }
+    // Ensure the HTTP handle completes too.
    let _ = http_handle.await;
-    let _ = grpc_handle.await;
    if let Some(h) = public_handle {
        let _ = h.await;
    }
@@ -774,10 +817,15 @@ async fn run(
    // Close all volumes (flush and release file handles) matching Go's Shutdown()
    state.store.write().unwrap().close();

+    #[cfg(unix)]
    if let Some(cpu_profile) = cpu_profile {
        cpu_profile.finish().map_err(std::io::Error::other)?;
    }

+    if let Some(err_msg) = server_err {
+        return Err(std::io::Error::other(err_msg).into());
+    }
+
    info!("Volume server stopped.");
    Ok(())
 }
@@ -1867,6 +1867,7 @@ impl VolumeServer for VolumeGrpcService {
        {
            let needle_header = resp.needle_header;
            let mut needle_body = resp.needle_body;
+            let resp_version = resp.version;

            if needle_header.is_empty() {
                continue;
@@ -1891,8 +1892,36 @@ impl VolumeServer for VolumeGrpcService {
            // Parse needle from header + body
            let mut n = Needle::default();
            n.read_header(&needle_header);
-            n.read_body_v2(&needle_body)
-                .map_err(|e| Status::internal(format!("parse needle body: {}", e)))?;
+
+            if n.size.0 < 0 {
+                return Err(Status::invalid_argument(format!(
+                    "unexpected negative needle size {} for needle {}",
+                    n.size.0, n.id.0
+                )));
+            } else if n.size.0 > 0 {
+                // Normal needle: parse the body fields (DataSize, Data, flags, etc.)
+                n.read_body_v2(&needle_body)
+                    .map_err(|e| Status::internal(format!("parse needle body: {}", e)))?;
+            } else {
+                // Delete tombstone (size == 0): body is checksum + timestamp
+                // (V3) or checksum only (V2) + padding. Validate minimum
+                // footer length for the protocol version.
+                use crate::storage::types::{
+                    NEEDLE_CHECKSUM_SIZE, TIMESTAMP_SIZE, VERSION_3, Version,
+                };
+                let version = Version(resp_version as u8);
+                let min_footer = if version >= VERSION_3 {
+                    NEEDLE_CHECKSUM_SIZE + TIMESTAMP_SIZE
+                } else {
+                    NEEDLE_CHECKSUM_SIZE
+                };
+                if needle_body.len() < min_footer {
+                    return Err(Status::invalid_argument(format!(
+                        "tombstone needle {} body too short: got {} bytes, need >= {} for version {}",
+                        n.id.0, needle_body.len(), min_footer, resp_version
+                    )));
+                }
+            }

            // Write needle to local volume
            let mut store = state.store.write().unwrap();
@@ -4047,11 +4076,12 @@ fn find_last_append_at_ns(idx_path: &str, dat_path: &str, version: u32) -> Optio
    let mut header = [0u8; 16];
    dat_file.read_exact(&mut header).ok()?;
    let needle_size = i32::from_be_bytes([header[12], header[13], header[14], header[15]]);
-    if needle_size <= 0 {
+    if needle_size < 0 {
        return None;
    }

    // Seek to tail: offset + 16 (header) + size -> checksum (4) + timestamp (8)
+    // For delete needles (size == 0), the tail is right after the header.
    let tail_offset = actual_offset as u64 + 16 + needle_size as u64;
    dat_file.seek(SeekFrom::Start(tail_offset)).ok()?;

@@ -1,9 +1,11 @@
+#[cfg(unix)]
 pub mod debug;
 pub mod grpc_client;
 pub mod grpc_server;
 pub mod handlers;
 pub mod heartbeat;
 pub mod memory_status;
+#[cfg(unix)]
 pub mod profiling;
 pub mod request_id;
 pub mod server_stats;
@@ -6,7 +6,7 @@
 use std::fs::File;
 use std::io;
 #[cfg(not(unix))]
-use std::io::{Seek, SeekFrom};
+use std::io::{Read, Seek, SeekFrom};

 use reed_solomon_erasure::galois_8::ReedSolomon;

@@ -249,11 +249,18 @@ func testConcurrentDirectoryOperations(t *testing.T, framework *FuseTestFramewor
 					return
 				}

-				// Create file in subdirectory
+				// Create file in subdirectory with retry for transient FUSE errors
 				testFile := filepath.Join(subDir, "test.txt")
 				content := []byte(fmt.Sprintf("Worker %d, Subdir %d", workerID, i))
-				if err := os.WriteFile(testFile, content, 0644); err != nil {
-					addError(fmt.Errorf("worker %d file %d: %v", workerID, i, err))
+				var writeErr error
+				for attempt := 0; attempt < 3; attempt++ {
+					if writeErr = os.WriteFile(testFile, content, 0644); writeErr == nil {
+						break
+					}
+					time.Sleep(100 * time.Millisecond)
+				}
+				if writeErr != nil {
+					addError(fmt.Errorf("worker %d file %d: %v", workerID, i, writeErr))
 					return
 				}
 			}
@@ -85,9 +85,19 @@ func testGitCloneAndPull(t *testing.T, mountPoint, localDir string) {
 	branch := gitOutput(t, localClone, "rev-parse", "--abbrev-ref", "HEAD")
 	gitRun(t, localClone, "push", "origin", branch)

+	// The bare repo lives on the FUSE mount and can briefly disappear after
+	// the push completes, and pushed pack objects may not be immediately
+	// consistent on the FUSE layer. Give the mount a chance to settle, then
+	// recover from the local clone if the remote is still missing.
+	if !waitForBareRepoEventually(t, bareRepo, 10*time.Second) {
+		t.Logf("bare repo %s did not stabilise after push; forcing recovery before clone", bareRepo)
+	}
+	refreshDirEntry(t, bareRepo)
+	time.Sleep(1 * time.Second)
+
 	// ---- Phase 3: Clone from mount bare repo into on-mount working dir ----
 	t.Log("Phase 3: clone from mount bare repo to on-mount working dir")
-	gitRun(t, "", "clone", bareRepo, mountClone)
+	ensureMountCloneFromBareWithRecovery(t, bareRepo, localClone, mountClone)

 	assertFileContains(t, filepath.Join(mountClone, "README.md"), "# Updated")
 	assertFileContains(t, filepath.Join(mountClone, "src/main.go"), "v2")
@@ -290,7 +300,7 @@ func waitForBareRepoEventually(t *testing.T, bareRepo string, timeout time.Durat
 	t.Helper()
 	deadline := time.Now().Add(timeout)
 	for time.Now().Before(deadline) {
-		if isBareRepo(bareRepo) {
+		if isBareRepoAccessible(bareRepo) {
 			return true
 		}
 		refreshDirEntry(t, bareRepo)
@@ -312,6 +322,14 @@ func isBareRepo(bareRepo string) bool {
 	return true
 }

+func isBareRepoAccessible(bareRepo string) bool {
+	if !isBareRepo(bareRepo) {
+		return false
+	}
+	out, err := tryGitCommand("", "--git-dir="+bareRepo, "rev-parse", "--is-bare-repository")
+	return err == nil && out == "true"
+}
+
 func ensureMountClone(t *testing.T, bareRepo, mountClone string) {
 	t.Helper()
 	require.NoError(t, tryEnsureMountClone(bareRepo, mountClone))
@@ -320,7 +338,7 @@ func ensureMountClone(t *testing.T, bareRepo, mountClone string) {
 // tryEnsureBareRepo verifies the bare repo on the FUSE mount exists.
 // If it has vanished, it re-creates it from the local clone.
 func tryEnsureBareRepo(bareRepo, localClone string) error {
-	if _, err := os.Stat(filepath.Join(bareRepo, "HEAD")); err == nil {
+	if isBareRepoAccessible(bareRepo) {
 		return nil
 	}
 	branch, err := tryGitCommand(localClone, "rev-parse", "--abbrev-ref", "HEAD")
@@ -342,6 +360,36 @@ func tryEnsureBareRepo(bareRepo, localClone string) error {
 	return nil
 }

+func ensureMountCloneFromBareWithRecovery(t *testing.T, bareRepo, localClone, mountClone string) {
+	t.Helper()
+	const maxAttempts = 3
+	var lastErr error
+	for attempt := 1; attempt <= maxAttempts; attempt++ {
+		if lastErr = tryEnsureMountCloneFromBare(bareRepo, localClone, mountClone); lastErr == nil {
+			return
+		}
+		if attempt == maxAttempts {
+			require.NoError(t, lastErr, "git clone %s %s failed after %d recovery attempts", bareRepo, mountClone, maxAttempts)
+		}
+		t.Logf("clone recovery attempt %d: %v — removing clone for re-create", attempt, lastErr)
+		os.RemoveAll(mountClone)
+		time.Sleep(2 * time.Second)
+	}
+}
+
+func tryEnsureMountCloneFromBare(bareRepo, localClone, mountClone string) error {
+	if err := tryEnsureBareRepo(bareRepo, localClone); err != nil {
+		return fmt.Errorf("ensure bare repo: %w", err)
+	}
+	if err := tryEnsureMountClone(bareRepo, mountClone); err != nil {
+		return fmt.Errorf("ensure mount clone: %w", err)
+	}
+	if _, err := tryGitCommand(mountClone, "rev-parse", "HEAD"); err != nil {
+		return fmt.Errorf("verify mount clone: %w", err)
+	}
+	return nil
+}
+
 // tryEnsureMountClone is like ensureMountClone but returns an error instead
 // of failing the test, for use in recovery loops.
 func tryEnsureMountClone(bareRepo, mountClone string) error {
@@ -550,3 +598,33 @@ func TestTryEnsureBareRepoPreservesCurrentBranch(t *testing.T) {
 	assert.Equal(t, branch, restoredHead, "clone from recovered bare repo should check out the current branch")
 	assertFileContains(t, filepath.Join(restoredClone, "README.md"), "hello recovery")
 }
+
+func TestEnsureMountCloneFromBareWithRecoveryRecreatesMissingBareRepo(t *testing.T) {
+	tempDir, err := os.MkdirTemp("", "git_mount_clone_recovery_")
+	require.NoError(t, err)
+	defer os.RemoveAll(tempDir)
+
+	bareRepo := filepath.Join(tempDir, "repo.git")
+	localClone := filepath.Join(tempDir, "clone")
+	mountClone := filepath.Join(tempDir, "mount-clone")
+
+	gitRun(t, "", "init", "--bare", bareRepo)
+	gitRun(t, "", "clone", bareRepo, localClone)
+	gitRun(t, localClone, "config", "user.email", "test@seaweedfs.test")
+	gitRun(t, localClone, "config", "user.name", "Test")
+
+	writeFile(t, localClone, "README.md", "hello clone recovery\n")
+	gitRun(t, localClone, "add", "README.md")
+	gitRun(t, localClone, "commit", "-m", "initial commit")
+
+	branch := gitOutput(t, localClone, "rev-parse", "--abbrev-ref", "HEAD")
+	gitRun(t, localClone, "push", "origin", branch)
+
+	require.NoError(t, os.RemoveAll(bareRepo))
+
+	ensureMountCloneFromBareWithRecovery(t, bareRepo, localClone, mountClone)
+
+	head := gitOutput(t, mountClone, "rev-parse", "--abbrev-ref", "HEAD")
+	assert.Equal(t, branch, head, "recovered clone should stay on the pushed branch")
+	assertFileContains(t, filepath.Join(mountClone, "README.md"), "hello clone recovery")
+}
@@ -43,25 +43,25 @@ require (
 	github.com/andybalholm/cascadia v1.3.3 // indirect
 	github.com/appscode/go-querystring v0.0.0-20170504095604-0126cfb3f1dc // indirect
 	github.com/aws/aws-sdk-go v1.55.8 // indirect
-	github.com/aws/aws-sdk-go-v2 v1.41.4 // indirect
+	github.com/aws/aws-sdk-go-v2 v1.41.5 // indirect
 	github.com/aws/aws-sdk-go-v2/aws/protocol/eventstream v1.7.4 // indirect
-	github.com/aws/aws-sdk-go-v2/config v1.32.9 // indirect
-	github.com/aws/aws-sdk-go-v2/credentials v1.19.12 // indirect
-	github.com/aws/aws-sdk-go-v2/feature/ec2/imds v1.18.20 // indirect
+	github.com/aws/aws-sdk-go-v2/config v1.32.13 // indirect
+	github.com/aws/aws-sdk-go-v2/credentials v1.19.13 // indirect
+	github.com/aws/aws-sdk-go-v2/feature/ec2/imds v1.18.21 // indirect
 	github.com/aws/aws-sdk-go-v2/feature/s3/manager v1.20.12 // indirect
-	github.com/aws/aws-sdk-go-v2/internal/configsources v1.4.20 // indirect
-	github.com/aws/aws-sdk-go-v2/internal/endpoints/v2 v2.7.20 // indirect
-	github.com/aws/aws-sdk-go-v2/internal/ini v1.8.4 // indirect
+	github.com/aws/aws-sdk-go-v2/internal/configsources v1.4.21 // indirect
+	github.com/aws/aws-sdk-go-v2/internal/endpoints/v2 v2.7.21 // indirect
+	github.com/aws/aws-sdk-go-v2/internal/ini v1.8.6 // indirect
 	github.com/aws/aws-sdk-go-v2/internal/v4a v1.4.17 // indirect
 	github.com/aws/aws-sdk-go-v2/service/internal/accept-encoding v1.13.7 // indirect
 	github.com/aws/aws-sdk-go-v2/service/internal/checksum v1.9.8 // indirect
-	github.com/aws/aws-sdk-go-v2/service/internal/presigned-url v1.13.20 // indirect
+	github.com/aws/aws-sdk-go-v2/service/internal/presigned-url v1.13.21 // indirect
 	github.com/aws/aws-sdk-go-v2/service/internal/s3shared v1.19.17 // indirect
 	github.com/aws/aws-sdk-go-v2/service/s3 v1.96.0 // indirect
-	github.com/aws/aws-sdk-go-v2/service/signin v1.0.8 // indirect
-	github.com/aws/aws-sdk-go-v2/service/sso v1.30.13 // indirect
-	github.com/aws/aws-sdk-go-v2/service/ssooidc v1.35.17 // indirect
-	github.com/aws/aws-sdk-go-v2/service/sts v1.41.9 // indirect
+	github.com/aws/aws-sdk-go-v2/service/signin v1.0.9 // indirect
+	github.com/aws/aws-sdk-go-v2/service/sso v1.30.14 // indirect
+	github.com/aws/aws-sdk-go-v2/service/ssooidc v1.35.18 // indirect
+	github.com/aws/aws-sdk-go-v2/service/sts v1.41.10 // indirect
 	github.com/aws/smithy-go v1.24.2 // indirect
 	github.com/bahlo/generic-list-go v0.2.0 // indirect
 	github.com/beorn7/perks v1.0.1 // indirect
@@ -104,7 +104,7 @@ require (
 	github.com/go-chi/chi/v5 v5.2.5 // indirect
 	github.com/go-darwin/apfs v0.0.0-20211011131704-f84b94dbf348 // indirect
 	github.com/go-git/go-billy/v5 v5.6.2 // indirect
-	github.com/go-jose/go-jose/v4 v4.1.3 // indirect
+	github.com/go-jose/go-jose/v4 v4.1.4 // indirect
 	github.com/go-logr/logr v1.4.3 // indirect
 	github.com/go-logr/stdr v1.2.2 // indirect
 	github.com/go-ole/go-ole v1.3.0 // indirect
@@ -112,44 +112,44 @@ github.com/appscode/go-querystring v0.0.0-20170504095604-0126cfb3f1dc h1:LoL75er
 github.com/appscode/go-querystring v0.0.0-20170504095604-0126cfb3f1dc/go.mod h1:w648aMHEgFYS6xb0KVMMtZ2uMeemhiKCuD2vj6gY52A=
 github.com/aws/aws-sdk-go v1.55.8 h1:JRmEUbU52aJQZ2AjX4q4Wu7t4uZjOu71uyNmaWlUkJQ=
 github.com/aws/aws-sdk-go v1.55.8/go.mod h1:ZkViS9AqA6otK+JBBNH2++sx1sgxrPKcSzPPvQkUtXk=
-github.com/aws/aws-sdk-go-v2 v1.41.4 h1:10f50G7WyU02T56ox1wWXq+zTX9I1zxG46HYuG1hH/k=
-github.com/aws/aws-sdk-go-v2 v1.41.4/go.mod h1:mwsPRE8ceUUpiTgF7QmQIJ7lgsKUPQOUl3o72QBrE1o=
+github.com/aws/aws-sdk-go-v2 v1.41.5 h1:dj5kopbwUsVUVFgO4Fi5BIT3t4WyqIDjGKCangnV/yY=
+github.com/aws/aws-sdk-go-v2 v1.41.5/go.mod h1:mwsPRE8ceUUpiTgF7QmQIJ7lgsKUPQOUl3o72QBrE1o=
 github.com/aws/aws-sdk-go-v2/aws/protocol/eventstream v1.7.4 h1:489krEF9xIGkOaaX3CE/Be2uWjiXrkCH6gUX+bZA/BU=
 github.com/aws/aws-sdk-go-v2/aws/protocol/eventstream v1.7.4/go.mod h1:IOAPF6oT9KCsceNTvvYMNHy0+kMF8akOjeDvPENWxp4=
-github.com/aws/aws-sdk-go-v2/config v1.32.9 h1:ktda/mtAydeObvJXlHzyGpK1xcsLaP16zfUPDGoW90A=
-github.com/aws/aws-sdk-go-v2/config v1.32.9/go.mod h1:U+fCQ+9QKsLW786BCfEjYRj34VVTbPdsLP3CHSYXMOI=
-github.com/aws/aws-sdk-go-v2/credentials v1.19.12 h1:oqtA6v+y5fZg//tcTWahyN9PEn5eDU/Wpvc2+kJ4aY8=
-github.com/aws/aws-sdk-go-v2/credentials v1.19.12/go.mod h1:U3R1RtSHx6NB0DvEQFGyf/0sbrpJrluENHdPy1j/3TE=
-github.com/aws/aws-sdk-go-v2/feature/ec2/imds v1.18.20 h1:zOgq3uezl5nznfoK3ODuqbhVg1JzAGDUhXOsU0IDCAo=
-github.com/aws/aws-sdk-go-v2/feature/ec2/imds v1.18.20/go.mod h1:z/MVwUARehy6GAg/yQ1GO2IMl0k++cu1ohP9zo887wE=
+github.com/aws/aws-sdk-go-v2/config v1.32.13 h1:5KgbxMaS2coSWRrx9TX/QtWbqzgQkOdEa3sZPhBhCSg=
+github.com/aws/aws-sdk-go-v2/config v1.32.13/go.mod h1:8zz7wedqtCbw5e9Mi2doEwDyEgHcEE9YOJp6a8jdSMY=
+github.com/aws/aws-sdk-go-v2/credentials v1.19.13 h1:mA59E3fokBvyEGHKFdnpNNrvaR351cqiHgRg+JzOSRI=
+github.com/aws/aws-sdk-go-v2/credentials v1.19.13/go.mod h1:yoTXOQKea18nrM69wGF9jBdG4WocSZA1h38A+t/MAsk=
+github.com/aws/aws-sdk-go-v2/feature/ec2/imds v1.18.21 h1:NUS3K4BTDArQqNu2ih7yeDLaS3bmHD0YndtA6UP884g=
+github.com/aws/aws-sdk-go-v2/feature/ec2/imds v1.18.21/go.mod h1:YWNWJQNjKigKY1RHVJCuupeWDrrHjRqHm0N9rdrWzYI=
 github.com/aws/aws-sdk-go-v2/feature/s3/manager v1.20.12 h1:Zy6Tme1AA13kX8x3CnkHx5cqdGWGaj/anwOiWGnA0Xo=
 github.com/aws/aws-sdk-go-v2/feature/s3/manager v1.20.12/go.mod h1:ql4uXYKoTM9WUAUSmthY4AtPVrlTBZOvnBJTiCUdPxI=
-github.com/aws/aws-sdk-go-v2/internal/configsources v1.4.20 h1:CNXO7mvgThFGqOFgbNAP2nol2qAWBOGfqR/7tQlvLmc=
-github.com/aws/aws-sdk-go-v2/internal/configsources v1.4.20/go.mod h1:oydPDJKcfMhgfcgBUZaG+toBbwy8yPWubJXBVERtI4o=
-github.com/aws/aws-sdk-go-v2/internal/endpoints/v2 v2.7.20 h1:tN6W/hg+pkM+tf9XDkWUbDEjGLb+raoBMFsTodcoYKw=
-github.com/aws/aws-sdk-go-v2/internal/endpoints/v2 v2.7.20/go.mod h1:YJ898MhD067hSHA6xYCx5ts/jEd8BSOLtQDL3iZsvbc=
-github.com/aws/aws-sdk-go-v2/internal/ini v1.8.4 h1:WKuaxf++XKWlHWu9ECbMlha8WOEGm0OUEZqm4K/Gcfk=
-github.com/aws/aws-sdk-go-v2/internal/ini v1.8.4/go.mod h1:ZWy7j6v1vWGmPReu0iSGvRiise4YI5SkR3OHKTZ6Wuc=
+github.com/aws/aws-sdk-go-v2/internal/configsources v1.4.21 h1:Rgg6wvjjtX8bNHcvi9OnXWwcE0a2vGpbwmtICOsvcf4=
+github.com/aws/aws-sdk-go-v2/internal/configsources v1.4.21/go.mod h1:A/kJFst/nm//cyqonihbdpQZwiUhhzpqTsdbhDdRF9c=
+github.com/aws/aws-sdk-go-v2/internal/endpoints/v2 v2.7.21 h1:PEgGVtPoB6NTpPrBgqSE5hE/o47Ij9qk/SEZFbUOe9A=
+github.com/aws/aws-sdk-go-v2/internal/endpoints/v2 v2.7.21/go.mod h1:p+hz+PRAYlY3zcpJhPwXlLC4C+kqn70WIHwnzAfs6ps=
+github.com/aws/aws-sdk-go-v2/internal/ini v1.8.6 h1:qYQ4pzQ2Oz6WpQ8T3HvGHnZydA72MnLuFK9tJwmrbHw=
+github.com/aws/aws-sdk-go-v2/internal/ini v1.8.6/go.mod h1:O3h0IK87yXci+kg6flUKzJnWeziQUKciKrLjcatSNcY=
 github.com/aws/aws-sdk-go-v2/internal/v4a v1.4.17 h1:JqcdRG//czea7Ppjb+g/n4o8i/R50aTBHkA7vu0lK+k=
 github.com/aws/aws-sdk-go-v2/internal/v4a v1.4.17/go.mod h1:CO+WeGmIdj/MlPel2KwID9Gt7CNq4M65HUfBW97liM0=
 github.com/aws/aws-sdk-go-v2/service/internal/accept-encoding v1.13.7 h1:5EniKhLZe4xzL7a+fU3C2tfUN4nWIqlLesfrjkuPFTY=
 github.com/aws/aws-sdk-go-v2/service/internal/accept-encoding v1.13.7/go.mod h1:x0nZssQ3qZSnIcePWLvcoFisRXJzcTVvYpAAdYX8+GI=
 github.com/aws/aws-sdk-go-v2/service/internal/checksum v1.9.8 h1:Z5EiPIzXKewUQK0QTMkutjiaPVeVYXX7KIqhXu/0fXs=
 github.com/aws/aws-sdk-go-v2/service/internal/checksum v1.9.8/go.mod h1:FsTpJtvC4U1fyDXk7c71XoDv3HlRm8V3NiYLeYLh5YE=
-github.com/aws/aws-sdk-go-v2/service/internal/presigned-url v1.13.20 h1:2HvVAIq+YqgGotK6EkMf+KIEqTISmTYh5zLpYyeTo1Y=
-github.com/aws/aws-sdk-go-v2/service/internal/presigned-url v1.13.20/go.mod h1:V4X406Y666khGa8ghKmphma/7C0DAtEQYhkq9z4vpbk=
+github.com/aws/aws-sdk-go-v2/service/internal/presigned-url v1.13.21 h1:c31//R3xgIJMSC8S6hEVq+38DcvUlgFY0FM6mSI5oto=
+github.com/aws/aws-sdk-go-v2/service/internal/presigned-url v1.13.21/go.mod h1:r6+pf23ouCB718FUxaqzZdbpYFyDtehyZcmP5KL9FkA=
 github.com/aws/aws-sdk-go-v2/service/internal/s3shared v1.19.17 h1:bGeHBsGZx0Dvu/eJC0Lh9adJa3M1xREcndxLNZlve2U=
 github.com/aws/aws-sdk-go-v2/service/internal/s3shared v1.19.17/go.mod h1:dcW24lbU0CzHusTE8LLHhRLI42ejmINN8Lcr22bwh/g=
 github.com/aws/aws-sdk-go-v2/service/s3 v1.96.0 h1:oeu8VPlOre74lBA/PMhxa5vewaMIMmILM+RraSyB8KA=
 github.com/aws/aws-sdk-go-v2/service/s3 v1.96.0/go.mod h1:5jggDlZ2CLQhwJBiZJb4vfk4f0GxWdEDruWKEJ1xOdo=
-github.com/aws/aws-sdk-go-v2/service/signin v1.0.8 h1:0GFOLzEbOyZABS3PhYfBIx2rNBACYcKty+XGkTgw1ow=
-github.com/aws/aws-sdk-go-v2/service/signin v1.0.8/go.mod h1:LXypKvk85AROkKhOG6/YEcHFPoX+prKTowKnVdcaIxE=
-github.com/aws/aws-sdk-go-v2/service/sso v1.30.13 h1:kiIDLZ005EcKomYYITtfsjn7dtOwHDOFy7IbPXKek2o=
-github.com/aws/aws-sdk-go-v2/service/sso v1.30.13/go.mod h1:2h/xGEowcW/g38g06g3KpRWDlT+OTfxxI0o1KqayAB8=
-github.com/aws/aws-sdk-go-v2/service/ssooidc v1.35.17 h1:jzKAXIlhZhJbnYwHbvUQZEB8KfgAEuG0dc08Bkda7NU=
-github.com/aws/aws-sdk-go-v2/service/ssooidc v1.35.17/go.mod h1:Al9fFsXjv4KfbzQHGe6V4NZSZQXecFcvaIF4e70FoRA=
-github.com/aws/aws-sdk-go-v2/service/sts v1.41.9 h1:Cng+OOwCHmFljXIxpEVXAGMnBia8MSU6Ch5i9PgBkcU=
-github.com/aws/aws-sdk-go-v2/service/sts v1.41.9/go.mod h1:LrlIndBDdjA/EeXeyNBle+gyCwTlizzW5ycgWnvIxkk=
+github.com/aws/aws-sdk-go-v2/service/signin v1.0.9 h1:QKZH0S178gCmFEgst8hN0mCX1KxLgHBKKY/CLqwP8lg=
+github.com/aws/aws-sdk-go-v2/service/signin v1.0.9/go.mod h1:7yuQJoT+OoH8aqIxw9vwF+8KpvLZ8AWmvmUWHsGQZvI=
+github.com/aws/aws-sdk-go-v2/service/sso v1.30.14 h1:GcLE9ba5ehAQma6wlopUesYg/hbcOhFNWTjELkiWkh4=
+github.com/aws/aws-sdk-go-v2/service/sso v1.30.14/go.mod h1:WSvS1NLr7JaPunCXqpJnWk1Bjo7IxzZXrZi1QQCkuqM=
+github.com/aws/aws-sdk-go-v2/service/ssooidc v1.35.18 h1:mP49nTpfKtpXLt5SLn8Uv8z6W+03jYVoOSAl/c02nog=
+github.com/aws/aws-sdk-go-v2/service/ssooidc v1.35.18/go.mod h1:YO8TrYtFdl5w/4vmjL8zaBSsiNp3w0L1FfKVKenZT7w=
+github.com/aws/aws-sdk-go-v2/service/sts v1.41.10 h1:p8ogvvLugcR/zLBXTXrTkj0RYBUdErbMnAFFp12Lm/U=
+github.com/aws/aws-sdk-go-v2/service/sts v1.41.10/go.mod h1:60dv0eZJfeVXfbT1tFJinbHrDfSJ2GZl4Q//OSSNAVw=
 github.com/aws/smithy-go v1.24.2 h1:FzA3bu/nt/vDvmnkg+R8Xl46gmzEDam6mZ1hzmwXFng=
 github.com/aws/smithy-go v1.24.2/go.mod h1:YE2RhdIuDbA5E5bTdciG9KrW3+TiEONeUWCqxX9i1Fc=
 github.com/bahlo/generic-list-go v0.2.0 h1:5sz/EEAK+ls5wF+NeqDpk5+iNdMDXrh3z3nPnH1Wvgk=
@@ -284,8 +284,8 @@ github.com/go-git/go-billy/v5 v5.6.2/go.mod h1:rcFC2rAsp/erv7CMz9GczHcuD0D32fWzH
 github.com/go-gl/glfw v0.0.0-20190409004039-e6da0acd62b1/go.mod h1:vR7hzQXu2zJy9AVAgeJqvqgH9Q5CA+iKCZ2gyEVpxRU=
 github.com/go-gl/glfw/v3.3/glfw v0.0.0-20191125211704-12ad95a8df72/go.mod h1:tQ2UAYgL5IevRw8kRxooKSPJfGvJ9fJQFa0TUsXzTg8=
 github.com/go-gl/glfw/v3.3/glfw v0.0.0-20200222043503-6f7a984d4dc4/go.mod h1:tQ2UAYgL5IevRw8kRxooKSPJfGvJ9fJQFa0TUsXzTg8=
-github.com/go-jose/go-jose/v4 v4.1.3 h1:CVLmWDhDVRa6Mi/IgCgaopNosCaHz7zrMeF9MlZRkrs=
-github.com/go-jose/go-jose/v4 v4.1.3/go.mod h1:x4oUasVrzR7071A4TnHLGSPpNOm2a21K9Kf04k1rs08=
+github.com/go-jose/go-jose/v4 v4.1.4 h1:moDMcTHmvE6Groj34emNPLs/qtYXRVcd6S7NHbHz3kA=
+github.com/go-jose/go-jose/v4 v4.1.4/go.mod h1:x4oUasVrzR7071A4TnHLGSPpNOm2a21K9Kf04k1rs08=
 github.com/go-logr/logr v1.2.2/go.mod h1:jdQByPbusPIv2/zmleS9BjJVeZ6kBagPoEUsqbVz/1A=
 github.com/go-logr/logr v1.4.3 h1:CjnDlHq8ikf6E492q6eKboGOC0T8CDaOvkHCIg8idEI=
 github.com/go-logr/logr v1.4.3/go.mod h1:9T104GzyrTigFIr8wt5mBrctHMim0Nb2HLGrmQ40KvY=
@@ -707,8 +707,8 @@ github.com/xanzy/ssh-agent v0.3.3 h1:+/15pJfg/RsTxqYcX6fHqOXZwwMP+2VyYWJeWM2qQFM
 github.com/xanzy/ssh-agent v0.3.3/go.mod h1:6dzNDKs0J9rVPHPhaGCukekBHKqfl+L3KghI1Bc68Uw=
 github.com/xdg-go/pbkdf2 v1.0.0 h1:Su7DPu48wXMwC3bs7MCNG+z4FhcyEuz5dlvchbq0B0c=
 github.com/xdg-go/pbkdf2 v1.0.0/go.mod h1:jrpuAogTd400dnrH08LKmI/xc1MbPOebTwRqcT5RDeI=
-github.com/xdg-go/scram v1.1.2 h1:FHX5I5B4i4hKRVRBCFRxq1iQRej7WO3hhBuJf+UUySY=
-github.com/xdg-go/scram v1.1.2/go.mod h1:RT/sEzTbU5y00aCK8UOx6R7YryM0iF1N2MOmC3kKLN4=
+github.com/xdg-go/scram v1.2.0 h1:bYKF2AEwG5rqd1BumT4gAnvwU/M9nBp2pTSxeZw7Wvs=
+github.com/xdg-go/scram v1.2.0/go.mod h1:3dlrS0iBaWKYVt2ZfA4cj48umJZ+cAEbR6/SjLA88I8=
 github.com/xdg-go/stringprep v1.0.4 h1:XLI/Ng3O1Atzq0oBs3TWm+5ZVgkq2aqdlvP9JtoZ6c8=
 github.com/xdg-go/stringprep v1.0.4/go.mod h1:mPGuuIYwz7CmR2bT9j4GbQqutWS1zV24gijq1dTyGkM=
 github.com/xeipuuv/gojsonpointer v0.0.0-20180127040702-4e3ac2762d5f/go.mod h1:N2zxlSyiKSe5eX1tZViRH5QA0qijqEDrYZiPEAiq3wU=
@@ -21,7 +21,9 @@ import (
 	"github.com/aws/aws-sdk-go-v2/credentials"
 	"github.com/aws/aws-sdk-go-v2/service/s3"
 	"github.com/seaweedfs/seaweedfs/test/volume_server/framework"
+	"github.com/seaweedfs/seaweedfs/weed/cluster/lock_manager"
 	"github.com/seaweedfs/seaweedfs/weed/pb"
+	"github.com/seaweedfs/seaweedfs/weed/pb/filer_pb"
 	"github.com/seaweedfs/seaweedfs/weed/pb/master_pb"
 	"github.com/stretchr/testify/require"
 	"google.golang.org/grpc"
@@ -139,6 +141,7 @@ func startDistributedLockCluster(t *testing.T) *distributedLockCluster {
 		require.NoError(t, cluster.waitForTCP(cluster.filerGRPCAddress(i), 30*time.Second), "wait for filer %d grpc\n%s", i, cluster.tailLog(fmt.Sprintf("filer%d.log", i)))
 	}
 	require.NoError(t, cluster.waitForFilerCount(2, 30*time.Second), "wait for filer group registration")
+	require.NoError(t, cluster.waitForLockRingConverged(30*time.Second), "wait for lock ring convergence")

 	for i := 0; i < 2; i++ {
 		require.NoError(t, cluster.startS3(i), "start s3 %d", i)
@@ -411,6 +414,130 @@ func (c *distributedLockCluster) waitForFilerCount(expected int, timeout time.Du
 	return fmt.Errorf("timed out waiting for %d filers in group %q", expected, c.filerGroup)
 }

+// waitForLockRingConverged verifies that both filers have a consistent view of the
+// lock ring by acquiring the same lock through each filer and checking mutual exclusion.
+// This guards against the window between master seeing both filers and the filers
+// actually receiving the LockRingUpdate broadcast (delayed by the stabilization timer).
+//
+// A single arbitrary key could false-pass: if the key's real primary is filer0 and
+// filer0 has a stale ring (only sees itself), it still grants correctly because it IS
+// the primary. To catch the stale ring we must also test a key whose real primary is
+// filer1. So we generate one test key per primary filer using the same consistent-hash
+// ring as production, and require mutual exclusion for all of them.
+func (c *distributedLockCluster) waitForLockRingConverged(timeout time.Duration) error {
+	deadline := time.Now().Add(timeout)
+
+	owners := make([]pb.ServerAddress, 0, len(c.filerPorts))
+	for i := range c.filerPorts {
+		owners = append(owners, c.filerServerAddress(i))
+	}
+	ring := lock_manager.NewHashRing(lock_manager.DefaultVnodeCount)
+	ring.SetServers(owners)
+
+	attempt := 0
+	for time.Now().Before(deadline) {
+		// Generate one unique test key per primary filer
+		testKeys := c.convergenceKeysPerPrimary(ring, owners, attempt)
+		attempt++
+
+		allConverged := true
+		for _, key := range testKeys {
+			converged, _ := c.checkLockMutualExclusion(key)
+			if !converged {
+				allConverged = false
+				break
+			}
+		}
+		if allConverged {
+			return nil
+		}
+		time.Sleep(500 * time.Millisecond)
+	}
+	return fmt.Errorf("lock ring did not converge: both filers independently grant the same lock")
+}
+
+// convergenceKeysPerPrimary returns one lock key per distinct primary filer,
+// using the same consistent-hash ring as production routing.
+func (c *distributedLockCluster) convergenceKeysPerPrimary(ring *lock_manager.HashRing, owners []pb.ServerAddress, attempt int) []string {
+	keysByOwner := make(map[pb.ServerAddress]string, len(owners))
+	for i := 0; len(keysByOwner) < len(owners) && i < 1024; i++ {
+		candidate := fmt.Sprintf("convergence-%d-%d", attempt, i)
+		primary := ring.GetPrimary(candidate)
+		if _, exists := keysByOwner[primary]; !exists {
+			keysByOwner[primary] = candidate
+		}
+	}
+	keys := make([]string, 0, len(keysByOwner))
+	for _, k := range keysByOwner {
+		keys = append(keys, k)
+	}
+	return keys
+}
+
+// checkLockMutualExclusion acquires a lock via filer0, then tries the same lock via filer1.
+// Returns true if mutual exclusion holds (second attempt is denied).
+func (c *distributedLockCluster) checkLockMutualExclusion(testKey string) (bool, error) {
+	conn0, err := grpc.NewClient(c.filerGRPCAddress(0), grpc.WithTransportCredentials(insecure.NewCredentials()))
+	if err != nil {
+		return false, err
+	}
+	defer conn0.Close()
+
+	conn1, err := grpc.NewClient(c.filerGRPCAddress(1), grpc.WithTransportCredentials(insecure.NewCredentials()))
+	if err != nil {
+		return false, err
+	}
+	defer conn1.Close()
+
+	client0 := filer_pb.NewSeaweedFilerClient(conn0)
+	client1 := filer_pb.NewSeaweedFilerClient(conn1)
+
+	// Acquire lock via filer0
+	ctx0, cancel0 := context.WithTimeout(context.Background(), 2*time.Second)
+	defer cancel0()
+	resp0, err := client0.DistributedLock(ctx0, &filer_pb.LockRequest{
+		Name:          testKey,
+		SecondsToLock: 5,
+		Owner:         "convergence-filer0",
+	})
+	if err != nil || resp0.RenewToken == "" {
+		return false, fmt.Errorf("filer0 lock failed: err=%v resp=%v", err, resp0)
+	}
+	defer func() {
+		// Always release the lock we acquired
+		client0.DistributedUnlock(context.Background(), &filer_pb.UnlockRequest{
+			Name:       testKey,
+			RenewToken: resp0.RenewToken,
+		})
+	}()
+
+	// Try the same lock via filer1 - should be denied
+	ctx1, cancel1 := context.WithTimeout(context.Background(), 2*time.Second)
+	defer cancel1()
+	resp1, err := client1.DistributedLock(ctx1, &filer_pb.LockRequest{
+		Name:          testKey,
+		SecondsToLock: 5,
+		Owner:         "convergence-filer1",
+	})
+	if err != nil {
+		return false, err
+	}
+	if resp1.RenewToken != "" {
+		// Both filers granted the lock - ring not converged. Release the second lock.
+		client1.DistributedUnlock(context.Background(), &filer_pb.UnlockRequest{
+			Name:       testKey,
+			RenewToken: resp1.RenewToken,
+		})
+		return false, nil
+	}
+	// Verify the denial is specifically because the lock is already held,
+	// not due to a transient error that might give a false positive.
+	if !strings.Contains(resp1.Error, "lock already owned") {
+		return false, nil
+	}
+	return true, nil
+}
+
 func (c *distributedLockCluster) waitForHTTP(url string, timeout time.Duration) error {
 	client := &net.Dialer{Timeout: time.Second}
 	httpClient := &httpClientWithDialer{dialer: client}
@@ -6,7 +6,6 @@ import (
 	"errors"
 	"fmt"
 	"io"
-	"sort"
 	"strings"
 	"sync"
 	"testing"
@@ -15,9 +14,9 @@ import (
 	"github.com/aws/aws-sdk-go-v2/aws"
 	"github.com/aws/aws-sdk-go-v2/service/s3"
 	"github.com/aws/smithy-go"
+	"github.com/seaweedfs/seaweedfs/weed/cluster/lock_manager"
 	"github.com/seaweedfs/seaweedfs/weed/pb"
 	"github.com/seaweedfs/seaweedfs/weed/s3api/s3_constants"
-	"github.com/seaweedfs/seaweedfs/weed/util"
 	"github.com/stretchr/testify/assert"
 	"github.com/stretchr/testify/require"
 )
@@ -149,14 +148,15 @@ func (c *distributedLockCluster) findLockOwnerKeys(bucket, prefix string) map[pb
 	for i := range c.filerPorts {
 		owners = append(owners, c.filerServerAddress(i))
 	}
-	sort.Slice(owners, func(i, j int) bool {
-		return owners[i] < owners[j]
-	})
+
+	ring := lock_manager.NewHashRing(lock_manager.DefaultVnodeCount)
+	ring.SetServers(owners)

 	keysByOwner := make(map[pb.ServerAddress]string, len(owners))
 	for i := 0; i < 1024 && len(keysByOwner) < len(owners); i++ {
 		key := fmt.Sprintf("%s-%03d.txt", prefix, i)
-		lockOwner := ownerForObjectLock(bucket, key, owners)
+		lockKey := fmt.Sprintf("s3.object.write:/buckets/%s/%s", bucket, s3_constants.NormalizeObjectKey(key))
+		lockOwner := ring.GetPrimary(lockKey)
 		if _, exists := keysByOwner[lockOwner]; !exists {
 			keysByOwner[lockOwner] = key
 		}
@@ -164,15 +164,6 @@ func (c *distributedLockCluster) findLockOwnerKeys(bucket, prefix string) map[pb
 	return keysByOwner
 }

-func ownerForObjectLock(bucket, object string, owners []pb.ServerAddress) pb.ServerAddress {
-	lockKey := fmt.Sprintf("s3.object.write:/buckets/%s/%s", bucket, s3_constants.NormalizeObjectKey(object))
-	hash := util.HashStringToLong(lockKey)
-	if hash < 0 {
-		hash = -hash
-	}
-	return owners[hash%int64(len(owners))]
-}
-
 func lockOwnerLabel(owner pb.ServerAddress) string {
 	replacer := strings.NewReplacer(":", "_", ".", "_")
 	return "owner_" + replacer.Replace(string(owner))
@@ -0,0 +1,511 @@
+package iam
+
+import (
+	"encoding/xml"
+	"fmt"
+	"io"
+	"net/http"
+	"net/url"
+	"os"
+	"strings"
+	"testing"
+	"time"
+
+	"github.com/aws/aws-sdk-go/aws"
+	"github.com/aws/aws-sdk-go/aws/credentials"
+	"github.com/aws/aws-sdk-go/aws/session"
+	"github.com/aws/aws-sdk-go/service/s3"
+	v4 "github.com/aws/aws-sdk-go/aws/signer/v4"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+// GetFederationTokenTestResponse represents the STS GetFederationToken response
+type GetFederationTokenTestResponse struct {
+	XMLName xml.Name `xml:"GetFederationTokenResponse"`
+	Result  struct {
+		Credentials struct {
+			AccessKeyId     string `xml:"AccessKeyId"`
+			SecretAccessKey string `xml:"SecretAccessKey"`
+			SessionToken    string `xml:"SessionToken"`
+			Expiration      string `xml:"Expiration"`
+		} `xml:"Credentials"`
+		FederatedUser struct {
+			FederatedUserId string `xml:"FederatedUserId"`
+			Arn             string `xml:"Arn"`
+		} `xml:"FederatedUser"`
+	} `xml:"GetFederationTokenResult"`
+}
+
+func getTestCredentials() (string, string) {
+	accessKey := os.Getenv("STS_TEST_ACCESS_KEY")
+	if accessKey == "" {
+		accessKey = "admin"
+	}
+	secretKey := os.Getenv("STS_TEST_SECRET_KEY")
+	if secretKey == "" {
+		secretKey = "admin"
+	}
+	return accessKey, secretKey
+}
+
+// isGetFederationTokenImplemented checks if the running server supports GetFederationToken
+func isGetFederationTokenImplemented(t *testing.T) bool {
+	accessKey, secretKey := getTestCredentials()
+	resp, err := callSTSAPIWithSigV4(t, url.Values{
+		"Action":  {"GetFederationToken"},
+		"Version": {"2011-06-15"},
+		"Name":    {"probe"},
+	}, accessKey, secretKey)
+	if err != nil {
+		return false
+	}
+	defer resp.Body.Close()
+
+	body, _ := io.ReadAll(resp.Body)
+	var errResp STSErrorTestResponse
+	if xml.Unmarshal(body, &errResp) == nil {
+		if errResp.Error.Code == "InvalidAction" || errResp.Error.Code == "NotImplemented" {
+			return false
+		}
+	}
+	return true
+}
+
+// TestSTSGetFederationTokenValidation tests input validation for the GetFederationToken endpoint
+func TestSTSGetFederationTokenValidation(t *testing.T) {
+	if testing.Short() {
+		t.Skip("Skipping integration test in short mode")
+	}
+
+	if !isSTSEndpointRunning(t) {
+		t.Fatal("SeaweedFS STS endpoint is not running at", TestSTSEndpoint, "- please run 'make setup-all-tests' first")
+	}
+
+	if !isGetFederationTokenImplemented(t) {
+		t.Fatal("GetFederationToken action is not implemented in the running server")
+	}
+
+	accessKey, secretKey := getTestCredentials()
+
+	t.Run("missing_name", func(t *testing.T) {
+		resp, err := callSTSAPIWithSigV4(t, url.Values{
+			"Action":  {"GetFederationToken"},
+			"Version": {"2011-06-15"},
+			// Name is missing
+		}, accessKey, secretKey)
+		require.NoError(t, err)
+		defer resp.Body.Close()
+
+		body, _ := io.ReadAll(resp.Body)
+		var errResp STSErrorTestResponse
+		require.NoError(t, xml.Unmarshal(body, &errResp), "Failed to parse: %s", string(body))
+		assert.Equal(t, "MissingParameter", errResp.Error.Code)
+	})
+
+	t.Run("name_too_short", func(t *testing.T) {
+		resp, err := callSTSAPIWithSigV4(t, url.Values{
+			"Action":  {"GetFederationToken"},
+			"Version": {"2011-06-15"},
+			"Name":    {"A"},
+		}, accessKey, secretKey)
+		require.NoError(t, err)
+		defer resp.Body.Close()
+
+		body, _ := io.ReadAll(resp.Body)
+		var errResp STSErrorTestResponse
+		require.NoError(t, xml.Unmarshal(body, &errResp), "Failed to parse: %s", string(body))
+		assert.Equal(t, "InvalidParameterValue", errResp.Error.Code)
+	})
+
+	t.Run("name_too_long", func(t *testing.T) {
+		resp, err := callSTSAPIWithSigV4(t, url.Values{
+			"Action":  {"GetFederationToken"},
+			"Version": {"2011-06-15"},
+			"Name":    {strings.Repeat("A", 33)},
+		}, accessKey, secretKey)
+		require.NoError(t, err)
+		defer resp.Body.Close()
+
+		body, _ := io.ReadAll(resp.Body)
+		var errResp STSErrorTestResponse
+		require.NoError(t, xml.Unmarshal(body, &errResp), "Failed to parse: %s", string(body))
+		assert.Equal(t, "InvalidParameterValue", errResp.Error.Code)
+	})
+
+	t.Run("name_invalid_characters", func(t *testing.T) {
+		resp, err := callSTSAPIWithSigV4(t, url.Values{
+			"Action":  {"GetFederationToken"},
+			"Version": {"2011-06-15"},
+			"Name":    {"bad name"},
+		}, accessKey, secretKey)
+		require.NoError(t, err)
+		defer resp.Body.Close()
+
+		body, _ := io.ReadAll(resp.Body)
+		var errResp STSErrorTestResponse
+		require.NoError(t, xml.Unmarshal(body, &errResp), "Failed to parse: %s", string(body))
+		assert.Equal(t, "InvalidParameterValue", errResp.Error.Code)
+	})
+
+	t.Run("duration_too_short", func(t *testing.T) {
+		resp, err := callSTSAPIWithSigV4(t, url.Values{
+			"Action":          {"GetFederationToken"},
+			"Version":         {"2011-06-15"},
+			"Name":            {"TestApp"},
+			"DurationSeconds": {"100"},
+		}, accessKey, secretKey)
+		require.NoError(t, err)
+		defer resp.Body.Close()
+
+		body, _ := io.ReadAll(resp.Body)
+		var errResp STSErrorTestResponse
+		require.NoError(t, xml.Unmarshal(body, &errResp), "Failed to parse: %s", string(body))
+		assert.Equal(t, "InvalidParameterValue", errResp.Error.Code)
+	})
+
+	t.Run("duration_too_long", func(t *testing.T) {
+		resp, err := callSTSAPIWithSigV4(t, url.Values{
+			"Action":          {"GetFederationToken"},
+			"Version":         {"2011-06-15"},
+			"Name":            {"TestApp"},
+			"DurationSeconds": {"200000"},
+		}, accessKey, secretKey)
+		require.NoError(t, err)
+		defer resp.Body.Close()
+
+		body, _ := io.ReadAll(resp.Body)
+		var errResp STSErrorTestResponse
+		require.NoError(t, xml.Unmarshal(body, &errResp), "Failed to parse: %s", string(body))
+		assert.Equal(t, "InvalidParameterValue", errResp.Error.Code)
+	})
+
+	t.Run("malformed_policy", func(t *testing.T) {
+		resp, err := callSTSAPIWithSigV4(t, url.Values{
+			"Action":  {"GetFederationToken"},
+			"Version": {"2011-06-15"},
+			"Name":    {"TestApp"},
+			"Policy":  {"not-valid-json"},
+		}, accessKey, secretKey)
+		require.NoError(t, err)
+		defer resp.Body.Close()
+
+		body, _ := io.ReadAll(resp.Body)
+		var errResp STSErrorTestResponse
+		require.NoError(t, xml.Unmarshal(body, &errResp), "Failed to parse: %s", string(body))
+		assert.Equal(t, "MalformedPolicyDocument", errResp.Error.Code)
+	})
+
+	t.Run("anonymous_rejected", func(t *testing.T) {
+		// GetFederationToken requires SigV4, anonymous should fail
+		resp, err := callSTSAPI(t, url.Values{
+			"Action":  {"GetFederationToken"},
+			"Version": {"2011-06-15"},
+			"Name":    {"TestApp"},
+		})
+		require.NoError(t, err)
+		defer resp.Body.Close()
+
+		assert.NotEqual(t, http.StatusOK, resp.StatusCode)
+	})
+}
+
+// TestSTSGetFederationTokenRejectTemporaryCredentials tests that temporary
+// credentials (session tokens) are rejected by GetFederationToken
+func TestSTSGetFederationTokenRejectTemporaryCredentials(t *testing.T) {
+	if testing.Short() {
+		t.Skip("Skipping integration test in short mode")
+	}
+
+	if !isSTSEndpointRunning(t) {
+		t.Skip("SeaweedFS STS endpoint is not running at", TestSTSEndpoint)
+	}
+
+	if !isGetFederationTokenImplemented(t) {
+		t.Skip("GetFederationToken not implemented")
+	}
+
+	accessKey, secretKey := getTestCredentials()
+
+	// First, obtain temporary credentials via AssumeRole
+	resp, err := callSTSAPIWithSigV4(t, url.Values{
+		"Action":          {"AssumeRole"},
+		"Version":         {"2011-06-15"},
+		"RoleArn":         {"arn:aws:iam::role/admin"},
+		"RoleSessionName": {"temp-session"},
+	}, accessKey, secretKey)
+	require.NoError(t, err)
+	defer resp.Body.Close()
+
+	body, err := io.ReadAll(resp.Body)
+	require.NoError(t, err)
+
+	if resp.StatusCode != http.StatusOK {
+		t.Skipf("AssumeRole failed (may not be configured): status=%d body=%s", resp.StatusCode, string(body))
+	}
+
+	var assumeResp AssumeRoleTestResponse
+	require.NoError(t, xml.Unmarshal(body, &assumeResp), "Parse AssumeRole response: %s", string(body))
+
+	tempAccessKey := assumeResp.Result.Credentials.AccessKeyId
+	tempSecretKey := assumeResp.Result.Credentials.SecretAccessKey
+	tempSessionToken := assumeResp.Result.Credentials.SessionToken
+	require.NotEmpty(t, tempAccessKey)
+	require.NotEmpty(t, tempSessionToken)
+
+	// Now try GetFederationToken with the temporary credentials
+	// Include X-Amz-Security-Token header which marks this as a temp credential call
+	params := url.Values{
+		"Action":  {"GetFederationToken"},
+		"Version": {"2011-06-15"},
+		"Name":    {"ShouldFail"},
+	}
+
+	reqBody := params.Encode()
+	req, err := http.NewRequest(http.MethodPost, TestSTSEndpoint+"/", strings.NewReader(reqBody))
+	require.NoError(t, err)
+
+	req.Header.Set("Content-Type", "application/x-www-form-urlencoded")
+	req.Header.Set("X-Amz-Security-Token", tempSessionToken)
+
+	creds := credentials.NewStaticCredentials(tempAccessKey, tempSecretKey, tempSessionToken)
+	signer := v4.NewSigner(creds)
+	_, err = signer.Sign(req, strings.NewReader(reqBody), "sts", "us-east-1", time.Now())
+	require.NoError(t, err)
+
+	client := &http.Client{Timeout: 30 * time.Second}
+	resp2, err := client.Do(req)
+	require.NoError(t, err)
+	defer resp2.Body.Close()
+
+	body2, _ := io.ReadAll(resp2.Body)
+	assert.Equal(t, http.StatusForbidden, resp2.StatusCode,
+		"GetFederationToken should reject temporary credentials: %s", string(body2))
+	assert.Contains(t, string(body2), "temporary credentials",
+		"Error should mention temporary credentials")
+}
+
+// TestSTSGetFederationTokenSuccess tests a successful GetFederationToken call
+// and verifies the returned credentials can be used to access S3
+func TestSTSGetFederationTokenSuccess(t *testing.T) {
+	if testing.Short() {
+		t.Skip("Skipping integration test in short mode")
+	}
+
+	if !isSTSEndpointRunning(t) {
+		t.Skip("SeaweedFS STS endpoint is not running at", TestSTSEndpoint)
+	}
+
+	if !isGetFederationTokenImplemented(t) {
+		t.Skip("GetFederationToken not implemented")
+	}
+
+	accessKey, secretKey := getTestCredentials()
+
+	t.Run("basic_success", func(t *testing.T) {
+		resp, err := callSTSAPIWithSigV4(t, url.Values{
+			"Action":  {"GetFederationToken"},
+			"Version": {"2011-06-15"},
+			"Name":    {"AppClient"},
+		}, accessKey, secretKey)
+		require.NoError(t, err)
+		defer resp.Body.Close()
+
+		body, err := io.ReadAll(resp.Body)
+		require.NoError(t, err)
+		t.Logf("Response status: %d, body: %s", resp.StatusCode, string(body))
+
+		if resp.StatusCode != http.StatusOK {
+			var errResp STSErrorTestResponse
+			_ = xml.Unmarshal(body, &errResp)
+			t.Fatalf("GetFederationToken failed: code=%s message=%s", errResp.Error.Code, errResp.Error.Message)
+		}
+
+		var stsResp GetFederationTokenTestResponse
+		require.NoError(t, xml.Unmarshal(body, &stsResp), "Parse response: %s", string(body))
+
+		creds := stsResp.Result.Credentials
+		assert.NotEmpty(t, creds.AccessKeyId)
+		assert.NotEmpty(t, creds.SecretAccessKey)
+		assert.NotEmpty(t, creds.SessionToken)
+		assert.NotEmpty(t, creds.Expiration)
+
+		fedUser := stsResp.Result.FederatedUser
+		assert.Contains(t, fedUser.Arn, "federated-user/AppClient")
+		assert.Contains(t, fedUser.FederatedUserId, "AppClient")
+	})
+
+	t.Run("with_custom_duration", func(t *testing.T) {
+		resp, err := callSTSAPIWithSigV4(t, url.Values{
+			"Action":          {"GetFederationToken"},
+			"Version":         {"2011-06-15"},
+			"Name":            {"DurationTest"},
+			"DurationSeconds": {"3600"},
+		}, accessKey, secretKey)
+		require.NoError(t, err)
+		defer resp.Body.Close()
+
+		body, _ := io.ReadAll(resp.Body)
+		t.Logf("Response status: %d, body: %s", resp.StatusCode, string(body))
+
+		if resp.StatusCode == http.StatusOK {
+			var stsResp GetFederationTokenTestResponse
+			require.NoError(t, xml.Unmarshal(body, &stsResp))
+			assert.NotEmpty(t, stsResp.Result.Credentials.AccessKeyId)
+
+			// Verify expiration is roughly 1 hour from now
+			expTime, err := time.Parse(time.RFC3339, stsResp.Result.Credentials.Expiration)
+			require.NoError(t, err)
+			diff := time.Until(expTime)
+			assert.InDelta(t, 3600, diff.Seconds(), 60,
+				"Expiration should be ~1 hour from now")
+		}
+	})
+
+	t.Run("with_36_hour_duration", func(t *testing.T) {
+		// GetFederationToken allows up to 36 hours (unlike AssumeRole's 12h max)
+		resp, err := callSTSAPIWithSigV4(t, url.Values{
+			"Action":          {"GetFederationToken"},
+			"Version":         {"2011-06-15"},
+			"Name":            {"LongDuration"},
+			"DurationSeconds": {"129600"}, // 36 hours
+		}, accessKey, secretKey)
+		require.NoError(t, err)
+		defer resp.Body.Close()
+
+		body, _ := io.ReadAll(resp.Body)
+		if resp.StatusCode == http.StatusOK {
+			var stsResp GetFederationTokenTestResponse
+			require.NoError(t, xml.Unmarshal(body, &stsResp))
+
+			expTime, err := time.Parse(time.RFC3339, stsResp.Result.Credentials.Expiration)
+			require.NoError(t, err)
+			diff := time.Until(expTime)
+			assert.InDelta(t, 129600, diff.Seconds(), 60,
+				"Expiration should be ~36 hours from now")
+		} else {
+			// Duration should not cause a rejection
+			var errResp STSErrorTestResponse
+			_ = xml.Unmarshal(body, &errResp)
+			assert.NotContains(t, errResp.Error.Message, "DurationSeconds",
+				"36-hour duration should be accepted by GetFederationToken")
+		}
+	})
+}
+
+// TestSTSGetFederationTokenWithSessionPolicy tests that vended credentials
+// are scoped down by an inline session policy
+func TestSTSGetFederationTokenWithSessionPolicy(t *testing.T) {
+	if testing.Short() {
+		t.Skip("Skipping integration test in short mode")
+	}
+
+	if !isSTSEndpointRunning(t) {
+		t.Skip("SeaweedFS STS endpoint is not running at", TestSTSEndpoint)
+	}
+
+	if !isGetFederationTokenImplemented(t) {
+		t.Skip("GetFederationToken not implemented")
+	}
+
+	accessKey, secretKey := getTestCredentials()
+
+	// Create a test bucket using admin credentials
+	adminSess, err := session.NewSession(&aws.Config{
+		Region:           aws.String("us-east-1"),
+		Endpoint:         aws.String(TestSTSEndpoint),
+		DisableSSL:       aws.Bool(true),
+		S3ForcePathStyle: aws.Bool(true),
+		Credentials:      credentials.NewStaticCredentials(accessKey, secretKey, ""),
+	})
+	require.NoError(t, err)
+
+	adminS3 := s3.New(adminSess)
+	bucket := fmt.Sprintf("fed-token-test-%d", time.Now().UnixNano())
+
+	_, err = adminS3.CreateBucket(&s3.CreateBucketInput{Bucket: aws.String(bucket)})
+	require.NoError(t, err)
+	defer adminS3.DeleteBucket(&s3.DeleteBucketInput{Bucket: aws.String(bucket)})
+
+	_, err = adminS3.PutObject(&s3.PutObjectInput{
+		Bucket: aws.String(bucket),
+		Key:    aws.String("test.txt"),
+		Body:   strings.NewReader("hello"),
+	})
+	require.NoError(t, err)
+	defer adminS3.DeleteObject(&s3.DeleteObjectInput{Bucket: aws.String(bucket), Key: aws.String("test.txt")})
+
+	// Get federated credentials with a session policy that only allows GetObject
+	sessionPolicy := fmt.Sprintf(`{
+		"Version": "2012-10-17",
+		"Statement": [{
+			"Effect": "Allow",
+			"Action": ["s3:GetObject"],
+			"Resource": ["arn:aws:s3:::%s/*"]
+		}]
+	}`, bucket)
+
+	resp, err := callSTSAPIWithSigV4(t, url.Values{
+		"Action":  {"GetFederationToken"},
+		"Version": {"2011-06-15"},
+		"Name":    {"ScopedClient"},
+		"Policy":  {sessionPolicy},
+	}, accessKey, secretKey)
+	require.NoError(t, err)
+	defer resp.Body.Close()
+
+	body, _ := io.ReadAll(resp.Body)
+	t.Logf("GetFederationToken response: status=%d body=%s", resp.StatusCode, string(body))
+
+	if resp.StatusCode != http.StatusOK {
+		t.Skipf("GetFederationToken failed (may need IAM policy config): %s", string(body))
+	}
+
+	var stsResp GetFederationTokenTestResponse
+	require.NoError(t, xml.Unmarshal(body, &stsResp))
+
+	fedCreds := stsResp.Result.Credentials
+	require.NotEmpty(t, fedCreds.AccessKeyId)
+	require.NotEmpty(t, fedCreds.SessionToken)
+
+	// Create S3 client with the federated credentials
+	fedSess, err := session.NewSession(&aws.Config{
+		Region:           aws.String("us-east-1"),
+		Endpoint:         aws.String(TestSTSEndpoint),
+		DisableSSL:       aws.Bool(true),
+		S3ForcePathStyle: aws.Bool(true),
+		Credentials: credentials.NewStaticCredentials(
+			fedCreds.AccessKeyId, fedCreds.SecretAccessKey, fedCreds.SessionToken),
+	})
+	require.NoError(t, err)
+
+	fedS3 := s3.New(fedSess)
+
+	// GetObject should succeed (allowed by session policy)
+	getResp, err := fedS3.GetObject(&s3.GetObjectInput{
+		Bucket: aws.String(bucket),
+		Key:    aws.String("test.txt"),
+	})
+	if err == nil {
+		defer getResp.Body.Close()
+		t.Log("GetObject with federated credentials succeeded (as expected)")
+	} else {
+		t.Logf("GetObject with federated credentials: %v (session policy enforcement may vary)", err)
+	}
+
+	// PutObject should be denied (not allowed by session policy)
+	_, err = fedS3.PutObject(&s3.PutObjectInput{
+		Bucket: aws.String(bucket),
+		Key:    aws.String("denied.txt"),
+		Body:   strings.NewReader("should fail"),
+	})
+	if err != nil {
+		t.Log("PutObject correctly denied with federated credentials")
+		assert.Contains(t, err.Error(), "AccessDenied",
+			"PutObject should be denied by session policy")
+	} else {
+		// Clean up if unexpectedly succeeded
+		adminS3.DeleteObject(&s3.DeleteObjectInput{Bucket: aws.String(bucket), Key: aws.String("denied.txt")})
+		t.Log("PutObject unexpectedly succeeded — session policy enforcement may not be active")
+	}
+}
@@ -98,8 +98,8 @@ start-seaweedfs: check-binary
 	# Create S3 configuration with SSE-KMS support
 	@printf '{"identities":[{"name":"%s","credentials":[{"accessKey":"%s","secretKey":"%s"}],"actions":["Admin","Read","Write"]}],"kms":{"type":"%s","configs":{"keyId":"%s","encryptionContext":{},"bucketKey":false}}}' "$(ACCESS_KEY)" "$(ACCESS_KEY)" "$(SECRET_KEY)" "$(KMS_TYPE)" "$(KMS_KEY_ID)" > /tmp/seaweedfs-sse-s3.json
 	
-	# Start weed mini
-	@AWS_ACCESS_KEY_ID=$(ACCESS_KEY) AWS_SECRET_ACCESS_KEY=$(SECRET_KEY) $(SEAWEEDFS_BINARY) mini \
+	# Start weed mini (WEED_S3_SSE_KEY enables SSE-S3 encryption)
+	@AWS_ACCESS_KEY_ID=$(ACCESS_KEY) AWS_SECRET_ACCESS_KEY=$(SECRET_KEY) WEED_S3_SSE_KEY=test-sse-s3-key $(SEAWEEDFS_BINARY) mini \
 		-dir=/tmp/seaweedfs-test-sse \
 		-s3.port=$(S3_PORT) \
 		-s3.config=/tmp/seaweedfs-sse-s3.json \
@@ -337,8 +337,9 @@ start-seaweedfs-ci: check-binary
 	     s3-config-template.json > /tmp/seaweedfs-s3.json
 	
 	# Start weed mini with embedded S3 using the JSON config (with verbose logging)
+	# WEED_S3_SSE_KEY enables SSE-S3 encryption for testing (KEK derived via HKDF)
 	@echo "Starting weed mini with embedded S3..."
-	@AWS_ACCESS_KEY_ID=$(ACCESS_KEY) AWS_SECRET_ACCESS_KEY=$(SECRET_KEY) GLOG_v=4 $(SEAWEEDFS_BINARY) mini \
+	@AWS_ACCESS_KEY_ID=$(ACCESS_KEY) AWS_SECRET_ACCESS_KEY=$(SECRET_KEY) WEED_S3_SSE_KEY=test-sse-s3-key GLOG_v=4 $(SEAWEEDFS_BINARY) mini \
 		-dir=/tmp/seaweedfs-test-sse \
 		-s3.port=$(S3_PORT) \
 		-s3.config=/tmp/seaweedfs-s3.json \
@@ -482,7 +483,7 @@ test-volume-encryption: build-weed
 	     -e 's/SECRET_KEY_PLACEHOLDER/$(SECRET_KEY)/g' \
 	     s3-config-template.json > /tmp/seaweedfs-s3.json
 	@echo "Starting weed mini with S3 volume encryption..."
-	@AWS_ACCESS_KEY_ID=$(ACCESS_KEY) AWS_SECRET_ACCESS_KEY=$(SECRET_KEY) GLOG_v=4 $(SEAWEEDFS_BINARY) mini \
+	@AWS_ACCESS_KEY_ID=$(ACCESS_KEY) AWS_SECRET_ACCESS_KEY=$(SECRET_KEY) WEED_S3_SSE_KEY=test-sse-s3-key GLOG_v=4 $(SEAWEEDFS_BINARY) mini \
 		-dir=/tmp/seaweedfs-test-sse \
 		-s3.port=$(S3_PORT) \
 		-s3.config=/tmp/seaweedfs-s3.json \
@@ -289,6 +289,157 @@ func TestVersioningPaginationMultipleObjectsManyVersions(t *testing.T) {
 	})
 }

+// TestVersioningPaginationDeepDirectoryHierarchy tests that paginated ListObjectVersions
+// correctly skips directory subtrees before the key-marker. This reproduces the
+// real-world scenario where Veeam backup objects are spread across many subdirectories
+// (e.g., Mailboxes/<uuid>/ItemsData/<file>) and pagination becomes exponentially
+// slower as the marker advances through the tree.
+//
+// Run with: ENABLE_STRESS_TESTS=true go test -v -run TestVersioningPaginationDeepDirectoryHierarchy -timeout 10m
+func TestVersioningPaginationDeepDirectoryHierarchy(t *testing.T) {
+	if os.Getenv("ENABLE_STRESS_TESTS") != "true" {
+		t.Skip("Skipping stress test. Set ENABLE_STRESS_TESTS=true to run.")
+	}
+
+	client := getS3Client(t)
+	bucketName := getNewBucketName()
+
+	createBucket(t, client, bucketName)
+	defer deleteBucket(t, client, bucketName)
+
+	enableVersioning(t, client, bucketName)
+	checkVersioningStatus(t, client, bucketName, types.BucketVersioningStatusEnabled)
+
+	// Create a deep directory structure mimicking Veeam 365 backup layout:
+	// Backup/Organizations/<org>/Mailboxes/<mailbox>/ItemsData/<file>
+	numMailboxes := 20
+	filesPerMailbox := 5
+	totalObjects := numMailboxes * filesPerMailbox
+	orgPrefix := "Backup/Organizations/org-001/Mailboxes"
+
+	t.Logf("Creating %d objects across %d subdirectories (depth=6)...", totalObjects, numMailboxes)
+	startTime := time.Now()
+
+	allKeys := make([]string, 0, totalObjects)
+	for i := 0; i < numMailboxes; i++ {
+		mailboxId := fmt.Sprintf("mbx-%03d", i)
+		for j := 0; j < filesPerMailbox; j++ {
+			key := fmt.Sprintf("%s/%s/ItemsData/file-%03d.dat", orgPrefix, mailboxId, j)
+			_, err := client.PutObject(context.TODO(), &s3.PutObjectInput{
+				Bucket: aws.String(bucketName),
+				Key:    aws.String(key),
+				Body:   strings.NewReader(fmt.Sprintf("content-%d-%d", i, j)),
+			})
+			require.NoError(t, err)
+			allKeys = append(allKeys, key)
+		}
+	}
+	t.Logf("Created %d objects in %v", totalObjects, time.Since(startTime))
+
+	// Test 1: Paginate through all versions with a broad prefix and small maxKeys.
+	// This forces multiple pages that must skip earlier subdirectory trees.
+	t.Run("PaginateAcrossSubdirectories", func(t *testing.T) {
+		maxKeys := int32(10) // Force many pages to exercise marker skipping
+		var allVersions []types.ObjectVersion
+		var keyMarker, versionIdMarker *string
+		pageCount := 0
+		pageStartTimes := make([]time.Duration, 0)
+
+		for {
+			pageStart := time.Now()
+			resp, err := client.ListObjectVersions(context.TODO(), &s3.ListObjectVersionsInput{
+				Bucket:          aws.String(bucketName),
+				Prefix:          aws.String(orgPrefix + "/"),
+				MaxKeys:         aws.Int32(maxKeys),
+				KeyMarker:       keyMarker,
+				VersionIdMarker: versionIdMarker,
+			})
+			pageDuration := time.Since(pageStart)
+			require.NoError(t, err)
+			pageCount++
+			pageStartTimes = append(pageStartTimes, pageDuration)
+
+			allVersions = append(allVersions, resp.Versions...)
+			t.Logf("Page %d: %d versions in %v (marker: %v)",
+				pageCount, len(resp.Versions), pageDuration,
+				keyMarker)
+
+			if resp.IsTruncated == nil || !*resp.IsTruncated {
+				break
+			}
+			keyMarker = resp.NextKeyMarker
+			versionIdMarker = resp.NextVersionIdMarker
+		}
+
+		assert.Greater(t, pageCount, 1, "Should require multiple pages")
+
+		// Verify listed keys exactly match the keys we created (same elements, same order)
+		listedKeys := make([]string, 0, len(allVersions))
+		for _, v := range allVersions {
+			listedKeys = append(listedKeys, *v.Key)
+		}
+		assert.Equal(t, allKeys, listedKeys,
+			"Listed version keys should exactly match created keys")
+
+		// Check that later pages don't take dramatically longer than earlier ones.
+		// Before the fix, later pages were exponentially slower because they
+		// re-traversed the entire tree. With the fix, all pages should be similar.
+		if len(pageStartTimes) >= 4 {
+			firstQuarter := pageStartTimes[0]
+			lastQuarter := pageStartTimes[len(pageStartTimes)-1]
+			t.Logf("First page: %v, Last page: %v, Ratio: %.1fx",
+				firstQuarter, lastQuarter,
+				float64(lastQuarter)/float64(firstQuarter))
+			// Allow generous 10x ratio to avoid flakiness; before the fix
+			// the ratio was 100x+ on large datasets
+			if lastQuarter > firstQuarter*10 && lastQuarter > 500*time.Millisecond {
+				t.Errorf("Last page took %.1fx longer than first page (%v vs %v) — possible pagination regression",
+					float64(lastQuarter)/float64(firstQuarter), lastQuarter, firstQuarter)
+			}
+		}
+	})
+
+	// Test 2: Paginate with delimiter to verify CommonPrefixes interaction
+	t.Run("PaginateWithDelimiterAcrossSubdirs", func(t *testing.T) {
+		maxKeys := int32(5)
+		var allPrefixes []string
+		var keyMarker *string
+		pageCount := 0
+
+		for {
+			resp, err := client.ListObjectVersions(context.TODO(), &s3.ListObjectVersionsInput{
+				Bucket:    aws.String(bucketName),
+				Prefix:    aws.String(orgPrefix + "/"),
+				Delimiter: aws.String("/"),
+				MaxKeys:   aws.Int32(maxKeys),
+				KeyMarker: keyMarker,
+			})
+			require.NoError(t, err)
+			pageCount++
+
+			for _, cp := range resp.CommonPrefixes {
+				allPrefixes = append(allPrefixes, *cp.Prefix)
+			}
+
+			if resp.IsTruncated == nil || !*resp.IsTruncated {
+				break
+			}
+			keyMarker = resp.NextKeyMarker
+		}
+
+		assert.Greater(t, pageCount, 1, "Should require multiple pages with maxKeys=%d", maxKeys)
+
+		// Build the exact expected prefixes
+		expectedPrefixes := make([]string, 0, numMailboxes)
+		for i := 0; i < numMailboxes; i++ {
+			expectedPrefixes = append(expectedPrefixes,
+				fmt.Sprintf("%s/mbx-%03d/", orgPrefix, i))
+		}
+		assert.Equal(t, expectedPrefixes, allPrefixes,
+			"CommonPrefixes should exactly match expected mailbox prefixes")
+	})
+}
+
 // listAllVersions is a helper to list all versions of a specific object using pagination
 func listAllVersions(t *testing.T, client *s3.Client, bucketName, objectKey string) []types.ObjectVersion {
 	var allVersions []types.ObjectVersion
@@ -115,10 +115,13 @@ func NewTestEnvironment(t *testing.T) *TestEnvironment {

 	bindIP := testutil.FindBindIP()

-	masterPort, masterGrpcPort := testutil.MustFreePortPair(t, "Master")
-	volumePort, volumeGrpcPort := testutil.MustFreePortPair(t, "Volume")
-	filerPort, filerGrpcPort := testutil.MustFreePortPair(t, "Filer")
-	s3Port, s3GrpcPort := testutil.MustFreePortPair(t, "S3")
+	// Allocate all ports in a single batch to prevent the OS from recycling
+	// a released port, which can cause two services to get the same port.
+	ports := testutil.MustAllocatePorts(t, 8)
+	masterPort, masterGrpcPort := ports[0], ports[1]
+	volumePort, volumeGrpcPort := ports[2], ports[3]
+	filerPort, filerGrpcPort := ports[4], ports[5]
+	s3Port, s3GrpcPort := ports[6], ports[7]

 	return &TestEnvironment{
 		seaweedDir:     seaweedDir,
@@ -89,11 +89,14 @@ func NewTestEnvironment(t *testing.T) *TestEnvironment {

 	bindIP := testutil.FindBindIP()

-	masterPort, masterGrpcPort := testutil.MustFreePortPair(t, "Master")
-	volumePort, volumeGrpcPort := testutil.MustFreePortPair(t, "Volume")
-	filerPort, filerGrpcPort := testutil.MustFreePortPair(t, "Filer")
-	s3Port, s3GrpcPort := testutil.MustFreePortPair(t, "S3")
-	polarisPort, polarisAdminPort := testutil.MustFreePortPair(t, "Polaris")
+	// Allocate all ports in a single batch to prevent the OS from recycling
+	// a released port, which can cause two services to get the same port.
+	ports := testutil.MustAllocatePorts(t, 10)
+	masterPort, masterGrpcPort := ports[0], ports[1]
+	volumePort, volumeGrpcPort := ports[2], ports[3]
+	filerPort, filerGrpcPort := ports[4], ports[5]
+	s3Port, s3GrpcPort := ports[6], ports[7]
+	polarisPort, polarisAdminPort := ports[8], ports[9]

 	return &TestEnvironment{
 		seaweedDir:       seaweedDir,
@@ -93,10 +93,13 @@ func NewTestEnvironment(t *testing.T) *TestEnvironment {

 	bindIP := testutil.FindBindIP()

-	masterPort, masterGrpcPort := testutil.MustFreePortPair(t, "Master")
-	volumePort, volumeGrpcPort := testutil.MustFreePortPair(t, "Volume")
-	filerPort, filerGrpcPort := testutil.MustFreePortPair(t, "Filer")
-	s3Port, s3GrpcPort := testutil.MustFreePortPair(t, "S3") // Changed to use testutil.MustFreePortPair
+	// Allocate all ports in a single batch to prevent the OS from recycling
+	// a released port, which can cause two services to get the same port.
+	ports := testutil.MustAllocatePorts(t, 8)
+	masterPort, masterGrpcPort := ports[0], ports[1]
+	volumePort, volumeGrpcPort := ports[2], ports[3]
+	filerPort, filerGrpcPort := ports[4], ports[5]
+	s3Port, s3GrpcPort := ports[6], ports[7]

 	return &TestEnvironment{
 		seaweedDir:      seaweedDir,
@@ -15,33 +15,44 @@ func HasDocker() bool {
 	return cmd.Run() == nil
 }

+// MustFreePortPair is a convenience wrapper for tests that only need a single pair.
+// Prefer MustAllocatePorts when allocating multiple pairs to guarantee uniqueness.
 func MustFreePortPair(t *testing.T, name string) (int, int) {
-	httpPort, grpcPort, err := findAvailablePortPair()
-	if err != nil {
-		t.Fatalf("Failed to get free port pair for %s: %v", name, err)
-	}
-	return httpPort, grpcPort
+	ports := MustAllocatePorts(t, 2)
+	return ports[0], ports[1]
 }

-func findAvailablePortPair() (int, int, error) {
-	httpPort, err := GetFreePort()
+// MustAllocatePorts allocates count unique free ports atomically.
+// All listeners are held open until every port is obtained, preventing
+// the OS from recycling a port between successive allocations.
+func MustAllocatePorts(t *testing.T, count int) []int {
+	t.Helper()
+	ports, err := AllocatePorts(count)
 	if err != nil {
-		return 0, 0, err
+		t.Fatalf("Failed to allocate %d free ports: %v", count, err)
 	}
-	grpcPort, err := GetFreePort()
-	if err != nil {
-		return 0, 0, err
-	}
-	return httpPort, grpcPort, nil
+	return ports
 }

-func GetFreePort() (int, error) {
-	listener, err := net.Listen("tcp", "0.0.0.0:0")
-	if err != nil {
-		return 0, err
+// AllocatePorts allocates count unique free ports atomically.
+func AllocatePorts(count int) ([]int, error) {
+	listeners := make([]net.Listener, 0, count)
+	ports := make([]int, 0, count)
+	for i := 0; i < count; i++ {
+		l, err := net.Listen("tcp", "0.0.0.0:0")
+		if err != nil {
+			for _, ll := range listeners {
+				_ = ll.Close()
+			}
+			return nil, err
+		}
+		listeners = append(listeners, l)
+		ports = append(ports, l.Addr().(*net.TCPAddr).Port)
 	}
-	defer listener.Close()
-	return listener.Addr().(*net.TCPAddr).Port, nil
+	for _, l := range listeners {
+		_ = l.Close()
+	}
+	return ports, nil
 }

 func WaitForService(url string, timeout time.Duration) bool {
@@ -0,0 +1,368 @@
+package framework
+
+import (
+	"fmt"
+	"net"
+	"os"
+	"os/exec"
+	"path/filepath"
+	"strconv"
+	"sync"
+	"testing"
+
+	"github.com/seaweedfs/seaweedfs/test/volume_server/matrix"
+)
+
+// MixedVolumeCluster wraps a Go master + a mix of Go and Rust volume servers
+// for integration testing. Go servers occupy indices [0, goCount) and Rust
+// servers occupy indices [goCount, goCount+rustCount).
+type MixedVolumeCluster struct {
+	testingTB testing.TB
+	profile   matrix.Profile
+
+	weedBinary       string // Go weed binary (master + Go volume servers)
+	rustVolumeBinary string // Rust volume binary
+
+	baseDir   string
+	configDir string
+	logsDir   string
+	keepLogs  bool
+
+	masterPort     int
+	masterGrpcPort int
+
+	volumePorts     []int
+	volumeGrpcPorts []int
+	volumePubPorts  []int
+	isRust          []bool // which servers are Rust
+
+	masterCmd  *exec.Cmd
+	volumeCmds []*exec.Cmd
+
+	cleanupOnce sync.Once
+}
+
+// StartMixedVolumeCluster starts a cluster with 1 Go master, goCount Go volume
+// servers, and rustCount Rust volume servers. Go servers come first in the index.
+func StartMixedVolumeCluster(t testing.TB, profile matrix.Profile, goCount, rustCount int) *MixedVolumeCluster {
+	t.Helper()
+
+	if goCount < 0 || rustCount < 0 {
+		t.Fatalf("goCount and rustCount must be non-negative, got go=%d rust=%d", goCount, rustCount)
+	}
+	total := goCount + rustCount
+	if total < 2 {
+		t.Fatalf("need at least 2 volume servers, got %d", total)
+	}
+
+	weedBinary, err := FindOrBuildWeedBinary()
+	if err != nil {
+		t.Fatalf("resolve weed binary: %v", err)
+	}
+
+	// Only build the Rust binary when Rust servers are requested.
+	var rustBinary string
+	if rustCount > 0 {
+		rustBinary, err = FindOrBuildRustBinary()
+		if err != nil {
+			t.Skipf("skipping mixed cluster test: rust binary unavailable: %v", err)
+		}
+	}
+
+	baseDir, keepLogs, err := newWorkDir()
+	if err != nil {
+		t.Fatalf("create temp test directory: %v", err)
+	}
+
+	configDir := filepath.Join(baseDir, "config")
+	logsDir := filepath.Join(baseDir, "logs")
+	masterDataDir := filepath.Join(baseDir, "master")
+
+	dirs := []string{configDir, logsDir, masterDataDir}
+	for i := 0; i < total; i++ {
+		dirs = append(dirs, filepath.Join(baseDir, fmt.Sprintf("volume%d", i)))
+	}
+	for _, dir := range dirs {
+		if mkErr := os.MkdirAll(dir, 0o755); mkErr != nil {
+			t.Fatalf("create %s: %v", dir, mkErr)
+		}
+	}
+
+	if err = writeSecurityConfig(configDir, profile); err != nil {
+		t.Fatalf("write security config: %v", err)
+	}
+
+	masterPort, masterGrpcPort, err := allocateMasterPortPair()
+	if err != nil {
+		t.Fatalf("allocate master port pair: %v", err)
+	}
+
+	// 2 ports per server (admin, grpc); add 1 more when public port is split out.
+	portsPerServer := 2
+	if profile.SplitPublicPort {
+		portsPerServer = 3
+	}
+	ports, err := allocatePorts(total * portsPerServer)
+	if err != nil {
+		t.Fatalf("allocate volume ports: %v", err)
+	}
+
+	isRust := make([]bool, total)
+	for i := goCount; i < total; i++ {
+		isRust[i] = true
+	}
+
+	c := &MixedVolumeCluster{
+		testingTB:        t,
+		profile:          profile,
+		weedBinary:       weedBinary,
+		rustVolumeBinary: rustBinary,
+		baseDir:          baseDir,
+		configDir:        configDir,
+		logsDir:          logsDir,
+		keepLogs:         keepLogs,
+		masterPort:       masterPort,
+		masterGrpcPort:   masterGrpcPort,
+		volumePorts:      make([]int, total),
+		volumeGrpcPorts:  make([]int, total),
+		volumePubPorts:   make([]int, total),
+		isRust:           isRust,
+		volumeCmds:       make([]*exec.Cmd, total),
+	}
+
+	for i := 0; i < total; i++ {
+		baseIdx := i * portsPerServer
+		c.volumePorts[i] = ports[baseIdx]
+		c.volumeGrpcPorts[i] = ports[baseIdx+1]
+		if profile.SplitPublicPort {
+			c.volumePubPorts[i] = ports[baseIdx+2]
+		} else {
+			c.volumePubPorts[i] = c.volumePorts[i] // reuse admin port
+		}
+	}
+
+	// Start master
+	if err = c.startMaster(masterDataDir); err != nil {
+		c.Stop()
+		t.Fatalf("start master: %v", err)
+	}
+	helper := &Cluster{logsDir: logsDir}
+	if err = helper.waitForHTTP(c.MasterURL() + "/dir/status"); err != nil {
+		masterLog := helper.tailLog("master.log")
+		c.Stop()
+		t.Fatalf("wait for master readiness: %v\nmaster log tail:\n%s", err, masterLog)
+	}
+
+	// Start volume servers
+	for i := 0; i < total; i++ {
+		volumeDataDir := filepath.Join(baseDir, fmt.Sprintf("volume%d", i))
+		if isRust[i] {
+			err = c.startRustVolume(i, volumeDataDir)
+		} else {
+			err = c.startGoVolume(i, volumeDataDir)
+		}
+		if err != nil {
+			logTail := helper.tailLog(fmt.Sprintf("volume%d.log", i))
+			c.Stop()
+			t.Fatalf("start volume server %d (rust=%v): %v\nlog tail:\n%s", i, isRust[i], err, logTail)
+		}
+
+		// Rust uses /healthz, Go uses /status
+		healthURL := c.VolumeAdminURL(i) + "/status"
+		if isRust[i] {
+			healthURL = c.VolumeAdminURL(i) + "/healthz"
+		}
+		if err = helper.waitForHTTP(healthURL); err != nil {
+			logTail := helper.tailLog(fmt.Sprintf("volume%d.log", i))
+			c.Stop()
+			t.Fatalf("wait for volume server %d readiness: %v\nlog tail:\n%s", i, err, logTail)
+		}
+		if err = helper.waitForTCP(c.VolumeGRPCAddress(i)); err != nil {
+			logTail := helper.tailLog(fmt.Sprintf("volume%d.log", i))
+			c.Stop()
+			t.Fatalf("wait for volume server %d grpc readiness: %v\nlog tail:\n%s", i, err, logTail)
+		}
+	}
+
+	t.Cleanup(func() {
+		c.Stop()
+	})
+
+	return c
+}
+
+func (c *MixedVolumeCluster) Stop() {
+	if c == nil {
+		return
+	}
+	c.cleanupOnce.Do(func() {
+		for i := len(c.volumeCmds) - 1; i >= 0; i-- {
+			stopProcess(c.volumeCmds[i])
+		}
+		stopProcess(c.masterCmd)
+		if !c.keepLogs && !c.testingTB.Failed() {
+			_ = os.RemoveAll(c.baseDir)
+		} else if c.baseDir != "" {
+			c.testingTB.Logf("mixed volume server integration logs kept at %s", c.baseDir)
+		}
+	})
+}
+
+func (c *MixedVolumeCluster) startMaster(dataDir string) error {
+	logFile, err := os.Create(filepath.Join(c.logsDir, "master.log"))
+	if err != nil {
+		return err
+	}
+
+	args := []string{
+		"-config_dir=" + c.configDir,
+		"master",
+		"-ip=127.0.0.1",
+		"-port=" + strconv.Itoa(c.masterPort),
+		"-port.grpc=" + strconv.Itoa(c.masterGrpcPort),
+		"-mdir=" + dataDir,
+		"-peers=none",
+		"-volumeSizeLimitMB=" + strconv.Itoa(testVolumeSizeLimitMB),
+		"-defaultReplication=000",
+	}
+
+	c.masterCmd = exec.Command(c.weedBinary, args...)
+	c.masterCmd.Dir = c.baseDir
+	c.masterCmd.Stdout = logFile
+	c.masterCmd.Stderr = logFile
+	if err = c.masterCmd.Start(); err != nil {
+		logFile.Close()
+		return err
+	}
+	logFile.Close() // child inherited the fd
+	return nil
+}
+
+func (c *MixedVolumeCluster) startGoVolume(index int, dataDir string) error {
+	logName := fmt.Sprintf("volume%d.log", index)
+	logFile, err := os.Create(filepath.Join(c.logsDir, logName))
+	if err != nil {
+		return err
+	}
+
+	args := []string{
+		"-config_dir=" + c.configDir,
+		"volume",
+		"-ip=127.0.0.1",
+		"-port=" + strconv.Itoa(c.volumePorts[index]),
+		"-port.grpc=" + strconv.Itoa(c.volumeGrpcPorts[index]),
+		"-port.public=" + strconv.Itoa(c.volumePubPorts[index]),
+		"-dir=" + dataDir,
+		"-max=16",
+		"-master=127.0.0.1:" + strconv.Itoa(c.masterPort),
+		"-readMode=" + c.profile.ReadMode,
+		"-concurrentUploadLimitMB=" + strconv.Itoa(c.profile.ConcurrentUploadLimitMB),
+		"-concurrentDownloadLimitMB=" + strconv.Itoa(c.profile.ConcurrentDownloadLimitMB),
+	}
+	if c.profile.InflightUploadTimeout > 0 {
+		args = append(args, "-inflightUploadDataTimeout="+c.profile.InflightUploadTimeout.String())
+	}
+	if c.profile.InflightDownloadTimeout > 0 {
+		args = append(args, "-inflightDownloadDataTimeout="+c.profile.InflightDownloadTimeout.String())
+	}
+
+	cmd := exec.Command(c.weedBinary, args...)
+	cmd.Dir = c.baseDir
+	cmd.Stdout = logFile
+	cmd.Stderr = logFile
+	if err = cmd.Start(); err != nil {
+		logFile.Close()
+		return err
+	}
+	logFile.Close() // child inherited the fd
+	c.volumeCmds[index] = cmd
+	return nil
+}
+
+func (c *MixedVolumeCluster) startRustVolume(index int, dataDir string) error {
+	logName := fmt.Sprintf("volume%d.log", index)
+	logFile, err := os.Create(filepath.Join(c.logsDir, logName))
+	if err != nil {
+		return err
+	}
+
+	args := rustVolumeArgs(
+		c.profile,
+		c.configDir,
+		c.masterPort,
+		c.volumePorts[index],
+		c.volumeGrpcPorts[index],
+		c.volumePubPorts[index],
+		dataDir,
+	)
+
+	cmd := exec.Command(c.rustVolumeBinary, args...)
+	cmd.Dir = c.baseDir
+	cmd.Stdout = logFile
+	cmd.Stderr = logFile
+	if err = cmd.Start(); err != nil {
+		logFile.Close()
+		return err
+	}
+	logFile.Close() // child inherited the fd
+	c.volumeCmds[index] = cmd
+	return nil
+}
+
+// --- accessor methods (mirror MultiVolumeCluster) ---
+
+func (c *MixedVolumeCluster) MasterAddress() string {
+	return net.JoinHostPort("127.0.0.1", strconv.Itoa(c.masterPort))
+}
+
+func (c *MixedVolumeCluster) MasterURL() string {
+	return "http://" + c.MasterAddress()
+}
+
+func (c *MixedVolumeCluster) VolumeAdminAddress(index int) string {
+	if index < 0 || index >= len(c.volumePorts) {
+		return ""
+	}
+	return net.JoinHostPort("127.0.0.1", strconv.Itoa(c.volumePorts[index]))
+}
+
+func (c *MixedVolumeCluster) VolumePublicAddress(index int) string {
+	if index < 0 || index >= len(c.volumePubPorts) {
+		return ""
+	}
+	return net.JoinHostPort("127.0.0.1", strconv.Itoa(c.volumePubPorts[index]))
+}
+
+func (c *MixedVolumeCluster) VolumeGRPCAddress(index int) string {
+	if index < 0 || index >= len(c.volumeGrpcPorts) {
+		return ""
+	}
+	return net.JoinHostPort("127.0.0.1", strconv.Itoa(c.volumeGrpcPorts[index]))
+}
+
+func (c *MixedVolumeCluster) VolumeAdminURL(index int) string {
+	return "http://" + c.VolumeAdminAddress(index)
+}
+
+func (c *MixedVolumeCluster) VolumePublicURL(index int) string {
+	return "http://" + c.VolumePublicAddress(index)
+}
+
+func (c *MixedVolumeCluster) BaseDir() string {
+	return c.baseDir
+}
+
+// VolumeServerAddress returns SeaweedFS server address format: ip:httpPort.grpcPort
+func (c *MixedVolumeCluster) VolumeServerAddress(index int) string {
+	if index < 0 || index >= len(c.volumePorts) {
+		return ""
+	}
+	return fmt.Sprintf("%s.%d", c.VolumeAdminAddress(index), c.volumeGrpcPorts[index])
+}
+
+func (c *MixedVolumeCluster) IsRust(index int) bool {
+	if index < 0 || index >= len(c.isRust) {
+		return false
+	}
+	return c.isRust[index]
+}
@@ -0,0 +1,294 @@
+package volume_server_grpc_test
+
+import (
+	"context"
+	"net/http"
+	"strings"
+	"testing"
+	"time"
+
+	"github.com/seaweedfs/seaweedfs/test/volume_server/framework"
+	"github.com/seaweedfs/seaweedfs/test/volume_server/matrix"
+	"github.com/seaweedfs/seaweedfs/weed/pb/volume_server_pb"
+	"github.com/seaweedfs/seaweedfs/weed/storage/needle"
+)
+
+// TestEcDecodePreservesDeletedNeedles verifies that needles deleted via
+// VolumeEcBlobDelete (recorded in .ecj) are correctly excluded from the
+// decoded volume produced by VolumeEcShardsToVolume.
+func TestEcDecodePreservesDeletedNeedles(t *testing.T) {
+	if testing.Short() {
+		t.Skip("skipping integration test in short mode")
+	}
+
+	cluster := framework.StartVolumeCluster(t, matrix.P1())
+	conn, client := framework.DialVolumeServer(t, cluster.VolumeGRPCAddress())
+	defer conn.Close()
+
+	const (
+		volumeID = uint32(140)
+		keyA     = uint64(990020)
+		cookieA  = uint32(0xDA001122)
+		keyB     = uint64(990021)
+		cookieB  = uint32(0xDA003344)
+	)
+
+	framework.AllocateVolume(t, client, volumeID, "")
+
+	httpClient := framework.NewHTTPClient()
+	fidA := framework.NewFileID(volumeID, keyA, cookieA)
+	fidB := framework.NewFileID(volumeID, keyB, cookieB)
+	payloadA := []byte("needle-A-should-be-deleted-after-decode")
+	payloadB := []byte("needle-B-should-survive-decode")
+
+	// Upload two needles.
+	for _, tc := range []struct {
+		fid     string
+		payload []byte
+	}{
+		{fidA, payloadA},
+		{fidB, payloadB},
+	} {
+		resp := framework.UploadBytes(t, httpClient, cluster.VolumeAdminURL(), tc.fid, tc.payload)
+		_ = framework.ReadAllAndClose(t, resp)
+		if resp.StatusCode != http.StatusCreated {
+			t.Fatalf("upload %s: expected 201, got %d", tc.fid, resp.StatusCode)
+		}
+	}
+
+	ctx, cancel := context.WithTimeout(context.Background(), 60*time.Second)
+	defer cancel()
+
+	// EC encode.
+	_, err := client.VolumeEcShardsGenerate(ctx, &volume_server_pb.VolumeEcShardsGenerateRequest{
+		VolumeId: volumeID, Collection: "",
+	})
+	if err != nil {
+		t.Fatalf("VolumeEcShardsGenerate: %v", err)
+	}
+
+	// Mount all data shards so the EC volume is usable.
+	_, err = client.VolumeEcShardsMount(ctx, &volume_server_pb.VolumeEcShardsMountRequest{
+		VolumeId: volumeID, Collection: "",
+		ShardIds: []uint32{0, 1, 2, 3, 4, 5, 6, 7, 8, 9},
+	})
+	if err != nil {
+		t.Fatalf("VolumeEcShardsMount: %v", err)
+	}
+
+	// Delete needle A via EC path (writes to .ecj).
+	_, err = client.VolumeEcBlobDelete(ctx, &volume_server_pb.VolumeEcBlobDeleteRequest{
+		VolumeId: volumeID, Collection: "",
+		FileKey: keyA, Version: uint32(needle.GetCurrentVersion()),
+	})
+	if err != nil {
+		t.Fatalf("VolumeEcBlobDelete needle A: %v", err)
+	}
+
+	// Unmount the normal volume so decode writes fresh files.
+	_, err = client.VolumeUnmount(ctx, &volume_server_pb.VolumeUnmountRequest{VolumeId: volumeID})
+	if err != nil {
+		t.Fatalf("VolumeUnmount: %v", err)
+	}
+
+	// Decode EC shards back to a normal volume.
+	_, err = client.VolumeEcShardsToVolume(ctx, &volume_server_pb.VolumeEcShardsToVolumeRequest{
+		VolumeId: volumeID, Collection: "",
+	})
+	if err != nil {
+		t.Fatalf("VolumeEcShardsToVolume: %v", err)
+	}
+
+	// Re-mount the decoded volume.
+	_, err = client.VolumeMount(ctx, &volume_server_pb.VolumeMountRequest{VolumeId: volumeID})
+	if err != nil {
+		t.Fatalf("VolumeMount: %v", err)
+	}
+
+	// Needle A should be gone (deleted via .ecj before decode).
+	respA := framework.ReadBytes(t, httpClient, cluster.VolumeAdminURL(), fidA)
+	bodyA := framework.ReadAllAndClose(t, respA)
+	if respA.StatusCode >= 500 {
+		t.Fatalf("needle A read: server error %d: %s", respA.StatusCode, bodyA)
+	}
+	if respA.StatusCode != http.StatusNotFound {
+		t.Fatalf("needle A should be 404 after decode, got %d", respA.StatusCode)
+	}
+
+	// Needle B should still be readable.
+	respB := framework.ReadBytes(t, httpClient, cluster.VolumeAdminURL(), fidB)
+	bodyB := framework.ReadAllAndClose(t, respB)
+	if respB.StatusCode != http.StatusOK {
+		t.Fatalf("needle B read: expected 200, got %d", respB.StatusCode)
+	}
+	if string(bodyB) != string(payloadB) {
+		t.Fatalf("needle B payload mismatch: got %q, want %q", bodyB, payloadB)
+	}
+}
+
+// TestEcDecodeCollectsEcjFromPeer verifies that .ecj deletion entries from a
+// peer server that contributes no new data shards are still collected during
+// decode. This is the regression test for the fix in collectEcShards that
+// always copies .ecj from every shard location.
+//
+// Scenario:
+//   - Server 0 holds all 10 data shards (decode target).
+//   - Server 1 holds a copy of shard 0 (no new shards for server 0).
+//   - A needle is deleted ONLY on server 1 (server 0's .ecj is empty).
+//   - During decode on server 0, server 1's .ecj must be collected and applied.
+func TestEcDecodeCollectsEcjFromPeer(t *testing.T) {
+	if testing.Short() {
+		t.Skip("skipping integration test in short mode")
+	}
+
+	cluster := framework.StartMultiVolumeClusterAuto(t, matrix.P1(), 2)
+	conn0, client0 := framework.DialVolumeServer(t, cluster.VolumeGRPCAddress(0))
+	defer conn0.Close()
+	conn1, client1 := framework.DialVolumeServer(t, cluster.VolumeGRPCAddress(1))
+	defer conn1.Close()
+
+	const (
+		volumeID = uint32(141)
+		keyA     = uint64(990030)
+		cookieA  = uint32(0xDB001122)
+		keyB     = uint64(990031)
+		cookieB  = uint32(0xDB003344)
+	)
+
+	// Allocate and upload on server 0.
+	framework.AllocateVolume(t, client0, volumeID, "")
+
+	httpClient := framework.NewHTTPClient()
+	fidA := framework.NewFileID(volumeID, keyA, cookieA)
+	fidB := framework.NewFileID(volumeID, keyB, cookieB)
+	payloadB := []byte("needle-B-should-survive-peer-ecj-decode")
+
+	resp := framework.UploadBytes(t, httpClient, cluster.VolumeAdminURL(0), fidA, []byte("needle-A-deleted-on-peer"))
+	_ = framework.ReadAllAndClose(t, resp)
+	if resp.StatusCode != http.StatusCreated {
+		t.Fatalf("upload A: expected 201, got %d", resp.StatusCode)
+	}
+	resp = framework.UploadBytes(t, httpClient, cluster.VolumeAdminURL(0), fidB, payloadB)
+	_ = framework.ReadAllAndClose(t, resp)
+	if resp.StatusCode != http.StatusCreated {
+		t.Fatalf("upload B: expected 201, got %d", resp.StatusCode)
+	}
+
+	ctx, cancel := context.WithTimeout(context.Background(), 60*time.Second)
+	defer cancel()
+
+	// EC encode on server 0.
+	_, err := client0.VolumeEcShardsGenerate(ctx, &volume_server_pb.VolumeEcShardsGenerateRequest{
+		VolumeId: volumeID, Collection: "",
+	})
+	if err != nil {
+		t.Fatalf("VolumeEcShardsGenerate on server 0: %v", err)
+	}
+
+	// Build the SourceDataNode address for server 0 (format: host:adminPort.grpcPort).
+	sourceDataNode := cluster.VolumeAdminAddress(0) + "." +
+		strings.Split(cluster.VolumeGRPCAddress(0), ":")[1]
+
+	// Copy shard 0 + ecx + ecj from server 0 → server 1.
+	_, err = client1.VolumeEcShardsCopy(ctx, &volume_server_pb.VolumeEcShardsCopyRequest{
+		VolumeId:       volumeID,
+		Collection:     "",
+		SourceDataNode: sourceDataNode,
+		ShardIds:       []uint32{0},
+		CopyEcxFile:    true,
+		CopyEcjFile:    true,
+		CopyVifFile:    true,
+	})
+	if err != nil {
+		t.Fatalf("VolumeEcShardsCopy 0→1: %v", err)
+	}
+
+	// Mount shard 0 on server 1 so the EC volume can accept deletions.
+	_, err = client1.VolumeEcShardsMount(ctx, &volume_server_pb.VolumeEcShardsMountRequest{
+		VolumeId: volumeID, Collection: "",
+		ShardIds: []uint32{0},
+	})
+	if err != nil {
+		t.Fatalf("VolumeEcShardsMount on server 1: %v", err)
+	}
+
+	// Delete needle A on server 1 only (creates .ecj entry on server 1).
+	_, err = client1.VolumeEcBlobDelete(ctx, &volume_server_pb.VolumeEcBlobDeleteRequest{
+		VolumeId: volumeID, Collection: "",
+		FileKey: keyA, Version: uint32(needle.GetCurrentVersion()),
+	})
+	if err != nil {
+		t.Fatalf("VolumeEcBlobDelete needle A on server 1: %v", err)
+	}
+
+	// Mount all data shards on server 0 (the decode target).
+	_, err = client0.VolumeEcShardsMount(ctx, &volume_server_pb.VolumeEcShardsMountRequest{
+		VolumeId: volumeID, Collection: "",
+		ShardIds: []uint32{0, 1, 2, 3, 4, 5, 6, 7, 8, 9},
+	})
+	if err != nil {
+		t.Fatalf("VolumeEcShardsMount on server 0: %v", err)
+	}
+
+	// Collect .ecj from server 1 → server 0 with NO new shard IDs.
+	// This is the critical path: server 1 has shard 0 which server 0 already
+	// has, so needToCopyShardsInfo would be empty. Before the fix in
+	// collectEcShards, this copy would be skipped entirely, losing server 1's
+	// deletion entries.
+	server1DataNode := cluster.VolumeAdminAddress(1) + "." +
+		strings.Split(cluster.VolumeGRPCAddress(1), ":")[1]
+
+	_, err = client0.VolumeEcShardsCopy(ctx, &volume_server_pb.VolumeEcShardsCopyRequest{
+		VolumeId:       volumeID,
+		Collection:     "",
+		SourceDataNode: server1DataNode,
+		ShardIds:       []uint32{}, // No new shards — just .ecj.
+		CopyEcxFile:    false,
+		CopyEcjFile:    true,
+		CopyVifFile:    false,
+	})
+	if err != nil {
+		t.Fatalf("VolumeEcShardsCopy .ecj from server 1→0: %v", err)
+	}
+
+	// Unmount the normal volume before decode.
+	_, err = client0.VolumeUnmount(ctx, &volume_server_pb.VolumeUnmountRequest{VolumeId: volumeID})
+	if err != nil {
+		t.Fatalf("VolumeUnmount on server 0: %v", err)
+	}
+
+	// Decode on server 0. RebuildEcxFile should see needle A's deletion from
+	// the .ecj that was collected from server 1.
+	_, err = client0.VolumeEcShardsToVolume(ctx, &volume_server_pb.VolumeEcShardsToVolumeRequest{
+		VolumeId: volumeID, Collection: "",
+	})
+	if err != nil {
+		t.Fatalf("VolumeEcShardsToVolume on server 0: %v", err)
+	}
+
+	// Re-mount the decoded normal volume.
+	_, err = client0.VolumeMount(ctx, &volume_server_pb.VolumeMountRequest{VolumeId: volumeID})
+	if err != nil {
+		t.Fatalf("VolumeMount on server 0: %v", err)
+	}
+
+	// Needle A should be gone — its deletion was in server 1's .ecj.
+	respA := framework.ReadBytes(t, httpClient, cluster.VolumeAdminURL(0), fidA)
+	bodyA := framework.ReadAllAndClose(t, respA)
+	if respA.StatusCode >= 500 {
+		t.Fatalf("needle A read: server error %d: %s", respA.StatusCode, bodyA)
+	}
+	if respA.StatusCode != http.StatusNotFound {
+		t.Fatalf("needle A should be 404 (ecj from peer), got %d", respA.StatusCode)
+	}
+
+	// Needle B should still be readable.
+	respB := framework.ReadBytes(t, httpClient, cluster.VolumeAdminURL(0), fidB)
+	bodyB := framework.ReadAllAndClose(t, respB)
+	if respB.StatusCode != http.StatusOK {
+		t.Fatalf("needle B read: expected 200, got %d", respB.StatusCode)
+	}
+	if string(bodyB) != string(payloadB) {
+		t.Fatalf("needle B payload mismatch: got %q, want %q", bodyB, payloadB)
+	}
+}
@@ -0,0 +1,567 @@
+package volume_server_grpc_test
+
+import (
+	"bytes"
+	"context"
+	"fmt"
+	"io"
+	"net/http"
+	"testing"
+	"time"
+
+	"github.com/seaweedfs/seaweedfs/test/volume_server/framework"
+	"github.com/seaweedfs/seaweedfs/test/volume_server/matrix"
+	"github.com/seaweedfs/seaweedfs/weed/pb/volume_server_pb"
+)
+
+// deleteAndWaitForTombstone issues an HTTP DELETE for the given fid on
+// volumeURL, asserts a successful response, then polls with GET until
+// the file returns 404 (tombstone visible) or the timeout elapses.
+func deleteAndWaitForTombstone(t *testing.T, httpClient *http.Client, volumeURL, fid string) {
+	t.Helper()
+	req, err := http.NewRequest(http.MethodDelete, fmt.Sprintf("%s/%s", volumeURL, fid), nil)
+	if err != nil {
+		t.Fatalf("build delete request for %s: %v", fid, err)
+	}
+	resp := framework.DoRequest(t, httpClient, req)
+	framework.ReadAllAndClose(t, resp)
+	if resp.StatusCode != http.StatusAccepted && resp.StatusCode != http.StatusOK {
+		t.Fatalf("delete %s: expected 200 or 202, got %d", fid, resp.StatusCode)
+	}
+	// Poll until GET returns 404 (tombstone flushed to disk).
+	deadline := time.Now().Add(5 * time.Second)
+	for time.Now().Before(deadline) {
+		getResp := framework.ReadBytes(t, httpClient, volumeURL, fid)
+		status := getResp.StatusCode
+		framework.ReadAllAndClose(t, getResp)
+		if status == http.StatusNotFound {
+			return
+		}
+		time.Sleep(50 * time.Millisecond)
+	}
+	t.Fatalf("delete %s: tombstone not visible after 5s", fid)
+}
+
+// TestMixedBalanceCopyGoToRust verifies that VolumeCopy works from a Go volume
+// server to a Rust volume server. This is the core operation behind volume
+// balancing in a mixed Go+Rust cluster.
+func TestMixedBalanceCopyGoToRust(t *testing.T) {
+	if testing.Short() {
+		t.Skip("skipping integration test in short mode")
+	}
+
+	cluster := framework.StartMixedVolumeCluster(t, matrix.P1(), 1, 1)
+
+	// server 0 = Go, server 1 = Rust
+	conn0, goClient := framework.DialVolumeServer(t, cluster.VolumeGRPCAddress(0))
+	defer conn0.Close()
+	conn1, rustClient := framework.DialVolumeServer(t, cluster.VolumeGRPCAddress(1))
+	defer conn1.Close()
+
+	httpClient := framework.NewHTTPClient()
+
+	const volumeID = uint32(50)
+
+	// Allocate volume on Go server and upload test data
+	framework.AllocateVolume(t, goClient, volumeID, "")
+
+	testFiles := []struct {
+		key    uint64
+		cookie uint32
+		data   []byte
+	}{
+		{1, 0xAABBCCDD, []byte("hello from Go server")},
+		{2, 0x11223344, []byte("second file for balance test")},
+		{3, 0xDEADBEEF, make([]byte, 4096)}, // larger file
+	}
+
+	for _, f := range testFiles {
+		fid := framework.NewFileID(volumeID, f.key, f.cookie)
+		resp := framework.UploadBytes(t, httpClient, cluster.VolumeAdminURL(0), fid, f.data)
+		body := framework.ReadAllAndClose(t, resp)
+		if resp.StatusCode != http.StatusCreated {
+			t.Fatalf("upload %s: expected 201, got %d: %s", fid, resp.StatusCode, body)
+		}
+	}
+
+	ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
+	defer cancel()
+
+	// Read source volume status before copy
+	sourceStatus, err := goClient.ReadVolumeFileStatus(ctx, &volume_server_pb.ReadVolumeFileStatusRequest{
+		VolumeId: volumeID,
+	})
+	if err != nil {
+		t.Fatalf("ReadVolumeFileStatus on Go server: %v", err)
+	}
+	t.Logf("Source: dat=%d idx=%d files=%d version=%d",
+		sourceStatus.GetDatFileSize(), sourceStatus.GetIdxFileSize(),
+		sourceStatus.GetFileCount(), sourceStatus.GetVersion())
+
+	// Copy volume from Go (server 0) to Rust (server 1)
+	copyStream, err := rustClient.VolumeCopy(ctx, &volume_server_pb.VolumeCopyRequest{
+		VolumeId:       volumeID,
+		SourceDataNode: cluster.VolumeServerAddress(0),
+	})
+	if err != nil {
+		t.Fatalf("VolumeCopy call failed: %v", err)
+	}
+
+	var lastAppendAtNs uint64
+	for {
+		resp, recvErr := copyStream.Recv()
+		if recvErr != nil {
+			if recvErr != io.EOF {
+				t.Fatalf("VolumeCopy recv error: %v", recvErr)
+			}
+			break
+		}
+		if resp.GetLastAppendAtNs() != 0 {
+			lastAppendAtNs = resp.GetLastAppendAtNs()
+		}
+	}
+	t.Logf("VolumeCopy completed, lastAppendAtNs=%d", lastAppendAtNs)
+
+	// Verify: read volume status from Rust server
+	targetStatus, err := rustClient.ReadVolumeFileStatus(ctx, &volume_server_pb.ReadVolumeFileStatusRequest{
+		VolumeId: volumeID,
+	})
+	if err != nil {
+		t.Fatalf("ReadVolumeFileStatus on Rust server: %v", err)
+	}
+	t.Logf("Target: dat=%d idx=%d files=%d version=%d",
+		targetStatus.GetDatFileSize(), targetStatus.GetIdxFileSize(),
+		targetStatus.GetFileCount(), targetStatus.GetVersion())
+
+	if sourceStatus.GetDatFileSize() != targetStatus.GetDatFileSize() {
+		t.Fatalf("dat file size mismatch: source=%d target=%d",
+			sourceStatus.GetDatFileSize(), targetStatus.GetDatFileSize())
+	}
+	if sourceStatus.GetIdxFileSize() != targetStatus.GetIdxFileSize() {
+		t.Fatalf("idx file size mismatch: source=%d target=%d",
+			sourceStatus.GetIdxFileSize(), targetStatus.GetIdxFileSize())
+	}
+	if sourceStatus.GetFileCount() != targetStatus.GetFileCount() {
+		t.Fatalf("file count mismatch: source=%d target=%d",
+			sourceStatus.GetFileCount(), targetStatus.GetFileCount())
+	}
+
+	// Verify data can be read from Rust server
+	for _, f := range testFiles {
+		fid := framework.NewFileID(volumeID, f.key, f.cookie)
+		resp := framework.ReadBytes(t, httpClient, cluster.VolumeAdminURL(1), fid)
+		body := framework.ReadAllAndClose(t, resp)
+		if resp.StatusCode != http.StatusOK {
+			t.Fatalf("read %s from Rust server: expected 200, got %d", fid, resp.StatusCode)
+		}
+		if !bytes.Equal(body, f.data) {
+			t.Fatalf("read %s from Rust server: content mismatch (got %d bytes, want %d)", fid, len(body), len(f.data))
+		}
+	}
+}
+
+// TestMixedBalanceCopyRustToGo verifies that VolumeCopy works from a Rust
+// volume server to a Go volume server.
+func TestMixedBalanceCopyRustToGo(t *testing.T) {
+	if testing.Short() {
+		t.Skip("skipping integration test in short mode")
+	}
+
+	cluster := framework.StartMixedVolumeCluster(t, matrix.P1(), 1, 1)
+
+	conn0, goClient := framework.DialVolumeServer(t, cluster.VolumeGRPCAddress(0))
+	defer conn0.Close()
+	conn1, rustClient := framework.DialVolumeServer(t, cluster.VolumeGRPCAddress(1))
+	defer conn1.Close()
+
+	httpClient := framework.NewHTTPClient()
+
+	const volumeID = uint32(51)
+
+	// Allocate volume on Rust server and upload test data
+	framework.AllocateVolume(t, rustClient, volumeID, "")
+
+	testFiles := []struct {
+		key    uint64
+		cookie uint32
+		data   []byte
+	}{
+		{1, 0xAABBCCDD, []byte("hello from Rust server")},
+		{2, 0x11223344, []byte("second file for reverse balance")},
+	}
+
+	for _, f := range testFiles {
+		fid := framework.NewFileID(volumeID, f.key, f.cookie)
+		resp := framework.UploadBytes(t, httpClient, cluster.VolumeAdminURL(1), fid, f.data)
+		body := framework.ReadAllAndClose(t, resp)
+		if resp.StatusCode != http.StatusCreated {
+			t.Fatalf("upload %s: expected 201, got %d: %s", fid, resp.StatusCode, body)
+		}
+	}
+
+	ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
+	defer cancel()
+
+	sourceStatus, err := rustClient.ReadVolumeFileStatus(ctx, &volume_server_pb.ReadVolumeFileStatusRequest{
+		VolumeId: volumeID,
+	})
+	if err != nil {
+		t.Fatalf("ReadVolumeFileStatus on Rust server: %v", err)
+	}
+
+	// Copy volume from Rust (server 1) to Go (server 0)
+	copyStream, err := goClient.VolumeCopy(ctx, &volume_server_pb.VolumeCopyRequest{
+		VolumeId:       volumeID,
+		SourceDataNode: cluster.VolumeServerAddress(1),
+	})
+	if err != nil {
+		t.Fatalf("VolumeCopy call failed: %v", err)
+	}
+	for {
+		_, recvErr := copyStream.Recv()
+		if recvErr != nil {
+			if recvErr != io.EOF {
+				t.Fatalf("VolumeCopy recv error: %v", recvErr)
+			}
+			break
+		}
+	}
+
+	targetStatus, err := goClient.ReadVolumeFileStatus(ctx, &volume_server_pb.ReadVolumeFileStatusRequest{
+		VolumeId: volumeID,
+	})
+	if err != nil {
+		t.Fatalf("ReadVolumeFileStatus on Go server: %v", err)
+	}
+
+	if sourceStatus.GetDatFileSize() != targetStatus.GetDatFileSize() {
+		t.Fatalf("dat file size mismatch: source=%d target=%d",
+			sourceStatus.GetDatFileSize(), targetStatus.GetDatFileSize())
+	}
+	if sourceStatus.GetIdxFileSize() != targetStatus.GetIdxFileSize() {
+		t.Fatalf("idx file size mismatch: source=%d target=%d",
+			sourceStatus.GetIdxFileSize(), targetStatus.GetIdxFileSize())
+	}
+	if sourceStatus.GetFileCount() != targetStatus.GetFileCount() {
+		t.Fatalf("file count mismatch: source=%d target=%d",
+			sourceStatus.GetFileCount(), targetStatus.GetFileCount())
+	}
+
+	// Verify data can be read from Go server
+	for _, f := range testFiles {
+		fid := framework.NewFileID(volumeID, f.key, f.cookie)
+		resp := framework.ReadBytes(t, httpClient, cluster.VolumeAdminURL(0), fid)
+		body := framework.ReadAllAndClose(t, resp)
+		if resp.StatusCode != http.StatusOK {
+			t.Fatalf("read %s from Go server: expected 200, got %d", fid, resp.StatusCode)
+		}
+		if !bytes.Equal(body, f.data) {
+			t.Fatalf("read %s from Go server: content mismatch (got %d bytes, want %d)", fid, len(body), len(f.data))
+		}
+	}
+}
+
+// TestMixedBalanceCopyWithDeletes verifies that VolumeCopy correctly handles
+// volumes that have both active and deleted needles.
+func TestMixedBalanceCopyWithDeletes(t *testing.T) {
+	if testing.Short() {
+		t.Skip("skipping integration test in short mode")
+	}
+
+	cluster := framework.StartMixedVolumeCluster(t, matrix.P1(), 1, 1)
+
+	conn0, goClient := framework.DialVolumeServer(t, cluster.VolumeGRPCAddress(0))
+	defer conn0.Close()
+	conn1, rustClient := framework.DialVolumeServer(t, cluster.VolumeGRPCAddress(1))
+	defer conn1.Close()
+
+	httpClient := framework.NewHTTPClient()
+
+	const volumeID = uint32(52)
+	framework.AllocateVolume(t, goClient, volumeID, "")
+
+	// Upload files
+	for i := uint64(1); i <= 5; i++ {
+		fid := framework.NewFileID(volumeID, i, 0x12340000+uint32(i))
+		resp := framework.UploadBytes(t, httpClient, cluster.VolumeAdminURL(0), fid, []byte(fmt.Sprintf("file-%d", i)))
+		framework.ReadAllAndClose(t, resp)
+		if resp.StatusCode != http.StatusCreated {
+			t.Fatalf("upload file %d: expected 201, got %d", i, resp.StatusCode)
+		}
+	}
+
+	// Delete some files and wait for tombstones to be visible
+	for _, key := range []uint64{2, 4} {
+		fid := framework.NewFileID(volumeID, key, 0x12340000+uint32(key))
+		deleteAndWaitForTombstone(t, httpClient, cluster.VolumeAdminURL(0), fid)
+	}
+
+	ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
+	defer cancel()
+
+	sourceStatus, err := goClient.ReadVolumeFileStatus(ctx, &volume_server_pb.ReadVolumeFileStatusRequest{
+		VolumeId: volumeID,
+	})
+	if err != nil {
+		t.Fatalf("ReadVolumeFileStatus: %v", err)
+	}
+	t.Logf("Source after deletes: dat=%d idx=%d files=%d",
+		sourceStatus.GetDatFileSize(), sourceStatus.GetIdxFileSize(),
+		sourceStatus.GetFileCount())
+
+	// Copy volume from Go to Rust
+	copyStream, err := rustClient.VolumeCopy(ctx, &volume_server_pb.VolumeCopyRequest{
+		VolumeId:       volumeID,
+		SourceDataNode: cluster.VolumeServerAddress(0),
+	})
+	if err != nil {
+		t.Fatalf("VolumeCopy: %v", err)
+	}
+	var lastAppendAtNs uint64
+	for {
+		resp, recvErr := copyStream.Recv()
+		if recvErr != nil {
+			if recvErr != io.EOF {
+				t.Fatalf("VolumeCopy recv error: %v", recvErr)
+			}
+			break
+		}
+		if resp.GetLastAppendAtNs() != 0 {
+			lastAppendAtNs = resp.GetLastAppendAtNs()
+		}
+	}
+	if lastAppendAtNs == 0 {
+		t.Fatalf("VolumeCopy did not return a lastAppendAtNs timestamp")
+	}
+	t.Logf("VolumeCopy completed, lastAppendAtNs=%d", lastAppendAtNs)
+
+	targetStatusAfterCopy, err := rustClient.ReadVolumeFileStatus(ctx, &volume_server_pb.ReadVolumeFileStatusRequest{
+		VolumeId: volumeID,
+	})
+	if err != nil {
+		t.Fatalf("ReadVolumeFileStatus on Rust: %v", err)
+	}
+	t.Logf("Target after copy: dat=%d idx=%d files=%d",
+		targetStatusAfterCopy.GetDatFileSize(), targetStatusAfterCopy.GetIdxFileSize(),
+		targetStatusAfterCopy.GetFileCount())
+
+	if sourceStatus.GetDatFileSize() != targetStatusAfterCopy.GetDatFileSize() {
+		t.Fatalf("dat file size mismatch: source=%d target=%d",
+			sourceStatus.GetDatFileSize(), targetStatusAfterCopy.GetDatFileSize())
+	}
+	if sourceStatus.GetIdxFileSize() != targetStatusAfterCopy.GetIdxFileSize() {
+		t.Fatalf("idx file size mismatch: source=%d target=%d",
+			sourceStatus.GetIdxFileSize(), targetStatusAfterCopy.GetIdxFileSize())
+	}
+
+	// Tail from the copy checkpoint — source is unchanged (deletes happened
+	// before copy), so tailing should not append any data.
+	_, err = rustClient.VolumeTailReceiver(ctx, &volume_server_pb.VolumeTailReceiverRequest{
+		VolumeId:           volumeID,
+		SinceNs:            lastAppendAtNs,
+		IdleTimeoutSeconds: 3,
+		SourceVolumeServer: cluster.VolumeServerAddress(0),
+	})
+	if err != nil {
+		t.Fatalf("VolumeTailReceiver: %v", err)
+	}
+
+	targetStatusAfterTail, err := rustClient.ReadVolumeFileStatus(ctx, &volume_server_pb.ReadVolumeFileStatusRequest{
+		VolumeId: volumeID,
+	})
+	if err != nil {
+		t.Fatalf("ReadVolumeFileStatus after tail: %v", err)
+	}
+	if targetStatusAfterTail.GetDatFileSize() != targetStatusAfterCopy.GetDatFileSize() {
+		t.Fatalf("dat grew after tail: before=%d after=%d",
+			targetStatusAfterCopy.GetDatFileSize(), targetStatusAfterTail.GetDatFileSize())
+	}
+	if targetStatusAfterTail.GetIdxFileSize() != targetStatusAfterCopy.GetIdxFileSize() {
+		t.Fatalf("idx grew after tail: before=%d after=%d",
+			targetStatusAfterCopy.GetIdxFileSize(), targetStatusAfterTail.GetIdxFileSize())
+	}
+
+	// Verify surviving files are readable from Rust
+	for _, key := range []uint64{1, 3, 5} {
+		fid := framework.NewFileID(volumeID, key, 0x12340000+uint32(key))
+		resp := framework.ReadBytes(t, httpClient, cluster.VolumeAdminURL(1), fid)
+		body := framework.ReadAllAndClose(t, resp)
+		if resp.StatusCode != http.StatusOK {
+			t.Fatalf("read surviving file %d from Rust: expected 200, got %d", key, resp.StatusCode)
+		}
+		expected := fmt.Sprintf("file-%d", key)
+		if string(body) != expected {
+			t.Fatalf("data mismatch for file %d: got %q want %q", key, body, expected)
+		}
+	}
+
+	// Verify deleted files return 404
+	for _, key := range []uint64{2, 4} {
+		fid := framework.NewFileID(volumeID, key, 0x12340000+uint32(key))
+		resp := framework.ReadBytes(t, httpClient, cluster.VolumeAdminURL(1), fid)
+		framework.ReadAllAndClose(t, resp)
+		if resp.StatusCode != http.StatusNotFound {
+			t.Fatalf("read deleted file %d from Rust: expected 404, got %d", key, resp.StatusCode)
+		}
+	}
+}
+
+// TestMixedBalanceFullMoveGoToRust exercises the complete volume balance move
+// flow: mark readonly → copy → tail → verify sizes → delete source.
+// This mirrors the steps in balance_task.go Execute().
+func TestMixedBalanceFullMoveGoToRust(t *testing.T) {
+	if testing.Short() {
+		t.Skip("skipping integration test in short mode")
+	}
+
+	cluster := framework.StartMixedVolumeCluster(t, matrix.P1(), 1, 1)
+
+	conn0, goClient := framework.DialVolumeServer(t, cluster.VolumeGRPCAddress(0))
+	defer conn0.Close()
+	conn1, rustClient := framework.DialVolumeServer(t, cluster.VolumeGRPCAddress(1))
+	defer conn1.Close()
+
+	httpClient := framework.NewHTTPClient()
+
+	const volumeID = uint32(53)
+	framework.AllocateVolume(t, goClient, volumeID, "")
+
+	// Upload test data
+	for i := uint64(1); i <= 5; i++ {
+		fid := framework.NewFileID(volumeID, i, 0xABCD0000+uint32(i))
+		resp := framework.UploadBytes(t, httpClient, cluster.VolumeAdminURL(0), fid, []byte(fmt.Sprintf("balance-move-file-%d", i)))
+		framework.ReadAllAndClose(t, resp)
+		if resp.StatusCode != http.StatusCreated {
+			t.Fatalf("upload file %d: expected 201, got %d", i, resp.StatusCode)
+		}
+	}
+
+	ctx, cancel := context.WithTimeout(context.Background(), 60*time.Second)
+	defer cancel()
+
+	// Step 1: Copy volume to Rust server (source stays writable so we can
+	// delete after copy to exercise the tail tombstone path)
+	copyStream, err := rustClient.VolumeCopy(ctx, &volume_server_pb.VolumeCopyRequest{
+		VolumeId:       volumeID,
+		SourceDataNode: cluster.VolumeServerAddress(0),
+	})
+	if err != nil {
+		t.Fatalf("VolumeCopy: %v", err)
+	}
+	var lastAppendAtNs uint64
+	for {
+		resp, recvErr := copyStream.Recv()
+		if recvErr != nil {
+			if recvErr != io.EOF {
+				t.Fatalf("VolumeCopy recv: %v", recvErr)
+			}
+			break
+		}
+		if resp.GetLastAppendAtNs() != 0 {
+			lastAppendAtNs = resp.GetLastAppendAtNs()
+		}
+	}
+	t.Logf("Copy done, lastAppendAtNs=%d", lastAppendAtNs)
+
+	// Step 2: Delete file 3 on the source AFTER copy. This creates a
+	// tombstone needle that the tail step must propagate to the Rust server.
+	deleteAndWaitForTombstone(t, httpClient, cluster.VolumeAdminURL(0),
+		framework.NewFileID(volumeID, 3, 0xABCD0003))
+
+	// Step 3: Mark source readonly so no further writes arrive during tail
+	_, err = goClient.VolumeMarkReadonly(ctx, &volume_server_pb.VolumeMarkReadonlyRequest{
+		VolumeId: volumeID,
+	})
+	if err != nil {
+		t.Fatalf("mark readonly: %v", err)
+	}
+
+	// Read source status (the reference for post-tail verification)
+	sourceStatus, err := goClient.ReadVolumeFileStatus(ctx, &volume_server_pb.ReadVolumeFileStatusRequest{
+		VolumeId: volumeID,
+	})
+	if err != nil {
+		t.Fatalf("read source status: %v", err)
+	}
+	t.Logf("Source: dat=%d idx=%d files=%d",
+		sourceStatus.GetDatFileSize(), sourceStatus.GetIdxFileSize(),
+		sourceStatus.GetFileCount())
+
+	// Step 4: Tail for updates — this must pick up the delete tombstone
+	_, err = rustClient.VolumeTailReceiver(ctx, &volume_server_pb.VolumeTailReceiverRequest{
+		VolumeId:           volumeID,
+		SinceNs:            lastAppendAtNs,
+		IdleTimeoutSeconds: 5,
+		SourceVolumeServer: cluster.VolumeServerAddress(0),
+	})
+	if err != nil {
+		t.Fatalf("VolumeTailReceiver: %v", err)
+	}
+
+	// Step 5: Verify dat/idx sizes match after tail.
+	// We compare file sizes (byte-level correctness) rather than file_count
+	// because the tail writes the tombstone via the write path (which
+	// increments file_count) while the source used the delete path (which
+	// only increments deletion_count). This is consistent with Go behavior.
+	targetStatus, err := rustClient.ReadVolumeFileStatus(ctx, &volume_server_pb.ReadVolumeFileStatusRequest{
+		VolumeId: volumeID,
+	})
+	if err != nil {
+		t.Fatalf("read target status: %v", err)
+	}
+	t.Logf("Target: dat=%d idx=%d files=%d",
+		targetStatus.GetDatFileSize(), targetStatus.GetIdxFileSize(),
+		targetStatus.GetFileCount())
+
+	if sourceStatus.GetDatFileSize() != targetStatus.GetDatFileSize() {
+		t.Fatalf("dat size mismatch after tail: source=%d target=%d",
+			sourceStatus.GetDatFileSize(), targetStatus.GetDatFileSize())
+	}
+	if sourceStatus.GetIdxFileSize() != targetStatus.GetIdxFileSize() {
+		t.Fatalf("idx size mismatch after tail: source=%d target=%d",
+			sourceStatus.GetIdxFileSize(), targetStatus.GetIdxFileSize())
+	}
+
+	// Step 6: Delete volume from source
+	_, err = goClient.VolumeDelete(ctx, &volume_server_pb.VolumeDeleteRequest{
+		VolumeId: volumeID,
+	})
+	if err != nil {
+		t.Fatalf("delete source volume: %v", err)
+	}
+
+	// Verify source volume is gone
+	_, err = goClient.ReadVolumeFileStatus(ctx, &volume_server_pb.ReadVolumeFileStatusRequest{
+		VolumeId: volumeID,
+	})
+	if err == nil {
+		t.Fatalf("expected error reading deleted source volume, got nil")
+	}
+	t.Logf("Source volume deleted successfully (error as expected: %v)", err)
+
+	// Verify all surviving data is readable from Rust (the new home)
+	for _, key := range []uint64{1, 2, 4, 5} {
+		fid := framework.NewFileID(volumeID, key, 0xABCD0000+uint32(key))
+		resp := framework.ReadBytes(t, httpClient, cluster.VolumeAdminURL(1), fid)
+		body := framework.ReadAllAndClose(t, resp)
+		if resp.StatusCode != http.StatusOK {
+			t.Fatalf("read file %d from Rust after move: expected 200, got %d", key, resp.StatusCode)
+		}
+		expected := fmt.Sprintf("balance-move-file-%d", key)
+		if string(body) != expected {
+			t.Fatalf("data mismatch for file %d after move: got %q want %q", key, body, expected)
+		}
+	}
+
+	// Verify deleted file is 404 on Rust (tombstone propagated via tail)
+	{
+		fid := framework.NewFileID(volumeID, 3, 0xABCD0003)
+		resp := framework.ReadBytes(t, httpClient, cluster.VolumeAdminURL(1), fid)
+		framework.ReadAllAndClose(t, resp)
+		if resp.StatusCode != http.StatusNotFound {
+			t.Fatalf("deleted file 3 on Rust: expected 404, got %d", resp.StatusCode)
+		}
+	}
+
+	t.Logf("Full balance move completed: volume %d moved from Go to Rust, source purged", volumeID)
+}
@@ -45,6 +45,7 @@ type ObjectStoreUser struct {
 	SecretKey   string   `json:"secret_key"`
 	Permissions []string `json:"permissions"`
 	PolicyNames []string `json:"policy_names"`
+	IsStatic    bool     `json:"is_static"` // loaded from static config file, not editable
 }

 type ObjectStoreUsersData struct {
@@ -867,6 +867,24 @@ func (s *AdminServer) DeleteS3Bucket(bucketName string) error {
 	})
 }

+// IsStaticUser checks if a user is a static identity by loading the
+// configuration from the credential manager and checking the IsStatic flag.
+func (s *AdminServer) IsStaticUser(username string) bool {
+	if s.credentialManager == nil {
+		return false
+	}
+	s3cfg, err := s.credentialManager.LoadConfiguration(context.Background())
+	if err != nil {
+		return false
+	}
+	for _, ident := range s3cfg.Identities {
+		if ident.Name == username {
+			return ident.IsStatic
+		}
+	}
+	return false
+}
+
 // GetObjectStoreUsers retrieves object store users from identity.json
 func (s *AdminServer) GetObjectStoreUsers(ctx context.Context) ([]ObjectStoreUser, error) {
 	if s.credentialManager == nil {
@@ -890,6 +908,7 @@ func (s *AdminServer) GetObjectStoreUsers(ctx context.Context) ([]ObjectStoreUse
 		user := ObjectStoreUser{
 			Username:    identity.Name,
 			Permissions: identity.Actions,
+			IsStatic:    identity.IsStatic,
 		}

 		// Set email from account if available
@@ -943,7 +962,7 @@ func (s *AdminServer) GetClusterMasters() (*ClusterMastersData, error) {
 			leaderCount++
 		}

-		masterMap[master.Address] = masterInfo
+		masterMap[masterInfo.Address] = masterInfo
 	}

 	// Then, get additional master information from Raft cluster
@@ -955,11 +974,11 @@ func (s *AdminServer) GetClusterMasters() (*ClusterMastersData, error) {

 		// Process each raft server
 		for _, server := range resp.ClusterServers {
-			address := server.Address
-			httpAddress := pb.ServerAddress(address).ToHttpAddress()
+			// Raft stores gRPC addresses, convert to HTTP address
+			httpAddress := pb.GrpcAddressToServerAddress(server.Address)

 			// Update existing master info or create new one
-			if masterInfo, exists := masterMap[address]; exists {
+			if masterInfo, exists := masterMap[httpAddress]; exists {
 				// Update existing master with raft data
 				masterInfo.IsLeader = server.IsLeader
 				masterInfo.Suffrage = server.Suffrage
@@ -970,7 +989,7 @@ func (s *AdminServer) GetClusterMasters() (*ClusterMastersData, error) {
 					IsLeader: server.IsLeader,
 					Suffrage: server.Suffrage,
 				}
-				masterMap[address] = masterInfo
+				masterMap[httpAddress] = masterInfo
 			}

 			if server.IsLeader {
@@ -1646,159 +1665,6 @@ func (as *AdminServer) GetConfigPersistence() *ConfigPersistence {
 	return as.configPersistence
 }

-// convertJSONToMaintenanceConfig converts JSON map to protobuf MaintenanceConfig
-func convertJSONToMaintenanceConfig(jsonConfig map[string]interface{}) (*maintenance.MaintenanceConfig, error) {
-	config := &maintenance.MaintenanceConfig{}
-
-	// Helper function to get int32 from interface{}
-	getInt32 := func(key string) (int32, error) {
-		if val, ok := jsonConfig[key]; ok {
-			switch v := val.(type) {
-			case int:
-				return int32(v), nil
-			case int32:
-				return v, nil
-			case int64:
-				return int32(v), nil
-			case float64:
-				return int32(v), nil
-			default:
-				return 0, fmt.Errorf("invalid type for %s: expected number, got %T", key, v)
-			}
-		}
-		return 0, nil
-	}
-
-	// Helper function to get bool from interface{}
-	getBool := func(key string) bool {
-		if val, ok := jsonConfig[key]; ok {
-			if b, ok := val.(bool); ok {
-				return b
-			}
-		}
-		return false
-	}
-
-	var err error
-
-	// Convert basic fields
-	config.Enabled = getBool("enabled")
-
-	if config.ScanIntervalSeconds, err = getInt32("scan_interval_seconds"); err != nil {
-		return nil, err
-	}
-	if config.WorkerTimeoutSeconds, err = getInt32("worker_timeout_seconds"); err != nil {
-		return nil, err
-	}
-	if config.TaskTimeoutSeconds, err = getInt32("task_timeout_seconds"); err != nil {
-		return nil, err
-	}
-	if config.RetryDelaySeconds, err = getInt32("retry_delay_seconds"); err != nil {
-		return nil, err
-	}
-	if config.MaxRetries, err = getInt32("max_retries"); err != nil {
-		return nil, err
-	}
-	if config.CleanupIntervalSeconds, err = getInt32("cleanup_interval_seconds"); err != nil {
-		return nil, err
-	}
-	if config.TaskRetentionSeconds, err = getInt32("task_retention_seconds"); err != nil {
-		return nil, err
-	}
-
-	// Convert policy if present
-	if policyData, ok := jsonConfig["policy"]; ok {
-		if policyMap, ok := policyData.(map[string]interface{}); ok {
-			policy := &maintenance.MaintenancePolicy{}
-
-			if globalMaxConcurrent, err := getInt32FromMap(policyMap, "global_max_concurrent"); err != nil {
-				return nil, err
-			} else {
-				policy.GlobalMaxConcurrent = globalMaxConcurrent
-			}
-
-			if defaultRepeatIntervalSeconds, err := getInt32FromMap(policyMap, "default_repeat_interval_seconds"); err != nil {
-				return nil, err
-			} else {
-				policy.DefaultRepeatIntervalSeconds = defaultRepeatIntervalSeconds
-			}
-
-			if defaultCheckIntervalSeconds, err := getInt32FromMap(policyMap, "default_check_interval_seconds"); err != nil {
-				return nil, err
-			} else {
-				policy.DefaultCheckIntervalSeconds = defaultCheckIntervalSeconds
-			}
-
-			// Convert task policies if present
-			if taskPoliciesData, ok := policyMap["task_policies"]; ok {
-				if taskPoliciesMap, ok := taskPoliciesData.(map[string]interface{}); ok {
-					policy.TaskPolicies = make(map[string]*maintenance.TaskPolicy)
-
-					for taskType, taskPolicyData := range taskPoliciesMap {
-						if taskPolicyMap, ok := taskPolicyData.(map[string]interface{}); ok {
-							taskPolicy := &maintenance.TaskPolicy{}
-
-							taskPolicy.Enabled = getBoolFromMap(taskPolicyMap, "enabled")
-
-							if maxConcurrent, err := getInt32FromMap(taskPolicyMap, "max_concurrent"); err != nil {
-								return nil, err
-							} else {
-								taskPolicy.MaxConcurrent = maxConcurrent
-							}
-
-							if repeatIntervalSeconds, err := getInt32FromMap(taskPolicyMap, "repeat_interval_seconds"); err != nil {
-								return nil, err
-							} else {
-								taskPolicy.RepeatIntervalSeconds = repeatIntervalSeconds
-							}
-
-							if checkIntervalSeconds, err := getInt32FromMap(taskPolicyMap, "check_interval_seconds"); err != nil {
-								return nil, err
-							} else {
-								taskPolicy.CheckIntervalSeconds = checkIntervalSeconds
-							}
-
-							policy.TaskPolicies[taskType] = taskPolicy
-						}
-					}
-				}
-			}
-
-			config.Policy = policy
-		}
-	}
-
-	return config, nil
-}
-
-// Helper functions for map conversion
-func getInt32FromMap(m map[string]interface{}, key string) (int32, error) {
-	if val, ok := m[key]; ok {
-		switch v := val.(type) {
-		case int:
-			return int32(v), nil
-		case int32:
-			return v, nil
-		case int64:
-			return int32(v), nil
-		case float64:
-			return int32(v), nil
-		default:
-			return 0, fmt.Errorf("invalid type for %s: expected number, got %T", key, v)
-		}
-	}
-	return 0, nil
-}
-
-func getBoolFromMap(m map[string]interface{}, key string) bool {
-	if val, ok := m[key]; ok {
-		if b, ok := val.(bool); ok {
-			return b
-		}
-	}
-	return false
-}
-
 type collectionStats struct {
 	PhysicalSize int64
 	LogicalSize  int64
@@ -361,26 +361,6 @@ func normalizeQuotaUnit(unit string) (string, error) {
 	}
 }

-// Helper function to convert bytes to appropriate unit and size
-func convertBytesToQuota(bytes int64) (int64, string) {
-	if bytes == 0 {
-		return 0, "MB"
-	}
-
-	// Convert to TB if >= 1TB
-	if bytes >= 1024*1024*1024*1024 && bytes%(1024*1024*1024*1024) == 0 {
-		return bytes / (1024 * 1024 * 1024 * 1024), "TB"
-	}
-
-	// Convert to GB if >= 1GB
-	if bytes >= 1024*1024*1024 && bytes%(1024*1024*1024) == 0 {
-		return bytes / (1024 * 1024 * 1024), "GB"
-	}
-
-	// Convert to MB (default)
-	return bytes / (1024 * 1024), "MB"
-}
-
 // SetBucketQuota sets the quota for a bucket
 func (s *AdminServer) SetBucketQuota(bucketName string, quotaBytes int64, quotaEnabled bool) error {
 	return s.WithFilerClient(func(client filer_pb.SeaweedFilerClient) error {
@@ -506,18 +506,6 @@ func getShardCount(ecIndexBits uint32) int {
 	return count
 }

-// getMissingShards returns a slice of missing shard IDs for a volume
-// Assumes default 10+4 EC configuration (14 total shards)
-func getMissingShards(ecIndexBits uint32) []int {
-	var missing []int
-	for i := 0; i < erasure_coding.TotalShardsCount; i++ {
-		if (ecIndexBits & (1 << uint(i))) == 0 {
-			missing = append(missing, i)
-		}
-	}
-	return missing
-}
-
 // sortEcShards sorts EC shards based on the specified field and order
 func sortEcShards(shards []EcShardWithInfo, sortBy string, sortOrder string) {
 	sort.Slice(shards, func(i, j int) bool {
@@ -430,67 +430,6 @@ func (s *AdminServer) GetConsumerGroupOffsets(namespace, topicName string) ([]Co
 	return offsets, nil
 }

-// convertRecordTypeToSchemaFields converts a protobuf RecordType to SchemaFieldInfo slice
-func convertRecordTypeToSchemaFields(recordType *schema_pb.RecordType) []SchemaFieldInfo {
-	var schemaFields []SchemaFieldInfo
-
-	if recordType == nil || recordType.Fields == nil {
-		return schemaFields
-	}
-
-	for _, field := range recordType.Fields {
-		schemaField := SchemaFieldInfo{
-			Name:     field.Name,
-			Type:     getFieldTypeString(field.Type),
-			Required: field.IsRequired,
-		}
-		schemaFields = append(schemaFields, schemaField)
-	}
-
-	return schemaFields
-}
-
-// getFieldTypeString converts a protobuf Type to a human-readable string
-func getFieldTypeString(fieldType *schema_pb.Type) string {
-	if fieldType == nil {
-		return "unknown"
-	}
-
-	switch kind := fieldType.Kind.(type) {
-	case *schema_pb.Type_ScalarType:
-		return getScalarTypeString(kind.ScalarType)
-	case *schema_pb.Type_RecordType:
-		return "record"
-	case *schema_pb.Type_ListType:
-		elementType := getFieldTypeString(kind.ListType.ElementType)
-		return fmt.Sprintf("list<%s>", elementType)
-	default:
-		return "unknown"
-	}
-}
-
-// getScalarTypeString converts a protobuf ScalarType to a string
-func getScalarTypeString(scalarType schema_pb.ScalarType) string {
-	switch scalarType {
-	case schema_pb.ScalarType_BOOL:
-		return "bool"
-	case schema_pb.ScalarType_INT32:
-		return "int32"
-	case schema_pb.ScalarType_INT64:
-		return "int64"
-	case schema_pb.ScalarType_FLOAT:
-		return "float"
-	case schema_pb.ScalarType_DOUBLE:
-		return "double"
-	case schema_pb.ScalarType_BYTES:
-		return "bytes"
-	case schema_pb.ScalarType_STRING:
-		return "string"
-	default:
-		return "unknown"
-	}
-}
-
 // convertTopicPublishers converts protobuf TopicPublisher slice to PublisherInfo slice
 func convertTopicPublishers(publishers []*mq_pb.TopicPublisher) []PublisherInfo {
 	publisherInfos := make([]PublisherInfo, 0, len(publishers))
@@ -2,8 +2,6 @@ package dash

 import (
 	"context"
-	"crypto/rand"
-	"encoding/hex"
 	"encoding/json"
 	"errors"
 	"fmt"
@@ -851,43 +849,6 @@ func normalizeTimeout(timeoutSeconds int, defaultTimeout, maxTimeout time.Durati
 	return timeout
 }

-func buildJobSpecFromProposal(jobType string, proposal *plugin_pb.JobProposal, index int) *plugin_pb.JobSpec {
-	now := timestamppb.Now()
-	suffix := make([]byte, 4)
-	if _, err := rand.Read(suffix); err != nil {
-		// Fallback to simpler ID if rand fails
-		suffix = []byte(fmt.Sprintf("%d", index))
-	}
-	jobID := fmt.Sprintf("%s-%d-%s", jobType, now.AsTime().UnixNano(), hex.EncodeToString(suffix))
-
-	jobSpec := &plugin_pb.JobSpec{
-		JobId:      jobID,
-		JobType:    jobType,
-		Priority:   plugin_pb.JobPriority_JOB_PRIORITY_NORMAL,
-		CreatedAt:  now,
-		Labels:     make(map[string]string),
-		Parameters: make(map[string]*plugin_pb.ConfigValue),
-		DedupeKey:  "",
-	}
-
-	if proposal != nil {
-		jobSpec.Summary = proposal.Summary
-		jobSpec.Detail = proposal.Detail
-		if proposal.Priority != plugin_pb.JobPriority_JOB_PRIORITY_UNSPECIFIED {
-			jobSpec.Priority = proposal.Priority
-		}
-		jobSpec.DedupeKey = proposal.DedupeKey
-		jobSpec.Parameters = plugin.CloneConfigValueMap(proposal.Parameters)
-		if proposal.Labels != nil {
-			for k, v := range proposal.Labels {
-				jobSpec.Labels[k] = v
-			}
-		}
-	}
-
-	return jobSpec
-}
-
 func applyDescriptorDefaultsToPersistedConfig(
 	config *plugin_pb.PersistedJobTypeConfig,
 	descriptor *plugin_pb.JobTypeDescriptor,
@@ -115,32 +115,6 @@ func TestExpirePluginJobAPI(t *testing.T) {
 	})
 }

-func TestBuildJobSpecFromProposalDoesNotReuseProposalID(t *testing.T) {
-	t.Parallel()
-
-	proposal := &plugin_pb.JobProposal{
-		ProposalId: "vacuum-2",
-		DedupeKey:  "vacuum:2",
-		JobType:    "vacuum",
-	}
-
-	jobA := buildJobSpecFromProposal("vacuum", proposal, 0)
-	jobB := buildJobSpecFromProposal("vacuum", proposal, 1)
-
-	if jobA.JobId == proposal.ProposalId {
-		t.Fatalf("job id must not reuse proposal id: %s", jobA.JobId)
-	}
-	if jobB.JobId == proposal.ProposalId {
-		t.Fatalf("job id must not reuse proposal id: %s", jobB.JobId)
-	}
-	if jobA.JobId == jobB.JobId {
-		t.Fatalf("job ids must be unique across jobs: %s", jobA.JobId)
-	}
-	if jobA.DedupeKey != proposal.DedupeKey {
-		t.Fatalf("dedupe key must be preserved: got=%s want=%s", jobA.DedupeKey, proposal.DedupeKey)
-	}
-}
-
 func TestApplyDescriptorDefaultsToPersistedConfigBackfillsAdminDefaults(t *testing.T) {
 	t.Parallel()

@@ -447,11 +447,6 @@ type QueueStats = maintenance.QueueStats
 type WorkerDetailsData = maintenance.WorkerDetailsData
 type WorkerPerformance = maintenance.WorkerPerformance

-// GetTaskIcon returns the icon CSS class for a task type from its UI provider
-func GetTaskIcon(taskType MaintenanceTaskType) string {
-	return maintenance.GetTaskIcon(taskType)
-}
-
 // Status constants (these are still static)
 const (
 	TaskStatusPending    = maintenance.TaskStatusPending
@@ -3,13 +3,13 @@ package dash
 import (
 	"context"
 	"crypto/rand"
-	"encoding/base64"
 	"errors"
 	"fmt"
 	"strings"
 	"time"

 	"github.com/seaweedfs/seaweedfs/weed/credential"
+	"github.com/seaweedfs/seaweedfs/weed/iam"
 	"github.com/seaweedfs/seaweedfs/weed/pb/iam_pb"
 )

@@ -175,7 +175,7 @@ func (s *AdminServer) GetObjectStoreUserDetails(username string) (*UserDetails,

 	ctx := context.Background()

-	// Get user using credential manager
+	// Get user using credential manager (resolves static users via filer gRPC)
 	identity, err := s.credentialManager.GetUser(ctx, username)
 	if err != nil {
 		if err == credential.ErrUserNotFound {
@@ -435,10 +435,13 @@ func generateAccessKey() string {
 }

 func generateSecretKey() string {
-	// Generate 40-character secret key (AWS standard)
-	b := make([]byte, 30) // 30 bytes = 40 characters in base64
-	rand.Read(b)
-	return base64.StdEncoding.EncodeToString(b)
+	// Use the IAM helper to generate URL-safe secret keys (no +, / characters)
+	// that won't break S3 signature authentication
+	key, err := iam.GenerateSecretAccessKey()
+	if err != nil {
+		panic(fmt.Sprintf("failed to generate secret key: %v", err))
+	}
+	return key
 }

 func generateAccountId() string {
@@ -74,9 +74,16 @@ func TestGenerateSecretKey(t *testing.T) {
 	key1 := generateSecretKey()
 	key2 := generateSecretKey()

-	// Check length (base64 encoding of 30 bytes = 40 characters)
-	if len(key1) != 40 {
-		t.Errorf("Expected secret key length 40, got %d", len(key1))
+	// Check length (IAM standard secret key length)
+	if len(key1) != 42 {
+		t.Errorf("Expected secret key length 42, got %d", len(key1))
+	}
+
+	// Check that key contains only URL-safe characters (no +, /)
+	for _, c := range key1 {
+		if c == '+' || c == '/' || c == '=' {
+			t.Errorf("Secret key contains non-URL-safe character: %c", c)
+		}
 	}

 	// Check uniqueness
@@ -787,15 +787,6 @@ func (s *WorkerGrpcServer) RequestTaskLogsFromAllWorkers(taskID string, maxEntri
 	return results, nil
 }

-// convertTaskParameters converts task parameters to protobuf format
-func convertTaskParameters(params map[string]interface{}) map[string]string {
-	result := make(map[string]string)
-	for key, value := range params {
-		result[key] = fmt.Sprintf("%v", value)
-	}
-	return result
-}
-
 func findClientAddress(ctx context.Context) string {
 	// fmt.Printf("FromContext %+v\n", ctx)
 	pr, ok := peer.FromContext(ctx)
@@ -312,29 +312,6 @@ func (h *ClusterHandlers) ShowClusterFilers(w http.ResponseWriter, r *http.Reque
 	}
 }

-// ShowClusterBrokers renders the cluster message brokers page
-func (h *ClusterHandlers) ShowClusterBrokers(w http.ResponseWriter, r *http.Request) {
-	// Get cluster brokers data
-	brokersData, err := h.adminServer.GetClusterBrokers()
-	if err != nil {
-		writeJSONError(w, http.StatusInternalServerError, "Failed to get cluster brokers: "+err.Error())
-		return
-	}
-
-	username := usernameOrDefault(r)
-	brokersData.Username = username
-
-	// Render HTML template
-	w.Header().Set("Content-Type", "text/html")
-	brokersComponent := app.ClusterBrokers(*brokersData)
-	viewCtx := layout.NewViewContext(r, username, dash.CSRFTokenFromContext(r.Context()))
-	layoutComponent := layout.Layout(viewCtx, brokersComponent)
-	if err := layoutComponent.Render(r.Context(), w); err != nil {
-		writeJSONError(w, http.StatusInternalServerError, "Failed to render template: "+err.Error())
-		return
-	}
-}
-
 // GetClusterTopology returns the cluster topology as JSON
 func (h *ClusterHandlers) GetClusterTopology(w http.ResponseWriter, r *http.Request) {
 	topology, err := h.adminServer.GetClusterTopology()
@@ -78,34 +78,6 @@ func (h *MessageQueueHandlers) ShowTopics(w http.ResponseWriter, r *http.Request
 	}
 }

-// ShowSubscribers renders the message queue subscribers page
-func (h *MessageQueueHandlers) ShowSubscribers(w http.ResponseWriter, r *http.Request) {
-	// Get subscribers data
-	subscribersData, err := h.adminServer.GetSubscribers()
-	if err != nil {
-		writeJSONError(w, http.StatusInternalServerError, "Failed to get subscribers: "+err.Error())
-		return
-	}
-
-	// Set username
-	username := dash.UsernameFromContext(r.Context())
-	if username == "" {
-		username = "admin"
-	}
-	subscribersData.Username = username
-
-	// Render HTML template
-	w.Header().Set("Content-Type", "text/html")
-	subscribersComponent := app.Subscribers(*subscribersData)
-	viewCtx := layout.NewViewContext(r, username, dash.CSRFTokenFromContext(r.Context()))
-	layoutComponent := layout.Layout(viewCtx, subscribersComponent)
-	err = layoutComponent.Render(r.Context(), w)
-	if err != nil {
-		writeJSONError(w, http.StatusInternalServerError, "Failed to render template: "+err.Error())
-		return
-	}
-}
-
 // ShowTopicDetails renders the topic details page
 func (h *MessageQueueHandlers) ShowTopicDetails(w http.ResponseWriter, r *http.Request) {
 	// Get topic parameters from URL
@@ -93,6 +93,11 @@ func (h *UserHandlers) UpdateUser(w http.ResponseWriter, r *http.Request) {
 		return
 	}

+	if h.adminServer.IsStaticUser(username) {
+		writeJSONError(w, http.StatusForbidden, "Cannot modify static user "+username+" (loaded from config file)")
+		return
+	}
+
 	var req dash.UpdateUserRequest
 	if err := decodeJSONBody(newJSONMaxReader(w, r), &req); err != nil {
 		writeJSONError(w, http.StatusBadRequest, "Invalid request: "+err.Error())
@@ -120,6 +125,11 @@ func (h *UserHandlers) DeleteUser(w http.ResponseWriter, r *http.Request) {
 		return
 	}

+	if h.adminServer.IsStaticUser(username) {
+		writeJSONError(w, http.StatusForbidden, "Cannot delete static user "+username+" (loaded from config file)")
+		return
+	}
+
 	err := h.adminServer.DeleteObjectStoreUser(username)
 	if err != nil {
 		glog.Errorf("Failed to delete user %s: %v", username, err)
@@ -1,124 +0,0 @@
-package maintenance
-
-import (
-	"fmt"
-
-	"github.com/seaweedfs/seaweedfs/weed/pb/worker_pb"
-)
-
-// VerifyProtobufConfig demonstrates that the protobuf configuration system is working
-func VerifyProtobufConfig() error {
-	// Create configuration manager
-	configManager := NewMaintenanceConfigManager()
-	config := configManager.GetConfig()
-
-	// Verify basic configuration
-	if !config.Enabled {
-		return fmt.Errorf("expected config to be enabled by default")
-	}
-
-	if config.ScanIntervalSeconds != 30*60 {
-		return fmt.Errorf("expected scan interval to be 1800 seconds, got %d", config.ScanIntervalSeconds)
-	}
-
-	// Verify policy configuration
-	if config.Policy == nil {
-		return fmt.Errorf("expected policy to be configured")
-	}
-
-	if config.Policy.GlobalMaxConcurrent != 4 {
-		return fmt.Errorf("expected global max concurrent to be 4, got %d", config.Policy.GlobalMaxConcurrent)
-	}
-
-	// Verify task policies
-	vacuumPolicy := config.Policy.TaskPolicies["vacuum"]
-	if vacuumPolicy == nil {
-		return fmt.Errorf("expected vacuum policy to be configured")
-	}
-
-	if !vacuumPolicy.Enabled {
-		return fmt.Errorf("expected vacuum policy to be enabled")
-	}
-
-	// Verify typed configuration access
-	vacuumConfig := vacuumPolicy.GetVacuumConfig()
-	if vacuumConfig == nil {
-		return fmt.Errorf("expected vacuum config to be accessible")
-	}
-
-	if vacuumConfig.GarbageThreshold != 0.3 {
-		return fmt.Errorf("expected garbage threshold to be 0.3, got %f", vacuumConfig.GarbageThreshold)
-	}
-
-	// Verify helper functions work
-	if !IsTaskEnabled(config.Policy, "vacuum") {
-		return fmt.Errorf("expected vacuum task to be enabled via helper function")
-	}
-
-	maxConcurrent := GetMaxConcurrent(config.Policy, "vacuum")
-	if maxConcurrent != 2 {
-		return fmt.Errorf("expected vacuum max concurrent to be 2, got %d", maxConcurrent)
-	}
-
-	// Verify erasure coding configuration
-	ecPolicy := config.Policy.TaskPolicies["erasure_coding"]
-	if ecPolicy == nil {
-		return fmt.Errorf("expected EC policy to be configured")
-	}
-
-	ecConfig := ecPolicy.GetErasureCodingConfig()
-	if ecConfig == nil {
-		return fmt.Errorf("expected EC config to be accessible")
-	}
-
-	// Verify configurable EC fields only
-	if ecConfig.FullnessRatio <= 0 || ecConfig.FullnessRatio > 1 {
-		return fmt.Errorf("expected EC config to have valid fullness ratio (0-1), got %f", ecConfig.FullnessRatio)
-	}
-
-	return nil
-}
-
-// GetProtobufConfigSummary returns a summary of the current protobuf configuration
-func GetProtobufConfigSummary() string {
-	configManager := NewMaintenanceConfigManager()
-	config := configManager.GetConfig()
-
-	summary := fmt.Sprintf("SeaweedFS Protobuf Maintenance Configuration:\n")
-	summary += fmt.Sprintf("  Enabled: %v\n", config.Enabled)
-	summary += fmt.Sprintf("  Scan Interval: %d seconds\n", config.ScanIntervalSeconds)
-	summary += fmt.Sprintf("  Max Retries: %d\n", config.MaxRetries)
-	summary += fmt.Sprintf("  Global Max Concurrent: %d\n", config.Policy.GlobalMaxConcurrent)
-	summary += fmt.Sprintf("  Task Policies: %d configured\n", len(config.Policy.TaskPolicies))
-
-	for taskType, policy := range config.Policy.TaskPolicies {
-		summary += fmt.Sprintf("    %s: enabled=%v, max_concurrent=%d\n",
-			taskType, policy.Enabled, policy.MaxConcurrent)
-	}
-
-	return summary
-}
-
-// CreateCustomConfig demonstrates creating a custom protobuf configuration
-func CreateCustomConfig() *worker_pb.MaintenanceConfig {
-	return &worker_pb.MaintenanceConfig{
-		Enabled:             true,
-		ScanIntervalSeconds: 60 * 60, // 1 hour
-		MaxRetries:          5,
-		Policy: &worker_pb.MaintenancePolicy{
-			GlobalMaxConcurrent: 8,
-			TaskPolicies: map[string]*worker_pb.TaskPolicy{
-				"custom_vacuum": {
-					Enabled:       true,
-					MaxConcurrent: 4,
-					TaskConfig: &worker_pb.TaskPolicy_VacuumConfig{
-						VacuumConfig: &worker_pb.VacuumTaskConfig{
-							GarbageThreshold:  0.5,
-							MinVolumeAgeHours: 48,
-						},
-					},
-				},
-			},
-		},
-	}
-}
@@ -1,24 +1,9 @@
 package maintenance

 import (
-	"fmt"
-	"time"
-
 	"github.com/seaweedfs/seaweedfs/weed/pb/worker_pb"
 )

-// MaintenanceConfigManager handles protobuf-based configuration
-type MaintenanceConfigManager struct {
-	config *worker_pb.MaintenanceConfig
-}
-
-// NewMaintenanceConfigManager creates a new config manager with defaults
-func NewMaintenanceConfigManager() *MaintenanceConfigManager {
-	return &MaintenanceConfigManager{
-		config: DefaultMaintenanceConfigProto(),
-	}
-}
-
 // DefaultMaintenanceConfigProto returns default configuration as protobuf
 func DefaultMaintenanceConfigProto() *worker_pb.MaintenanceConfig {
 	return &worker_pb.MaintenanceConfig{
@@ -34,253 +19,3 @@ func DefaultMaintenanceConfigProto() *worker_pb.MaintenanceConfig {
 		Policy: nil,
 	}
 }
-
-// GetConfig returns the current configuration
-func (mcm *MaintenanceConfigManager) GetConfig() *worker_pb.MaintenanceConfig {
-	return mcm.config
-}
-
-// Type-safe configuration accessors
-
-// GetVacuumConfig returns vacuum-specific configuration for a task type
-func (mcm *MaintenanceConfigManager) GetVacuumConfig(taskType string) *worker_pb.VacuumTaskConfig {
-	if policy := mcm.getTaskPolicy(taskType); policy != nil {
-		if vacuumConfig := policy.GetVacuumConfig(); vacuumConfig != nil {
-			return vacuumConfig
-		}
-	}
-	// Return defaults if not configured
-	return &worker_pb.VacuumTaskConfig{
-		GarbageThreshold:  0.3,
-		MinVolumeAgeHours: 24,
-	}
-}
-
-// GetErasureCodingConfig returns EC-specific configuration for a task type
-func (mcm *MaintenanceConfigManager) GetErasureCodingConfig(taskType string) *worker_pb.ErasureCodingTaskConfig {
-	if policy := mcm.getTaskPolicy(taskType); policy != nil {
-		if ecConfig := policy.GetErasureCodingConfig(); ecConfig != nil {
-			return ecConfig
-		}
-	}
-	// Return defaults if not configured
-	return &worker_pb.ErasureCodingTaskConfig{
-		FullnessRatio:    0.95,
-		QuietForSeconds:  3600,
-		MinVolumeSizeMb:  100,
-		CollectionFilter: "",
-	}
-}
-
-// GetBalanceConfig returns balance-specific configuration for a task type
-func (mcm *MaintenanceConfigManager) GetBalanceConfig(taskType string) *worker_pb.BalanceTaskConfig {
-	if policy := mcm.getTaskPolicy(taskType); policy != nil {
-		if balanceConfig := policy.GetBalanceConfig(); balanceConfig != nil {
-			return balanceConfig
-		}
-	}
-	// Return defaults if not configured
-	return &worker_pb.BalanceTaskConfig{
-		ImbalanceThreshold: 0.2,
-		MinServerCount:     2,
-	}
-}
-
-// GetReplicationConfig returns replication-specific configuration for a task type
-func (mcm *MaintenanceConfigManager) GetReplicationConfig(taskType string) *worker_pb.ReplicationTaskConfig {
-	if policy := mcm.getTaskPolicy(taskType); policy != nil {
-		if replicationConfig := policy.GetReplicationConfig(); replicationConfig != nil {
-			return replicationConfig
-		}
-	}
-	// Return defaults if not configured
-	return &worker_pb.ReplicationTaskConfig{
-		TargetReplicaCount: 2,
-	}
-}
-
-// Typed convenience methods for getting task configurations
-
-// GetVacuumTaskConfigForType returns vacuum configuration for a specific task type
-func (mcm *MaintenanceConfigManager) GetVacuumTaskConfigForType(taskType string) *worker_pb.VacuumTaskConfig {
-	return GetVacuumTaskConfig(mcm.config.Policy, MaintenanceTaskType(taskType))
-}
-
-// GetErasureCodingTaskConfigForType returns erasure coding configuration for a specific task type
-func (mcm *MaintenanceConfigManager) GetErasureCodingTaskConfigForType(taskType string) *worker_pb.ErasureCodingTaskConfig {
-	return GetErasureCodingTaskConfig(mcm.config.Policy, MaintenanceTaskType(taskType))
-}
-
-// GetBalanceTaskConfigForType returns balance configuration for a specific task type
-func (mcm *MaintenanceConfigManager) GetBalanceTaskConfigForType(taskType string) *worker_pb.BalanceTaskConfig {
-	return GetBalanceTaskConfig(mcm.config.Policy, MaintenanceTaskType(taskType))
-}
-
-// GetReplicationTaskConfigForType returns replication configuration for a specific task type
-func (mcm *MaintenanceConfigManager) GetReplicationTaskConfigForType(taskType string) *worker_pb.ReplicationTaskConfig {
-	return GetReplicationTaskConfig(mcm.config.Policy, MaintenanceTaskType(taskType))
-}
-
-// Helper methods
-
-func (mcm *MaintenanceConfigManager) getTaskPolicy(taskType string) *worker_pb.TaskPolicy {
-	if mcm.config.Policy != nil && mcm.config.Policy.TaskPolicies != nil {
-		return mcm.config.Policy.TaskPolicies[taskType]
-	}
-	return nil
-}
-
-// IsTaskEnabled returns whether a task type is enabled
-func (mcm *MaintenanceConfigManager) IsTaskEnabled(taskType string) bool {
-	if policy := mcm.getTaskPolicy(taskType); policy != nil {
-		return policy.Enabled
-	}
-	return false
-}
-
-// GetMaxConcurrent returns the max concurrent limit for a task type
-func (mcm *MaintenanceConfigManager) GetMaxConcurrent(taskType string) int32 {
-	if policy := mcm.getTaskPolicy(taskType); policy != nil {
-		return policy.MaxConcurrent
-	}
-	return 1 // Default
-}
-
-// GetRepeatInterval returns the repeat interval for a task type in seconds
-func (mcm *MaintenanceConfigManager) GetRepeatInterval(taskType string) int32 {
-	if policy := mcm.getTaskPolicy(taskType); policy != nil {
-		return policy.RepeatIntervalSeconds
-	}
-	return mcm.config.Policy.DefaultRepeatIntervalSeconds
-}
-
-// GetCheckInterval returns the check interval for a task type in seconds
-func (mcm *MaintenanceConfigManager) GetCheckInterval(taskType string) int32 {
-	if policy := mcm.getTaskPolicy(taskType); policy != nil {
-		return policy.CheckIntervalSeconds
-	}
-	return mcm.config.Policy.DefaultCheckIntervalSeconds
-}
-
-// Duration accessor methods
-
-// GetScanInterval returns the scan interval as a time.Duration
-func (mcm *MaintenanceConfigManager) GetScanInterval() time.Duration {
-	return time.Duration(mcm.config.ScanIntervalSeconds) * time.Second
-}
-
-// GetWorkerTimeout returns the worker timeout as a time.Duration
-func (mcm *MaintenanceConfigManager) GetWorkerTimeout() time.Duration {
-	return time.Duration(mcm.config.WorkerTimeoutSeconds) * time.Second
-}
-
-// GetTaskTimeout returns the task timeout as a time.Duration
-func (mcm *MaintenanceConfigManager) GetTaskTimeout() time.Duration {
-	return time.Duration(mcm.config.TaskTimeoutSeconds) * time.Second
-}
-
-// GetRetryDelay returns the retry delay as a time.Duration
-func (mcm *MaintenanceConfigManager) GetRetryDelay() time.Duration {
-	return time.Duration(mcm.config.RetryDelaySeconds) * time.Second
-}
-
-// GetCleanupInterval returns the cleanup interval as a time.Duration
-func (mcm *MaintenanceConfigManager) GetCleanupInterval() time.Duration {
-	return time.Duration(mcm.config.CleanupIntervalSeconds) * time.Second
-}
-
-// GetTaskRetention returns the task retention period as a time.Duration
-func (mcm *MaintenanceConfigManager) GetTaskRetention() time.Duration {
-	return time.Duration(mcm.config.TaskRetentionSeconds) * time.Second
-}
-
-// ValidateMaintenanceConfigWithSchema validates protobuf maintenance configuration using ConfigField rules
-func ValidateMaintenanceConfigWithSchema(config *worker_pb.MaintenanceConfig) error {
-	if config == nil {
-		return fmt.Errorf("configuration cannot be nil")
-	}
-
-	// Get the schema to access field validation rules
-	schema := GetMaintenanceConfigSchema()
-
-	// Validate each field individually using the ConfigField rules
-	if err := validateFieldWithSchema(schema, "enabled", config.Enabled); err != nil {
-		return err
-	}
-
-	if err := validateFieldWithSchema(schema, "scan_interval_seconds", int(config.ScanIntervalSeconds)); err != nil {
-		return err
-	}
-
-	if err := validateFieldWithSchema(schema, "worker_timeout_seconds", int(config.WorkerTimeoutSeconds)); err != nil {
-		return err
-	}
-
-	if err := validateFieldWithSchema(schema, "task_timeout_seconds", int(config.TaskTimeoutSeconds)); err != nil {
-		return err
-	}
-
-	if err := validateFieldWithSchema(schema, "retry_delay_seconds", int(config.RetryDelaySeconds)); err != nil {
-		return err
-	}
-
-	if err := validateFieldWithSchema(schema, "max_retries", int(config.MaxRetries)); err != nil {
-		return err
-	}
-
-	if err := validateFieldWithSchema(schema, "cleanup_interval_seconds", int(config.CleanupIntervalSeconds)); err != nil {
-		return err
-	}
-
-	if err := validateFieldWithSchema(schema, "task_retention_seconds", int(config.TaskRetentionSeconds)); err != nil {
-		return err
-	}
-
-	// Validate policy fields if present
-	if config.Policy != nil {
-		// Note: These field names might need to be adjusted based on the actual schema
-		if err := validatePolicyField("global_max_concurrent", int(config.Policy.GlobalMaxConcurrent)); err != nil {
-			return err
-		}
-
-		if err := validatePolicyField("default_repeat_interval_seconds", int(config.Policy.DefaultRepeatIntervalSeconds)); err != nil {
-			return err
-		}
-
-		if err := validatePolicyField("default_check_interval_seconds", int(config.Policy.DefaultCheckIntervalSeconds)); err != nil {
-			return err
-		}
-	}
-
-	return nil
-}
-
-// validateFieldWithSchema validates a single field using its ConfigField definition
-func validateFieldWithSchema(schema *MaintenanceConfigSchema, fieldName string, value interface{}) error {
-	field := schema.GetFieldByName(fieldName)
-	if field == nil {
-		// Field not in schema, skip validation
-		return nil
-	}
-
-	return field.ValidateValue(value)
-}
-
-// validatePolicyField validates policy fields (simplified validation for now)
-func validatePolicyField(fieldName string, value int) error {
-	switch fieldName {
-	case "global_max_concurrent":
-		if value < 1 || value > 20 {
-			return fmt.Errorf("Global Max Concurrent must be between 1 and 20, got %d", value)
-		}
-	case "default_repeat_interval":
-		if value < 1 || value > 168 {
-			return fmt.Errorf("Default Repeat Interval must be between 1 and 168 hours, got %d", value)
-		}
-	case "default_check_interval":
-		if value < 1 || value > 168 {
-			return fmt.Errorf("Default Check Interval must be between 1 and 168 hours, got %d", value)
-		}
-	}
-	return nil
-}
@@ -1055,28 +1055,6 @@ func (mq *MaintenanceQueue) getMaxConcurrentForTaskType(taskType MaintenanceTask
 	return 1
 }

-// getRunningTasks returns all currently running tasks
-func (mq *MaintenanceQueue) getRunningTasks() []*MaintenanceTask {
-	var runningTasks []*MaintenanceTask
-	for _, task := range mq.tasks {
-		if task.Status == TaskStatusAssigned || task.Status == TaskStatusInProgress {
-			runningTasks = append(runningTasks, task)
-		}
-	}
-	return runningTasks
-}
-
-// getAvailableWorkers returns all workers that can take more work
-func (mq *MaintenanceQueue) getAvailableWorkers() []*MaintenanceWorker {
-	var availableWorkers []*MaintenanceWorker
-	for _, worker := range mq.workers {
-		if worker.Status == "active" && worker.CurrentLoad < worker.MaxConcurrent {
-			availableWorkers = append(availableWorkers, worker)
-		}
-	}
-	return availableWorkers
-}
-
 // trackPendingOperation adds a task to the pending operations tracker
 func (mq *MaintenanceQueue) trackPendingOperation(task *MaintenanceTask) {
 	if mq.integration == nil {
@@ -2,15 +2,11 @@ package maintenance

 import (
 	"html/template"
-	"sort"
 	"sync"
 	"time"

-	"github.com/seaweedfs/seaweedfs/weed/glog"
 	"github.com/seaweedfs/seaweedfs/weed/pb/master_pb"
 	"github.com/seaweedfs/seaweedfs/weed/pb/worker_pb"
-	"github.com/seaweedfs/seaweedfs/weed/worker/tasks"
-	"github.com/seaweedfs/seaweedfs/weed/worker/types"
 )

 // AdminClient interface defines what the maintenance system needs from the admin server
@@ -21,51 +17,6 @@ type AdminClient interface {
 // MaintenanceTaskType represents different types of maintenance operations
 type MaintenanceTaskType string

-// GetRegisteredMaintenanceTaskTypes returns all registered task types as MaintenanceTaskType values
-// sorted alphabetically for consistent menu ordering
-func GetRegisteredMaintenanceTaskTypes() []MaintenanceTaskType {
-	typesRegistry := tasks.GetGlobalTypesRegistry()
-	var taskTypes []MaintenanceTaskType
-
-	for workerTaskType := range typesRegistry.GetAllDetectors() {
-		maintenanceTaskType := MaintenanceTaskType(string(workerTaskType))
-		taskTypes = append(taskTypes, maintenanceTaskType)
-	}
-
-	// Sort task types alphabetically to ensure consistent menu ordering
-	sort.Slice(taskTypes, func(i, j int) bool {
-		return string(taskTypes[i]) < string(taskTypes[j])
-	})
-
-	return taskTypes
-}
-
-// GetMaintenanceTaskType returns a specific task type if it's registered, or empty string if not found
-func GetMaintenanceTaskType(taskTypeName string) MaintenanceTaskType {
-	typesRegistry := tasks.GetGlobalTypesRegistry()
-
-	for workerTaskType := range typesRegistry.GetAllDetectors() {
-		if string(workerTaskType) == taskTypeName {
-			return MaintenanceTaskType(taskTypeName)
-		}
-	}
-
-	return MaintenanceTaskType("")
-}
-
-// IsMaintenanceTaskTypeRegistered checks if a task type is registered
-func IsMaintenanceTaskTypeRegistered(taskType MaintenanceTaskType) bool {
-	typesRegistry := tasks.GetGlobalTypesRegistry()
-
-	for workerTaskType := range typesRegistry.GetAllDetectors() {
-		if string(workerTaskType) == string(taskType) {
-			return true
-		}
-	}
-
-	return false
-}
-
 // MaintenanceTaskPriority represents task execution priority
 type MaintenanceTaskPriority int

@@ -200,14 +151,6 @@ func GetTaskPolicy(mp *MaintenancePolicy, taskType MaintenanceTaskType) *TaskPol
 	return mp.TaskPolicies[string(taskType)]
 }

-// SetTaskPolicy sets the policy for a specific task type
-func SetTaskPolicy(mp *MaintenancePolicy, taskType MaintenanceTaskType, policy *TaskPolicy) {
-	if mp.TaskPolicies == nil {
-		mp.TaskPolicies = make(map[string]*TaskPolicy)
-	}
-	mp.TaskPolicies[string(taskType)] = policy
-}
-
 // IsTaskEnabled returns whether a task type is enabled
 func IsTaskEnabled(mp *MaintenancePolicy, taskType MaintenanceTaskType) bool {
 	policy := GetTaskPolicy(mp, taskType)
@@ -235,84 +178,6 @@ func GetRepeatInterval(mp *MaintenancePolicy, taskType MaintenanceTaskType) int
 	return int(policy.RepeatIntervalSeconds)
 }

-// GetVacuumTaskConfig returns the vacuum task configuration
-func GetVacuumTaskConfig(mp *MaintenancePolicy, taskType MaintenanceTaskType) *worker_pb.VacuumTaskConfig {
-	policy := GetTaskPolicy(mp, taskType)
-	if policy == nil {
-		return nil
-	}
-	return policy.GetVacuumConfig()
-}
-
-// GetErasureCodingTaskConfig returns the erasure coding task configuration
-func GetErasureCodingTaskConfig(mp *MaintenancePolicy, taskType MaintenanceTaskType) *worker_pb.ErasureCodingTaskConfig {
-	policy := GetTaskPolicy(mp, taskType)
-	if policy == nil {
-		return nil
-	}
-	return policy.GetErasureCodingConfig()
-}
-
-// GetBalanceTaskConfig returns the balance task configuration
-func GetBalanceTaskConfig(mp *MaintenancePolicy, taskType MaintenanceTaskType) *worker_pb.BalanceTaskConfig {
-	policy := GetTaskPolicy(mp, taskType)
-	if policy == nil {
-		return nil
-	}
-	return policy.GetBalanceConfig()
-}
-
-// GetReplicationTaskConfig returns the replication task configuration
-func GetReplicationTaskConfig(mp *MaintenancePolicy, taskType MaintenanceTaskType) *worker_pb.ReplicationTaskConfig {
-	policy := GetTaskPolicy(mp, taskType)
-	if policy == nil {
-		return nil
-	}
-	return policy.GetReplicationConfig()
-}
-
-// Note: GetTaskConfig was removed - use typed getters: GetVacuumTaskConfig, GetErasureCodingTaskConfig, GetBalanceTaskConfig, or GetReplicationTaskConfig
-
-// SetVacuumTaskConfig sets the vacuum task configuration
-func SetVacuumTaskConfig(mp *MaintenancePolicy, taskType MaintenanceTaskType, config *worker_pb.VacuumTaskConfig) {
-	policy := GetTaskPolicy(mp, taskType)
-	if policy != nil {
-		policy.TaskConfig = &worker_pb.TaskPolicy_VacuumConfig{
-			VacuumConfig: config,
-		}
-	}
-}
-
-// SetErasureCodingTaskConfig sets the erasure coding task configuration
-func SetErasureCodingTaskConfig(mp *MaintenancePolicy, taskType MaintenanceTaskType, config *worker_pb.ErasureCodingTaskConfig) {
-	policy := GetTaskPolicy(mp, taskType)
-	if policy != nil {
-		policy.TaskConfig = &worker_pb.TaskPolicy_ErasureCodingConfig{
-			ErasureCodingConfig: config,
-		}
-	}
-}
-
-// SetBalanceTaskConfig sets the balance task configuration
-func SetBalanceTaskConfig(mp *MaintenancePolicy, taskType MaintenanceTaskType, config *worker_pb.BalanceTaskConfig) {
-	policy := GetTaskPolicy(mp, taskType)
-	if policy != nil {
-		policy.TaskConfig = &worker_pb.TaskPolicy_BalanceConfig{
-			BalanceConfig: config,
-		}
-	}
-}
-
-// SetReplicationTaskConfig sets the replication task configuration
-func SetReplicationTaskConfig(mp *MaintenancePolicy, taskType MaintenanceTaskType, config *worker_pb.ReplicationTaskConfig) {
-	policy := GetTaskPolicy(mp, taskType)
-	if policy != nil {
-		policy.TaskConfig = &worker_pb.TaskPolicy_ReplicationConfig{
-			ReplicationConfig: config,
-		}
-	}
-}
-
 // SetTaskConfig sets a configuration value for a task type (legacy method - use typed setters above)
 // Note: SetTaskConfig was removed - use typed setters: SetVacuumTaskConfig, SetErasureCodingTaskConfig, SetBalanceTaskConfig, or SetReplicationTaskConfig

@@ -475,180 +340,6 @@ type ClusterReplicationTask struct {
 	Metadata        map[string]string `json:"metadata,omitempty"`
 }

-// BuildMaintenancePolicyFromTasks creates a maintenance policy with configurations
-// from all registered tasks using their UI providers
-func BuildMaintenancePolicyFromTasks() *MaintenancePolicy {
-	policy := &MaintenancePolicy{
-		TaskPolicies:                 make(map[string]*TaskPolicy),
-		GlobalMaxConcurrent:          4,
-		DefaultRepeatIntervalSeconds: 6 * 3600,  // 6 hours in seconds
-		DefaultCheckIntervalSeconds:  12 * 3600, // 12 hours in seconds
-	}
-
-	// Get all registered task types from the UI registry
-	uiRegistry := tasks.GetGlobalUIRegistry()
-	typesRegistry := tasks.GetGlobalTypesRegistry()
-
-	for taskType, provider := range uiRegistry.GetAllProviders() {
-		// Convert task type to maintenance task type
-		maintenanceTaskType := MaintenanceTaskType(string(taskType))
-
-		// Get the default configuration from the UI provider
-		defaultConfig := provider.GetCurrentConfig()
-
-		// Create task policy from UI configuration
-		taskPolicy := &TaskPolicy{
-			Enabled:               true, // Default enabled
-			MaxConcurrent:         2,    // Default concurrency
-			RepeatIntervalSeconds: policy.DefaultRepeatIntervalSeconds,
-			CheckIntervalSeconds:  policy.DefaultCheckIntervalSeconds,
-		}
-
-		// Extract configuration using TaskConfig interface - no more map conversions!
-		if taskConfig, ok := defaultConfig.(interface{ ToTaskPolicy() *worker_pb.TaskPolicy }); ok {
-			// Use protobuf directly for clean, type-safe config extraction
-			pbTaskPolicy := taskConfig.ToTaskPolicy()
-			taskPolicy.Enabled = pbTaskPolicy.Enabled
-			taskPolicy.MaxConcurrent = pbTaskPolicy.MaxConcurrent
-			if pbTaskPolicy.RepeatIntervalSeconds > 0 {
-				taskPolicy.RepeatIntervalSeconds = pbTaskPolicy.RepeatIntervalSeconds
-			}
-			if pbTaskPolicy.CheckIntervalSeconds > 0 {
-				taskPolicy.CheckIntervalSeconds = pbTaskPolicy.CheckIntervalSeconds
-			}
-		}
-
-		// Also get defaults from scheduler if available (using types.TaskScheduler explicitly)
-		var scheduler types.TaskScheduler = typesRegistry.GetScheduler(taskType)
-		if scheduler != nil {
-			if taskPolicy.MaxConcurrent <= 0 {
-				taskPolicy.MaxConcurrent = int32(scheduler.GetMaxConcurrent())
-			}
-			// Convert default repeat interval to seconds
-			if repeatInterval := scheduler.GetDefaultRepeatInterval(); repeatInterval > 0 {
-				taskPolicy.RepeatIntervalSeconds = int32(repeatInterval.Seconds())
-			}
-		}
-
-		// Also get defaults from detector if available (using types.TaskDetector explicitly)
-		var detector types.TaskDetector = typesRegistry.GetDetector(taskType)
-		if detector != nil {
-			// Convert scan interval to check interval (seconds)
-			if scanInterval := detector.ScanInterval(); scanInterval > 0 {
-				taskPolicy.CheckIntervalSeconds = int32(scanInterval.Seconds())
-			}
-		}
-
-		policy.TaskPolicies[string(maintenanceTaskType)] = taskPolicy
-		glog.V(3).Infof("Built policy for task type %s: enabled=%v, max_concurrent=%d",
-			maintenanceTaskType, taskPolicy.Enabled, taskPolicy.MaxConcurrent)
-	}
-
-	glog.V(2).Infof("Built maintenance policy with %d task configurations", len(policy.TaskPolicies))
-	return policy
-}
-
-// SetPolicyFromTasks sets the maintenance policy from registered tasks
-func SetPolicyFromTasks(policy *MaintenancePolicy) {
-	if policy == nil {
-		return
-	}
-
-	// Build new policy from tasks
-	newPolicy := BuildMaintenancePolicyFromTasks()
-
-	// Copy task policies
-	policy.TaskPolicies = newPolicy.TaskPolicies
-
-	glog.V(1).Infof("Updated maintenance policy with %d task configurations from registered tasks", len(policy.TaskPolicies))
-}
-
-// GetTaskIcon returns the icon CSS class for a task type from its UI provider
-func GetTaskIcon(taskType MaintenanceTaskType) string {
-	typesRegistry := tasks.GetGlobalTypesRegistry()
-	uiRegistry := tasks.GetGlobalUIRegistry()
-
-	// Convert MaintenanceTaskType to TaskType
-	for workerTaskType := range typesRegistry.GetAllDetectors() {
-		if string(workerTaskType) == string(taskType) {
-			// Get the UI provider for this task type
-			provider := uiRegistry.GetProvider(workerTaskType)
-			if provider != nil {
-				return provider.GetIcon()
-			}
-			break
-		}
-	}
-
-	// Default icon if no UI provider found
-	return "fas fa-cog text-muted"
-}
-
-// GetTaskDisplayName returns the display name for a task type from its UI provider
-func GetTaskDisplayName(taskType MaintenanceTaskType) string {
-	typesRegistry := tasks.GetGlobalTypesRegistry()
-	uiRegistry := tasks.GetGlobalUIRegistry()
-
-	// Convert MaintenanceTaskType to TaskType
-	for workerTaskType := range typesRegistry.GetAllDetectors() {
-		if string(workerTaskType) == string(taskType) {
-			// Get the UI provider for this task type
-			provider := uiRegistry.GetProvider(workerTaskType)
-			if provider != nil {
-				return provider.GetDisplayName()
-			}
-			break
-		}
-	}
-
-	// Fallback to the task type string
-	return string(taskType)
-}
-
-// GetTaskDescription returns the description for a task type from its UI provider
-func GetTaskDescription(taskType MaintenanceTaskType) string {
-	typesRegistry := tasks.GetGlobalTypesRegistry()
-	uiRegistry := tasks.GetGlobalUIRegistry()
-
-	// Convert MaintenanceTaskType to TaskType
-	for workerTaskType := range typesRegistry.GetAllDetectors() {
-		if string(workerTaskType) == string(taskType) {
-			// Get the UI provider for this task type
-			provider := uiRegistry.GetProvider(workerTaskType)
-			if provider != nil {
-				return provider.GetDescription()
-			}
-			break
-		}
-	}
-
-	// Fallback to a generic description
-	return "Configure detailed settings for " + string(taskType) + " tasks."
-}
-
-// BuildMaintenanceMenuItems creates menu items for all registered task types
-func BuildMaintenanceMenuItems() []*MaintenanceMenuItem {
-	var menuItems []*MaintenanceMenuItem
-
-	// Get all registered task types
-	registeredTypes := GetRegisteredMaintenanceTaskTypes()
-
-	for _, taskType := range registeredTypes {
-		menuItem := &MaintenanceMenuItem{
-			TaskType:    taskType,
-			DisplayName: GetTaskDisplayName(taskType),
-			Description: GetTaskDescription(taskType),
-			Icon:        GetTaskIcon(taskType),
-			IsEnabled:   IsMaintenanceTaskTypeRegistered(taskType),
-			Path:        "/maintenance/config/" + string(taskType),
-		}
-
-		menuItems = append(menuItems, menuItem)
-	}
-
-	return menuItems
-}
-
 // Helper functions to extract configuration fields

 // Note: Removed getVacuumConfigField, getErasureCodingConfigField, getBalanceConfigField, getReplicationConfigField
@@ -1,421 +0,0 @@
-package maintenance
-
-import (
-	"context"
-	"fmt"
-	"os"
-	"sync"
-	"time"
-
-	"github.com/seaweedfs/seaweedfs/weed/glog"
-	"github.com/seaweedfs/seaweedfs/weed/worker"
-	"github.com/seaweedfs/seaweedfs/weed/worker/tasks"
-	"github.com/seaweedfs/seaweedfs/weed/worker/types"
-
-	// Import task packages to trigger their auto-registration
-	_ "github.com/seaweedfs/seaweedfs/weed/worker/tasks/balance"
-	_ "github.com/seaweedfs/seaweedfs/weed/worker/tasks/erasure_coding"
-	_ "github.com/seaweedfs/seaweedfs/weed/worker/tasks/vacuum"
-)
-
-// MaintenanceWorkerService manages maintenance task execution
-// TaskExecutor defines the function signature for task execution
-type TaskExecutor func(*MaintenanceWorkerService, *MaintenanceTask) error
-
-// TaskExecutorFactory creates a task executor for a given worker service
-type TaskExecutorFactory func() TaskExecutor
-
-// Global registry for task executor factories
-var taskExecutorFactories = make(map[MaintenanceTaskType]TaskExecutorFactory)
-var executorRegistryMutex sync.RWMutex
-var executorRegistryInitOnce sync.Once
-
-// initializeExecutorFactories dynamically registers executor factories for all auto-registered task types
-func initializeExecutorFactories() {
-	executorRegistryInitOnce.Do(func() {
-		// Get all registered task types from the global registry
-		typesRegistry := tasks.GetGlobalTypesRegistry()
-
-		var taskTypes []MaintenanceTaskType
-		for workerTaskType := range typesRegistry.GetAllDetectors() {
-			// Convert types.TaskType to MaintenanceTaskType by string conversion
-			maintenanceTaskType := MaintenanceTaskType(string(workerTaskType))
-			taskTypes = append(taskTypes, maintenanceTaskType)
-		}
-
-		// Register generic executor for all task types
-		for _, taskType := range taskTypes {
-			RegisterTaskExecutorFactory(taskType, createGenericTaskExecutor)
-		}
-
-		glog.V(1).Infof("Dynamically registered generic task executor for %d task types: %v", len(taskTypes), taskTypes)
-	})
-}
-
-// RegisterTaskExecutorFactory registers a factory function for creating task executors
-func RegisterTaskExecutorFactory(taskType MaintenanceTaskType, factory TaskExecutorFactory) {
-	executorRegistryMutex.Lock()
-	defer executorRegistryMutex.Unlock()
-	taskExecutorFactories[taskType] = factory
-	glog.V(2).Infof("Registered executor factory for task type: %s", taskType)
-}
-
-// GetTaskExecutorFactory returns the factory for a task type
-func GetTaskExecutorFactory(taskType MaintenanceTaskType) (TaskExecutorFactory, bool) {
-	// Ensure executor factories are initialized
-	initializeExecutorFactories()
-
-	executorRegistryMutex.RLock()
-	defer executorRegistryMutex.RUnlock()
-	factory, exists := taskExecutorFactories[taskType]
-	return factory, exists
-}
-
-// GetSupportedExecutorTaskTypes returns all task types with registered executor factories
-func GetSupportedExecutorTaskTypes() []MaintenanceTaskType {
-	// Ensure executor factories are initialized
-	initializeExecutorFactories()
-
-	executorRegistryMutex.RLock()
-	defer executorRegistryMutex.RUnlock()
-
-	taskTypes := make([]MaintenanceTaskType, 0, len(taskExecutorFactories))
-	for taskType := range taskExecutorFactories {
-		taskTypes = append(taskTypes, taskType)
-	}
-	return taskTypes
-}
-
-// createGenericTaskExecutor creates a generic task executor that uses the task registry
-func createGenericTaskExecutor() TaskExecutor {
-	return func(mws *MaintenanceWorkerService, task *MaintenanceTask) error {
-		return mws.executeGenericTask(task)
-	}
-}
-
-// init does minimal initialization - actual registration happens lazily
-func init() {
-	// Executor factory registration will happen lazily when first accessed
-	glog.V(1).Infof("Maintenance worker initialized - executor factories will be registered on first access")
-}
-
-type MaintenanceWorkerService struct {
-	workerID      string
-	address       string
-	adminServer   string
-	capabilities  []MaintenanceTaskType
-	maxConcurrent int
-	currentTasks  map[string]*MaintenanceTask
-	queue         *MaintenanceQueue
-	adminClient   AdminClient
-	running       bool
-	stopChan      chan struct{}
-
-	// Task execution registry
-	taskExecutors map[MaintenanceTaskType]TaskExecutor
-
-	// Task registry for creating task instances
-	taskRegistry *tasks.TaskRegistry
-}
-
-// NewMaintenanceWorkerService creates a new maintenance worker service
-func NewMaintenanceWorkerService(workerID, address, adminServer string) *MaintenanceWorkerService {
-	// Get all registered maintenance task types dynamically
-	capabilities := GetRegisteredMaintenanceTaskTypes()
-
-	worker := &MaintenanceWorkerService{
-		workerID:      workerID,
-		address:       address,
-		adminServer:   adminServer,
-		capabilities:  capabilities,
-		maxConcurrent: 2, // Default concurrent task limit
-		currentTasks:  make(map[string]*MaintenanceTask),
-		stopChan:      make(chan struct{}),
-		taskExecutors: make(map[MaintenanceTaskType]TaskExecutor),
-		taskRegistry:  tasks.GetGlobalTaskRegistry(), // Use global registry with auto-registered tasks
-	}
-
-	// Initialize task executor registry
-	worker.initializeTaskExecutors()
-
-	glog.V(1).Infof("Created maintenance worker with %d registered task types", len(worker.taskRegistry.GetAll()))
-
-	return worker
-}
-
-// executeGenericTask executes a task using the task registry instead of hardcoded methods
-func (mws *MaintenanceWorkerService) executeGenericTask(task *MaintenanceTask) error {
-	glog.V(2).Infof("Executing generic task %s: %s for volume %d", task.ID, task.Type, task.VolumeID)
-
-	// Validate that task has proper typed parameters
-	if task.TypedParams == nil {
-		return fmt.Errorf("task %s has no typed parameters - task was not properly planned (insufficient destinations)", task.ID)
-	}
-
-	// Convert MaintenanceTask to types.TaskType
-	taskType := types.TaskType(string(task.Type))
-
-	// Create task instance using the registry
-	taskInstance, err := mws.taskRegistry.Get(taskType).Create(task.TypedParams)
-	if err != nil {
-		return fmt.Errorf("failed to create task instance: %w", err)
-	}
-
-	// Update progress to show task has started
-	mws.updateTaskProgress(task.ID, 5)
-
-	// Execute the task
-	err = taskInstance.Execute(context.Background(), task.TypedParams)
-	if err != nil {
-		return fmt.Errorf("task execution failed: %w", err)
-	}
-
-	// Update progress to show completion
-	mws.updateTaskProgress(task.ID, 100)
-
-	glog.V(2).Infof("Generic task %s completed successfully", task.ID)
-	return nil
-}
-
-// initializeTaskExecutors sets up the task execution registry dynamically
-func (mws *MaintenanceWorkerService) initializeTaskExecutors() {
-	mws.taskExecutors = make(map[MaintenanceTaskType]TaskExecutor)
-
-	// Get all registered executor factories and create executors
-	executorRegistryMutex.RLock()
-	defer executorRegistryMutex.RUnlock()
-
-	for taskType, factory := range taskExecutorFactories {
-		executor := factory()
-		mws.taskExecutors[taskType] = executor
-		glog.V(3).Infof("Initialized executor for task type: %s", taskType)
-	}
-
-	glog.V(2).Infof("Initialized %d task executors", len(mws.taskExecutors))
-}
-
-// RegisterTaskExecutor allows dynamic registration of new task executors
-func (mws *MaintenanceWorkerService) RegisterTaskExecutor(taskType MaintenanceTaskType, executor TaskExecutor) {
-	if mws.taskExecutors == nil {
-		mws.taskExecutors = make(map[MaintenanceTaskType]TaskExecutor)
-	}
-	mws.taskExecutors[taskType] = executor
-	glog.V(1).Infof("Registered executor for task type: %s", taskType)
-}
-
-// GetSupportedTaskTypes returns all task types that this worker can execute
-func (mws *MaintenanceWorkerService) GetSupportedTaskTypes() []MaintenanceTaskType {
-	return GetSupportedExecutorTaskTypes()
-}
-
-// Start begins the worker service
-func (mws *MaintenanceWorkerService) Start() error {
-	mws.running = true
-
-	// Register with admin server
-	worker := &MaintenanceWorker{
-		ID:            mws.workerID,
-		Address:       mws.address,
-		Capabilities:  mws.capabilities,
-		MaxConcurrent: mws.maxConcurrent,
-	}
-
-	if mws.queue != nil {
-		mws.queue.RegisterWorker(worker)
-	}
-
-	// Start worker loop
-	go mws.workerLoop()
-
-	glog.Infof("Maintenance worker %s started at %s", mws.workerID, mws.address)
-	return nil
-}
-
-// Stop terminates the worker service
-func (mws *MaintenanceWorkerService) Stop() {
-	mws.running = false
-	close(mws.stopChan)
-
-	// Wait for current tasks to complete or timeout
-	timeout := time.NewTimer(30 * time.Second)
-	defer timeout.Stop()
-
-	for len(mws.currentTasks) > 0 {
-		select {
-		case <-timeout.C:
-			glog.Warningf("Worker %s stopping with %d tasks still running", mws.workerID, len(mws.currentTasks))
-			return
-		case <-time.After(time.Second):
-			// Check again
-		}
-	}
-
-	glog.Infof("Maintenance worker %s stopped", mws.workerID)
-}
-
-// workerLoop is the main worker event loop
-func (mws *MaintenanceWorkerService) workerLoop() {
-	heartbeatTicker := time.NewTicker(30 * time.Second)
-	defer heartbeatTicker.Stop()
-
-	taskRequestTicker := time.NewTicker(5 * time.Second)
-	defer taskRequestTicker.Stop()
-
-	for mws.running {
-		select {
-		case <-mws.stopChan:
-			return
-		case <-heartbeatTicker.C:
-			mws.sendHeartbeat()
-		case <-taskRequestTicker.C:
-			mws.requestTasks()
-		}
-	}
-}
-
-// sendHeartbeat sends heartbeat to admin server
-func (mws *MaintenanceWorkerService) sendHeartbeat() {
-	if mws.queue != nil {
-		mws.queue.UpdateWorkerHeartbeat(mws.workerID)
-	}
-}
-
-// requestTasks requests new tasks from the admin server
-func (mws *MaintenanceWorkerService) requestTasks() {
-	if len(mws.currentTasks) >= mws.maxConcurrent {
-		return // Already at capacity
-	}
-
-	if mws.queue != nil {
-		task := mws.queue.GetNextTask(mws.workerID, mws.capabilities)
-		if task != nil {
-			mws.executeTask(task)
-		}
-	}
-}
-
-// executeTask executes a maintenance task
-func (mws *MaintenanceWorkerService) executeTask(task *MaintenanceTask) {
-	mws.currentTasks[task.ID] = task
-
-	go func() {
-		defer func() {
-			delete(mws.currentTasks, task.ID)
-		}()
-
-		glog.Infof("Worker %s executing task %s: %s", mws.workerID, task.ID, task.Type)
-
-		// Execute task using dynamic executor registry
-		var err error
-		if executor, exists := mws.taskExecutors[task.Type]; exists {
-			err = executor(mws, task)
-		} else {
-			err = fmt.Errorf("unsupported task type: %s", task.Type)
-			glog.Errorf("No executor registered for task type: %s", task.Type)
-		}
-
-		// Report task completion
-		if mws.queue != nil {
-			errorMsg := ""
-			if err != nil {
-				errorMsg = err.Error()
-			}
-			mws.queue.CompleteTask(task.ID, errorMsg)
-		}
-
-		if err != nil {
-			glog.Errorf("Worker %s failed to execute task %s: %v", mws.workerID, task.ID, err)
-		} else {
-			glog.Infof("Worker %s completed task %s successfully", mws.workerID, task.ID)
-		}
-	}()
-}
-
-// updateTaskProgress updates the progress of a task
-func (mws *MaintenanceWorkerService) updateTaskProgress(taskID string, progress float64) {
-	if mws.queue != nil {
-		mws.queue.UpdateTaskProgress(taskID, progress)
-	}
-}
-
-// GetStatus returns the current status of the worker
-func (mws *MaintenanceWorkerService) GetStatus() map[string]interface{} {
-	return map[string]interface{}{
-		"worker_id":      mws.workerID,
-		"address":        mws.address,
-		"running":        mws.running,
-		"capabilities":   mws.capabilities,
-		"max_concurrent": mws.maxConcurrent,
-		"current_tasks":  len(mws.currentTasks),
-		"task_details":   mws.currentTasks,
-	}
-}
-
-// SetQueue sets the maintenance queue for the worker
-func (mws *MaintenanceWorkerService) SetQueue(queue *MaintenanceQueue) {
-	mws.queue = queue
-}
-
-// SetAdminClient sets the admin client for the worker
-func (mws *MaintenanceWorkerService) SetAdminClient(client AdminClient) {
-	mws.adminClient = client
-}
-
-// SetCapabilities sets the worker capabilities
-func (mws *MaintenanceWorkerService) SetCapabilities(capabilities []MaintenanceTaskType) {
-	mws.capabilities = capabilities
-}
-
-// SetMaxConcurrent sets the maximum concurrent tasks
-func (mws *MaintenanceWorkerService) SetMaxConcurrent(max int) {
-	mws.maxConcurrent = max
-}
-
-// SetHeartbeatInterval sets the heartbeat interval (placeholder for future use)
-func (mws *MaintenanceWorkerService) SetHeartbeatInterval(interval time.Duration) {
-	// Future implementation for configurable heartbeat
-}
-
-// SetTaskRequestInterval sets the task request interval (placeholder for future use)
-func (mws *MaintenanceWorkerService) SetTaskRequestInterval(interval time.Duration) {
-	// Future implementation for configurable task requests
-}
-
-// MaintenanceWorkerCommand represents a standalone maintenance worker command
-type MaintenanceWorkerCommand struct {
-	workerService *MaintenanceWorkerService
-}
-
-// NewMaintenanceWorkerCommand creates a new worker command
-func NewMaintenanceWorkerCommand(workerID, address, adminServer string) *MaintenanceWorkerCommand {
-	return &MaintenanceWorkerCommand{
-		workerService: NewMaintenanceWorkerService(workerID, address, adminServer),
-	}
-}
-
-// Run starts the maintenance worker as a standalone service
-func (mwc *MaintenanceWorkerCommand) Run() error {
-	// Generate or load persistent worker ID if not provided
-	if mwc.workerService.workerID == "" {
-		// Get current working directory for worker ID persistence
-		wd, err := os.Getwd()
-		if err != nil {
-			return fmt.Errorf("failed to get working directory: %w", err)
-		}
-
-		workerID, err := worker.GenerateOrLoadWorkerID(wd)
-		if err != nil {
-			return fmt.Errorf("failed to generate or load worker ID: %w", err)
-		}
-		mwc.workerService.workerID = workerID
-	}
-
-	// Start the worker service
-	err := mwc.workerService.Start()
-	if err != nil {
-		return fmt.Errorf("failed to start maintenance worker: %w", err)
-	}
-
-	// Wait for interrupt signal
-	select {}
-}
@@ -122,6 +122,7 @@ type Plugin struct {
 type streamSession struct {
 	workerID  string
 	outgoing  chan *plugin_pb.AdminToWorkerMessage
+	done      chan struct{}
 	closeOnce sync.Once
 }

@@ -274,6 +275,7 @@ func (r *Plugin) WorkerStream(stream plugin_pb.PluginControlService_WorkerStream
 	session := &streamSession{
 		workerID: workerID,
 		outgoing: make(chan *plugin_pb.AdminToWorkerMessage, r.outgoingBuffer),
+		done:     make(chan struct{}),
 	}
 	r.putSession(session)
 	defer r.cleanupSession(workerID)
@@ -908,8 +910,10 @@ func (r *Plugin) sendLoop(
 			return nil
 		case <-r.shutdownCh:
 			return nil
-		case msg, ok := <-session.outgoing:
-			if !ok {
+		case <-session.done:
+			return nil
+		case msg := <-session.outgoing:
+			if msg == nil {
 				return nil
 			}
 			if err := stream.Send(msg); err != nil {
@@ -930,6 +934,8 @@ func (r *Plugin) sendToWorker(workerID string, message *plugin_pb.AdminToWorkerM
 	select {
 	case <-r.shutdownCh:
 		return fmt.Errorf("plugin is shutting down")
+	case <-session.done:
+		return fmt.Errorf("worker %s session is closed", workerID)
 	case session.outgoing <- message:
 		return nil
 	case <-time.After(r.sendTimeout):
@@ -1425,7 +1431,7 @@ func CloneConfigValueMap(in map[string]*plugin_pb.ConfigValue) map[string]*plugi

 func (s *streamSession) close() {
 	s.closeOnce.Do(func() {
-		close(s.outgoing)
+		close(s.done)
 	})
 }

@@ -26,7 +26,7 @@ func TestRunDetectionSendsCancelOnContextDone(t *testing.T) {
 			{JobType: jobType, CanDetect: true, MaxDetectionConcurrency: 1},
 		},
 	})
-	session := &streamSession{workerID: workerID, outgoing: make(chan *plugin_pb.AdminToWorkerMessage, 4)}
+	session := &streamSession{workerID: workerID, outgoing: make(chan *plugin_pb.AdminToWorkerMessage, 4), done: make(chan struct{})}
 	pluginSvc.putSession(session)

 	ctx, cancel := context.WithCancel(context.Background())
@@ -77,7 +77,7 @@ func TestExecuteJobSendsCancelOnContextDone(t *testing.T) {
 			{JobType: jobType, CanExecute: true, MaxExecutionConcurrency: 1},
 		},
 	})
-	session := &streamSession{workerID: workerID, outgoing: make(chan *plugin_pb.AdminToWorkerMessage, 4)}
+	session := &streamSession{workerID: workerID, outgoing: make(chan *plugin_pb.AdminToWorkerMessage, 4), done: make(chan struct{})}
 	pluginSvc.putSession(session)

 	job := &plugin_pb.JobSpec{JobId: "job-1", JobType: jobType}
@@ -135,8 +135,8 @@ func TestAdminScriptExecutionBlocksOtherDetection(t *testing.T) {
 			{JobType: "vacuum", CanDetect: true, MaxDetectionConcurrency: 1},
 		},
 	})
-	adminSession := &streamSession{workerID: adminWorkerID, outgoing: make(chan *plugin_pb.AdminToWorkerMessage, 8)}
-	otherSession := &streamSession{workerID: otherWorkerID, outgoing: make(chan *plugin_pb.AdminToWorkerMessage, 8)}
+	adminSession := &streamSession{workerID: adminWorkerID, outgoing: make(chan *plugin_pb.AdminToWorkerMessage, 8), done: make(chan struct{})}
+	otherSession := &streamSession{workerID: otherWorkerID, outgoing: make(chan *plugin_pb.AdminToWorkerMessage, 8), done: make(chan struct{})}
 	pluginSvc.putSession(adminSession)
 	pluginSvc.putSession(otherSession)

@@ -214,8 +214,8 @@ func TestAdminScriptExecutionBlocksOtherExecution(t *testing.T) {
 			{JobType: "vacuum", CanExecute: true, MaxExecutionConcurrency: 1},
 		},
 	})
-	adminSession := &streamSession{workerID: adminWorkerID, outgoing: make(chan *plugin_pb.AdminToWorkerMessage, 8)}
-	otherSession := &streamSession{workerID: otherWorkerID, outgoing: make(chan *plugin_pb.AdminToWorkerMessage, 8)}
+	adminSession := &streamSession{workerID: adminWorkerID, outgoing: make(chan *plugin_pb.AdminToWorkerMessage, 8), done: make(chan struct{})}
+	otherSession := &streamSession{workerID: otherWorkerID, outgoing: make(chan *plugin_pb.AdminToWorkerMessage, 8), done: make(chan struct{})}
 	pluginSvc.putSession(adminSession)
 	pluginSvc.putSession(otherSession)

@@ -22,7 +22,7 @@ func TestRunDetectionIncludesLatestSuccessfulRun(t *testing.T) {
 			{JobType: jobType, CanDetect: true, MaxDetectionConcurrency: 1},
 		},
 	})
-	session := &streamSession{workerID: "worker-a", outgoing: make(chan *plugin_pb.AdminToWorkerMessage, 1)}
+	session := &streamSession{workerID: "worker-a", outgoing: make(chan *plugin_pb.AdminToWorkerMessage, 1), done: make(chan struct{})}
 	pluginSvc.putSession(session)

 	oldSuccess := time.Date(2026, 1, 1, 0, 0, 0, 0, time.UTC)
@@ -80,7 +80,7 @@ func TestRunDetectionOmitsLastSuccessfulRunWhenNoSuccessHistory(t *testing.T) {
 			{JobType: jobType, CanDetect: true, MaxDetectionConcurrency: 1},
 		},
 	})
-	session := &streamSession{workerID: "worker-a", outgoing: make(chan *plugin_pb.AdminToWorkerMessage, 1)}
+	session := &streamSession{workerID: "worker-a", outgoing: make(chan *plugin_pb.AdminToWorkerMessage, 1), done: make(chan struct{})}
 	pluginSvc.putSession(session)

 	if err := pluginSvc.store.AppendRunRecord(jobType, &JobRunRecord{
@@ -130,7 +130,7 @@ func TestRunDetectionWithReportCapturesDetectionActivities(t *testing.T) {
 			{JobType: jobType, CanDetect: true, MaxDetectionConcurrency: 1},
 		},
 	})
-	session := &streamSession{workerID: "worker-a", outgoing: make(chan *plugin_pb.AdminToWorkerMessage, 1)}
+	session := &streamSession{workerID: "worker-a", outgoing: make(chan *plugin_pb.AdminToWorkerMessage, 1), done: make(chan struct{})}
 	pluginSvc.putSession(session)

 	reportCh := make(chan *DetectionReport, 1)
@@ -210,7 +210,7 @@ func TestRunDetectionAdminScriptUsesLastCompletedRun(t *testing.T) {
 			{JobType: jobType, CanDetect: true, MaxDetectionConcurrency: 1},
 		},
 	})
-	session := &streamSession{workerID: "worker-admin-script", outgoing: make(chan *plugin_pb.AdminToWorkerMessage, 1)}
+	session := &streamSession{workerID: "worker-admin-script", outgoing: make(chan *plugin_pb.AdminToWorkerMessage, 1), done: make(chan struct{})}
 	pluginSvc.putSession(session)

 	successCompleted := time.Date(2026, 2, 1, 10, 0, 0, 0, time.UTC)
@@ -95,16 +95,6 @@ func (r *Plugin) laneSchedulerLoop(ls *schedulerLaneState) {
 	}
 }

-// schedulerLoop is kept for backward compatibility; it delegates to
-// laneSchedulerLoop with the default lane. New code should not call this.
-func (r *Plugin) schedulerLoop() {
-	ls := r.lanes[LaneDefault]
-	if ls == nil {
-		ls = newLaneState(LaneDefault)
-	}
-	r.laneSchedulerLoop(ls)
-}
-
 // runLaneSchedulerIteration runs one scheduling pass for a single lane,
 // processing only the job types assigned to that lane.
 //
@@ -229,82 +219,6 @@ func (r *Plugin) runLaneSchedulerIterationConcurrent(ls *schedulerLaneState, job
 	return hadJobs.Load()
 }

-// runSchedulerIteration is kept for backward compatibility. It runs a
-// single iteration across ALL job types (equivalent to the old single-loop
-// behavior). It is only used by the legacy schedulerLoop() fallback.
-func (r *Plugin) runSchedulerIteration() bool {
-	ls := r.lanes[LaneDefault]
-	if ls == nil {
-		ls = newLaneState(LaneDefault)
-	}
-	// For backward compat, the old function processes all job types.
-	r.expireStaleJobs(time.Now().UTC())
-
-	jobTypes := r.registry.DetectableJobTypes()
-	if len(jobTypes) == 0 {
-		r.setSchedulerLoopState("", "idle")
-		return false
-	}
-
-	r.setSchedulerLoopState("", "waiting_for_lock")
-	releaseLock, err := r.acquireAdminLock("plugin scheduler iteration")
-	if err != nil {
-		glog.Warningf("Plugin scheduler failed to acquire lock: %v", err)
-		r.setSchedulerLoopState("", "idle")
-		return false
-	}
-	if releaseLock != nil {
-		defer releaseLock()
-	}
-
-	active := make(map[string]struct{}, len(jobTypes))
-	hadJobs := false
-
-	for _, jobType := range jobTypes {
-		active[jobType] = struct{}{}
-
-		policy, enabled, err := r.loadSchedulerPolicy(jobType)
-		if err != nil {
-			glog.Warningf("Plugin scheduler failed to load policy for %s: %v", jobType, err)
-			continue
-		}
-		if !enabled {
-			r.clearSchedulerJobType(jobType)
-			continue
-		}
-		initialDelay := time.Duration(0)
-		if runInfo := r.snapshotSchedulerRun(jobType); runInfo.lastRunStartedAt.IsZero() {
-			initialDelay = 5 * time.Second
-		}
-		if !r.markDetectionDue(jobType, policy.DetectionInterval, initialDelay) {
-			continue
-		}
-
-		detected := r.runJobTypeIteration(jobType, policy)
-		if detected {
-			hadJobs = true
-		}
-	}
-
-	r.pruneSchedulerState(active)
-	r.pruneDetectorLeases(active)
-	r.setSchedulerLoopState("", "idle")
-	return hadJobs
-}
-
-// wakeLane wakes the scheduler goroutine for a specific lane.
-func (r *Plugin) wakeLane(lane SchedulerLane) {
-	if r == nil {
-		return
-	}
-	if ls, ok := r.lanes[lane]; ok {
-		select {
-		case ls.wakeCh <- struct{}{}:
-		default:
-		}
-	}
-}
-
 // wakeAllLanes wakes all lane scheduler goroutines.
 func (r *Plugin) wakeAllLanes() {
 	if r == nil {
@@ -210,16 +210,6 @@ func (r *Plugin) setSchedulerLoopStateForJobType(jobType, phase string) {
 	}
 }

-func (r *Plugin) recordSchedulerIterationComplete(hadJobs bool) {
-	if r == nil {
-		return
-	}
-	r.schedulerLoopMu.Lock()
-	r.schedulerLoopState.lastIterationHadJobs = hadJobs
-	r.schedulerLoopState.lastIterationCompleted = time.Now().UTC()
-	r.schedulerLoopMu.Unlock()
-}
-
 func (r *Plugin) snapshotSchedulerLoopState() schedulerLoopState {
 	if r == nil {
 		return schedulerLoopState{}
@@ -125,6 +125,9 @@ templ ObjectStoreUsers(data dash.ObjectStoreUsersData) {
                                                <div class="d-flex align-items-center">
                                                    <i class="fas fa-user me-2 text-muted"></i>
                                                    <strong>{user.Username}</strong>
+                                                    if user.IsStatic {
+                                                        <span class="badge bg-secondary ms-2" title="Loaded from config file (read-only)">static</span>
+                                                    }
                                                </div>
                                            </td>
                                            <td>{user.Email}</td>
@@ -133,24 +136,28 @@ templ ObjectStoreUsers(data dash.ObjectStoreUsersData) {
                                            </td>
                                            <td>
                                                <div class="btn-group btn-group-sm" role="group">
-                                                    <button type="button" class="btn btn-outline-info" 
+                                                    <button type="button" class="btn btn-outline-info"
                                                            data-action="show-user-details" data-username={ user.Username }>
                                                        <i class="fas fa-info-circle"></i>
                                                    </button>
-                                                    <button type="button" class="btn btn-outline-primary" 
-                                                            data-action="edit-user" data-username={ user.Username }>
-                                                        <i class="fas fa-edit"></i>
-                                                    </button>
-                                                    if user.Username != "anonymous" {
+                                                    if !user.IsStatic {
+                                                        <button type="button" class="btn btn-outline-primary"
+                                                                data-action="edit-user" data-username={ user.Username }>
+                                                            <i class="fas fa-edit"></i>
+                                                        </button>
+                                                    }
+                                                    if user.Username != "anonymous" && !user.IsStatic {
                                                        <button type="button" class="btn btn-outline-secondary"
                                                                data-action="manage-access-keys" data-username={ user.Username }>
                                                            <i class="fas fa-key"></i>
                                                        </button>
                                                    }
-                                                    <button type="button" class="btn btn-outline-danger" 
-                                                            data-action="delete-user" data-username={ user.Username }>
-                                                        <i class="fas fa-trash"></i>
-                                                    </button>
+                                                    if !user.IsStatic {
+                                                        <button type="button" class="btn btn-outline-danger"
+                                                                data-action="delete-user" data-username={ user.Username }>
+                                                            <i class="fas fa-trash"></i>
+                                                        </button>
+                                                    }
                                                </div>
                                            </td>
                                        </tr>
@@ -163,7 +163,7 @@ templ S3Buckets(data dash.S3BucketsData) {
                                    for _, bucket := range data.Buckets {
                                        <tr>
                                            <td>
-                                                <a href={templ.SafeURL(fmt.Sprintf("/files?path=/buckets/%s", bucket.Name))}
+                                                <a href={dash.PUrl(ctx, fmt.Sprintf("/files?path=/buckets/%s", bucket.Name))}
                                                   class="text-decoration-none">
                                                    <i class="fas fa-cube me-2"></i>
                                                    {bucket.Name}
@@ -236,7 +236,7 @@ templ S3Buckets(data dash.S3BucketsData) {
                                            </td>
                                            <td>
                                                <div class="btn-group btn-group-sm" role="group">
-                                                    <a href={templ.SafeURL(fmt.Sprintf("/files?path=/buckets/%s", bucket.Name))}
+                                                    <a href={dash.PUrl(ctx, fmt.Sprintf("/files?path=/buckets/%s", bucket.Name))}
                                                       class="btn btn-outline-success btn-sm"
                                                       title="Browse Files">
                                                        <i class="fas fa-folder-open"></i>
@@ -171,9 +171,9 @@ func S3Buckets(data dash.S3BucketsData) templ.Component {
 				return templ_7745c5c3_Err
 			}
 			var templ_7745c5c3_Var5 templ.SafeURL
-			templ_7745c5c3_Var5, templ_7745c5c3_Err = templ.JoinURLErrs(templ.SafeURL(fmt.Sprintf("/files?path=/buckets/%s", bucket.Name)))
+			templ_7745c5c3_Var5, templ_7745c5c3_Err = templ.JoinURLErrs(dash.PUrl(ctx, fmt.Sprintf("/files?path=/buckets/%s", bucket.Name)))
 			if templ_7745c5c3_Err != nil {
-				return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3_buckets.templ`, Line: 166, Col: 123}
+				return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3_buckets.templ`, Line: 166, Col: 124}
 			}
 			_, templ_7745c5c3_Err = templ_7745c5c3_Buffer.WriteString(templ.EscapeString(templ_7745c5c3_Var5))
 			if templ_7745c5c3_Err != nil {
@@ -439,9 +439,9 @@ func S3Buckets(data dash.S3BucketsData) templ.Component {
 				return templ_7745c5c3_Err
 			}
 			var templ_7745c5c3_Var19 templ.SafeURL
-			templ_7745c5c3_Var19, templ_7745c5c3_Err = templ.JoinURLErrs(templ.SafeURL(fmt.Sprintf("/files?path=/buckets/%s", bucket.Name)))
+			templ_7745c5c3_Var19, templ_7745c5c3_Err = templ.JoinURLErrs(dash.PUrl(ctx, fmt.Sprintf("/files?path=/buckets/%s", bucket.Name)))
 			if templ_7745c5c3_Err != nil {
-				return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3_buckets.templ`, Line: 239, Col: 127}
+				return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3_buckets.templ`, Line: 239, Col: 128}
 			}
 			_, templ_7745c5c3_Err = templ_7745c5c3_Buffer.WriteString(templ.EscapeString(templ_7745c5c3_Var19))
 			if templ_7745c5c3_Err != nil {
@@ -2,6 +2,7 @@ package app

 import (
 	"fmt"
+	"net/url"

 	"github.com/seaweedfs/seaweedfs/weed/admin/dash"
 	"github.com/seaweedfs/seaweedfs/weed/s3api/s3tables"
@@ -152,7 +153,7 @@ templ S3TablesBuckets(data dash.S3TablesBucketsData) {
 												<div class="btn-group btn-group-sm" role="group">
 													{{ bucketName, parseErr := s3tables.ParseBucketNameFromARN(bucket.ARN) }}
 													if parseErr == nil {
-														<a class="btn btn-outline-primary btn-sm" href={ templ.SafeURL(fmt.Sprintf("/object-store/s3tables/buckets/%s/namespaces", bucketName)) }>
+														<a class="btn btn-outline-primary btn-sm" href={ dash.PUrl(ctx, fmt.Sprintf("/object-store/s3tables/buckets/%s/namespaces", url.PathEscape(bucketName))) }>
 															<i class="fas fa-folder-open"></i>
 														</a>
 													} else {
@@ -10,6 +10,7 @@ import templruntime "github.com/a-h/templ/runtime"

 import (
 	"fmt"
+	"net/url"

 	"github.com/seaweedfs/seaweedfs/weed/admin/dash"
 	"github.com/seaweedfs/seaweedfs/weed/s3api/s3tables"
@@ -48,7 +49,7 @@ func S3TablesBuckets(data dash.S3TablesBucketsData) templ.Component {
 			var templ_7745c5c3_Var2 templ.SafeURL
 			templ_7745c5c3_Var2, templ_7745c5c3_Err = templ.JoinURLErrs(templ.SafeURL(fmt.Sprintf("http://localhost:%d/v1/config", data.IcebergPort)))
 			if templ_7745c5c3_Err != nil {
-				return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3tables_buckets.templ`, Line: 23, Col: 124}
+				return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3tables_buckets.templ`, Line: 24, Col: 124}
 			}
 			_, templ_7745c5c3_Err = templ_7745c5c3_Buffer.WriteString(templ.EscapeString(templ_7745c5c3_Var2))
 			if templ_7745c5c3_Err != nil {
@@ -66,7 +67,7 @@ func S3TablesBuckets(data dash.S3TablesBucketsData) templ.Component {
 		var templ_7745c5c3_Var3 string
 		templ_7745c5c3_Var3, templ_7745c5c3_Err = templ.JoinStringErrs(fmt.Sprintf("%d", data.IcebergPort))
 		if templ_7745c5c3_Err != nil {
-			return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3tables_buckets.templ`, Line: 30, Col: 91}
+			return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3tables_buckets.templ`, Line: 31, Col: 91}
 		}
 		_, templ_7745c5c3_Err = templ_7745c5c3_Buffer.WriteString(templ.EscapeString(templ_7745c5c3_Var3))
 		if templ_7745c5c3_Err != nil {
@@ -84,7 +85,7 @@ func S3TablesBuckets(data dash.S3TablesBucketsData) templ.Component {
 			var templ_7745c5c3_Var4 string
 			templ_7745c5c3_Var4, templ_7745c5c3_Err = templ.JoinStringErrs(fmt.Sprintf("%d", data.IcebergPort))
 			if templ_7745c5c3_Err != nil {
-				return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3tables_buckets.templ`, Line: 37, Col: 109}
+				return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3tables_buckets.templ`, Line: 38, Col: 109}
 			}
 			_, templ_7745c5c3_Err = templ_7745c5c3_Buffer.WriteString(templ.EscapeString(templ_7745c5c3_Var4))
 			if templ_7745c5c3_Err != nil {
@@ -97,7 +98,7 @@ func S3TablesBuckets(data dash.S3TablesBucketsData) templ.Component {
 			var templ_7745c5c3_Var5 string
 			templ_7745c5c3_Var5, templ_7745c5c3_Err = templ.JoinStringErrs(fmt.Sprintf("%d", data.IcebergPort))
 			if templ_7745c5c3_Err != nil {
-				return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3tables_buckets.templ`, Line: 39, Col: 107}
+				return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3tables_buckets.templ`, Line: 40, Col: 107}
 			}
 			_, templ_7745c5c3_Err = templ_7745c5c3_Buffer.WriteString(templ.EscapeString(templ_7745c5c3_Var5))
 			if templ_7745c5c3_Err != nil {
@@ -120,7 +121,7 @@ func S3TablesBuckets(data dash.S3TablesBucketsData) templ.Component {
 		var templ_7745c5c3_Var6 string
 		templ_7745c5c3_Var6, templ_7745c5c3_Err = templ.JoinStringErrs(fmt.Sprintf("%d", data.TotalBuckets))
 		if templ_7745c5c3_Err != nil {
-			return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3tables_buckets.templ`, Line: 69, Col: 47}
+			return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3tables_buckets.templ`, Line: 70, Col: 47}
 		}
 		_, templ_7745c5c3_Err = templ_7745c5c3_Buffer.WriteString(templ.EscapeString(templ_7745c5c3_Var6))
 		if templ_7745c5c3_Err != nil {
@@ -133,7 +134,7 @@ func S3TablesBuckets(data dash.S3TablesBucketsData) templ.Component {
 		var templ_7745c5c3_Var7 string
 		templ_7745c5c3_Var7, templ_7745c5c3_Err = templ.JoinStringErrs(data.LastUpdated.Format("2006-01-02 15:04"))
 		if templ_7745c5c3_Err != nil {
-			return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3tables_buckets.templ`, Line: 88, Col: 54}
+			return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3tables_buckets.templ`, Line: 89, Col: 54}
 		}
 		_, templ_7745c5c3_Err = templ_7745c5c3_Buffer.WriteString(templ.EscapeString(templ_7745c5c3_Var7))
 		if templ_7745c5c3_Err != nil {
@@ -147,7 +148,7 @@ func S3TablesBuckets(data dash.S3TablesBucketsData) templ.Component {
 			var templ_7745c5c3_Var8 string
 			templ_7745c5c3_Var8, templ_7745c5c3_Err = templ.JoinStringErrs(fmt.Sprintf("%d", data.IcebergPort))
 			if templ_7745c5c3_Err != nil {
-				return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3tables_buckets.templ`, Line: 108, Col: 47}
+				return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3tables_buckets.templ`, Line: 109, Col: 47}
 			}
 			_, templ_7745c5c3_Err = templ_7745c5c3_Buffer.WriteString(templ.EscapeString(templ_7745c5c3_Var8))
 			if templ_7745c5c3_Err != nil {
@@ -171,7 +172,7 @@ func S3TablesBuckets(data dash.S3TablesBucketsData) templ.Component {
 			var templ_7745c5c3_Var9 string
 			templ_7745c5c3_Var9, templ_7745c5c3_Err = templ.JoinStringErrs(bucket.Name)
 			if templ_7745c5c3_Err != nil {
-				return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3tables_buckets.templ`, Line: 146, Col: 28}
+				return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3tables_buckets.templ`, Line: 147, Col: 28}
 			}
 			_, templ_7745c5c3_Err = templ_7745c5c3_Buffer.WriteString(templ.EscapeString(templ_7745c5c3_Var9))
 			if templ_7745c5c3_Err != nil {
@@ -184,7 +185,7 @@ func S3TablesBuckets(data dash.S3TablesBucketsData) templ.Component {
 			var templ_7745c5c3_Var10 string
 			templ_7745c5c3_Var10, templ_7745c5c3_Err = templ.JoinStringErrs(bucket.OwnerAccountID)
 			if templ_7745c5c3_Err != nil {
-				return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3tables_buckets.templ`, Line: 147, Col: 38}
+				return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3tables_buckets.templ`, Line: 148, Col: 38}
 			}
 			_, templ_7745c5c3_Err = templ_7745c5c3_Buffer.WriteString(templ.EscapeString(templ_7745c5c3_Var10))
 			if templ_7745c5c3_Err != nil {
@@ -197,7 +198,7 @@ func S3TablesBuckets(data dash.S3TablesBucketsData) templ.Component {
 			var templ_7745c5c3_Var11 string
 			templ_7745c5c3_Var11, templ_7745c5c3_Err = templ.JoinStringErrs(bucket.ARN)
 			if templ_7745c5c3_Err != nil {
-				return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3tables_buckets.templ`, Line: 148, Col: 52}
+				return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3tables_buckets.templ`, Line: 149, Col: 52}
 			}
 			_, templ_7745c5c3_Err = templ_7745c5c3_Buffer.WriteString(templ.EscapeString(templ_7745c5c3_Var11))
 			if templ_7745c5c3_Err != nil {
@@ -210,7 +211,7 @@ func S3TablesBuckets(data dash.S3TablesBucketsData) templ.Component {
 			var templ_7745c5c3_Var12 string
 			templ_7745c5c3_Var12, templ_7745c5c3_Err = templ.JoinStringErrs(bucket.Name)
 			if templ_7745c5c3_Err != nil {
-				return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3tables_buckets.templ`, Line: 149, Col: 52}
+				return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3tables_buckets.templ`, Line: 150, Col: 52}
 			}
 			_, templ_7745c5c3_Err = templ_7745c5c3_Buffer.WriteString(templ.EscapeString(templ_7745c5c3_Var12))
 			if templ_7745c5c3_Err != nil {
@@ -223,7 +224,7 @@ func S3TablesBuckets(data dash.S3TablesBucketsData) templ.Component {
 			var templ_7745c5c3_Var13 string
 			templ_7745c5c3_Var13, templ_7745c5c3_Err = templ.JoinStringErrs(bucket.CreatedAt.Format("2006-01-02 15:04"))
 			if templ_7745c5c3_Err != nil {
-				return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3tables_buckets.templ`, Line: 150, Col: 60}
+				return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3tables_buckets.templ`, Line: 151, Col: 60}
 			}
 			_, templ_7745c5c3_Err = templ_7745c5c3_Buffer.WriteString(templ.EscapeString(templ_7745c5c3_Var13))
 			if templ_7745c5c3_Err != nil {
@@ -240,9 +241,9 @@ func S3TablesBuckets(data dash.S3TablesBucketsData) templ.Component {
 					return templ_7745c5c3_Err
 				}
 				var templ_7745c5c3_Var14 templ.SafeURL
-				templ_7745c5c3_Var14, templ_7745c5c3_Err = templ.JoinURLErrs(templ.SafeURL(fmt.Sprintf("/object-store/s3tables/buckets/%s/namespaces", bucketName)))
+				templ_7745c5c3_Var14, templ_7745c5c3_Err = templ.JoinURLErrs(dash.PUrl(ctx, fmt.Sprintf("/object-store/s3tables/buckets/%s/namespaces", url.PathEscape(bucketName))))
 				if templ_7745c5c3_Err != nil {
-					return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3tables_buckets.templ`, Line: 155, Col: 149}
+					return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3tables_buckets.templ`, Line: 156, Col: 166}
 				}
 				_, templ_7745c5c3_Err = templ_7745c5c3_Buffer.WriteString(templ.EscapeString(templ_7745c5c3_Var14))
 				if templ_7745c5c3_Err != nil {
@@ -265,7 +266,7 @@ func S3TablesBuckets(data dash.S3TablesBucketsData) templ.Component {
 			var templ_7745c5c3_Var15 string
 			templ_7745c5c3_Var15, templ_7745c5c3_Err = templ.JoinStringErrs(bucket.ARN)
 			if templ_7745c5c3_Err != nil {
-				return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3tables_buckets.templ`, Line: 163, Col: 122}
+				return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3tables_buckets.templ`, Line: 164, Col: 122}
 			}
 			_, templ_7745c5c3_Err = templ_7745c5c3_Buffer.WriteString(templ.EscapeString(templ_7745c5c3_Var15))
 			if templ_7745c5c3_Err != nil {
@@ -278,7 +279,7 @@ func S3TablesBuckets(data dash.S3TablesBucketsData) templ.Component {
 			var templ_7745c5c3_Var16 string
 			templ_7745c5c3_Var16, templ_7745c5c3_Err = templ.JoinStringErrs(bucket.ARN)
 			if templ_7745c5c3_Err != nil {
-				return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3tables_buckets.templ`, Line: 166, Col: 126}
+				return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3tables_buckets.templ`, Line: 167, Col: 126}
 			}
 			_, templ_7745c5c3_Err = templ_7745c5c3_Buffer.WriteString(templ.EscapeString(templ_7745c5c3_Var16))
 			if templ_7745c5c3_Err != nil {
@@ -291,7 +292,7 @@ func S3TablesBuckets(data dash.S3TablesBucketsData) templ.Component {
 			var templ_7745c5c3_Var17 string
 			templ_7745c5c3_Var17, templ_7745c5c3_Err = templ.JoinStringErrs(bucket.ARN)
 			if templ_7745c5c3_Err != nil {
-				return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3tables_buckets.templ`, Line: 169, Col: 128}
+				return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3tables_buckets.templ`, Line: 170, Col: 128}
 			}
 			_, templ_7745c5c3_Err = templ_7745c5c3_Buffer.WriteString(templ.EscapeString(templ_7745c5c3_Var17))
 			if templ_7745c5c3_Err != nil {
@@ -304,7 +305,7 @@ func S3TablesBuckets(data dash.S3TablesBucketsData) templ.Component {
 			var templ_7745c5c3_Var18 string
 			templ_7745c5c3_Var18, templ_7745c5c3_Err = templ.JoinStringErrs(bucket.Name)
 			if templ_7745c5c3_Err != nil {
-				return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3tables_buckets.templ`, Line: 169, Col: 161}
+				return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3tables_buckets.templ`, Line: 170, Col: 161}
 			}
 			_, templ_7745c5c3_Err = templ_7745c5c3_Buffer.WriteString(templ.EscapeString(templ_7745c5c3_Var18))
 			if templ_7745c5c3_Err != nil {
@@ -342,7 +343,7 @@ CREATE SECRET (

 SELECT * FROM iceberg_scan('s3://my-table-bucket/my-namespace/my-table');`)
 			if templ_7745c5c3_Err != nil {
-				return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3tables_buckets.templ`, Line: 219, Col: 74}
+				return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3tables_buckets.templ`, Line: 220, Col: 74}
 			}
 			_, templ_7745c5c3_Err = templ_7745c5c3_Buffer.WriteString(templ.EscapeString(templ_7745c5c3_Var19))
 			if templ_7745c5c3_Err != nil {
@@ -365,7 +366,7 @@ catalog = load_catalog(

 namespaces = catalog.list_namespaces()`)
 			if templ_7745c5c3_Err != nil {
-				return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3tables_buckets.templ`, Line: 235, Col: 39}
+				return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3tables_buckets.templ`, Line: 236, Col: 39}
 			}
 			_, templ_7745c5c3_Err = templ_7745c5c3_Buffer.WriteString(templ.EscapeString(templ_7745c5c3_Var20))
 			if templ_7745c5c3_Err != nil {
@@ -24,7 +24,7 @@ templ S3TablesTables(data dash.S3TablesTablesData) {
 	<div class="mb-3">
 		{{ bucketName, parseErr := s3tables.ParseBucketNameFromARN(data.BucketARN) }}
 		if parseErr == nil {
-			<a href={ templ.SafeURL(fmt.Sprintf("/object-store/s3tables/buckets/%s/namespaces", bucketName)) } class="btn btn-sm btn-outline-secondary">
+			<a href={ dash.PUrl(ctx, fmt.Sprintf("/object-store/s3tables/buckets/%s/namespaces", url.PathEscape(bucketName))) } class="btn btn-sm btn-outline-secondary">
 				<i class="fas fa-arrow-left me-1"></i>Back to Namespaces
 			</a>
 		} else {
@@ -126,7 +126,7 @@ templ S3TablesTables(data dash.S3TablesTablesData) {
 											<td>
 												<div class="btn-group btn-group-sm" role="group">
 													if parseErr == nil {
-														<a class="btn btn-outline-primary btn-sm" href={ templ.SafeURL(fmt.Sprintf("/object-store/s3tables/buckets/%s/namespaces/%s/tables/%s", url.PathEscape(bucketName), url.PathEscape(data.Namespace), url.PathEscape(tableName))) } title="View Iceberg Details">
+														<a class="btn btn-outline-primary btn-sm" href={ dash.PUrl(ctx, fmt.Sprintf("/object-store/s3tables/buckets/%s/namespaces/%s/tables/%s", url.PathEscape(bucketName), url.PathEscape(data.Namespace), url.PathEscape(tableName))) } title="View Iceberg Details">
 															<i class="fas fa-eye"></i>
 														</a>
 													} else {
@@ -48,9 +48,9 @@ func S3TablesTables(data dash.S3TablesTablesData) templ.Component {
 				return templ_7745c5c3_Err
 			}
 			var templ_7745c5c3_Var2 templ.SafeURL
-			templ_7745c5c3_Var2, templ_7745c5c3_Err = templ.JoinURLErrs(templ.SafeURL(fmt.Sprintf("/object-store/s3tables/buckets/%s/namespaces", bucketName)))
+			templ_7745c5c3_Var2, templ_7745c5c3_Err = templ.JoinURLErrs(dash.PUrl(ctx, fmt.Sprintf("/object-store/s3tables/buckets/%s/namespaces", url.PathEscape(bucketName))))
 			if templ_7745c5c3_Err != nil {
-				return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3tables_tables.templ`, Line: 27, Col: 99}
+				return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3tables_tables.templ`, Line: 27, Col: 116}
 			}
 			_, templ_7745c5c3_Err = templ_7745c5c3_Buffer.WriteString(templ.EscapeString(templ_7745c5c3_Var2))
 			if templ_7745c5c3_Err != nil {
@@ -308,9 +308,9 @@ func S3TablesTables(data dash.S3TablesTablesData) templ.Component {
 					return templ_7745c5c3_Err
 				}
 				var templ_7745c5c3_Var17 templ.SafeURL
-				templ_7745c5c3_Var17, templ_7745c5c3_Err = templ.JoinURLErrs(templ.SafeURL(fmt.Sprintf("/object-store/s3tables/buckets/%s/namespaces/%s/tables/%s", url.PathEscape(bucketName), url.PathEscape(data.Namespace), url.PathEscape(tableName))))
+				templ_7745c5c3_Var17, templ_7745c5c3_Err = templ.JoinURLErrs(dash.PUrl(ctx, fmt.Sprintf("/object-store/s3tables/buckets/%s/namespaces/%s/tables/%s", url.PathEscape(bucketName), url.PathEscape(data.Namespace), url.PathEscape(tableName))))
 				if templ_7745c5c3_Err != nil {
-					return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3tables_tables.templ`, Line: 129, Col: 237}
+					return templ.Error{Err: templ_7745c5c3_Err, FileName: `view/app/s3tables_tables.templ`, Line: 129, Col: 238}
 				}
 				_, templ_7745c5c3_Err = templ_7745c5c3_Buffer.WriteString(templ.EscapeString(templ_7745c5c3_Var17))
 				if templ_7745c5c3_Err != nil {
@@ -6,20 +6,6 @@ import (
 	"strings"
 )

-// getStatusColor returns Bootstrap color class for status
-func getStatusColor(status string) string {
-	switch status {
-	case "active", "healthy":
-		return "success"
-	case "warning":
-		return "warning"
-	case "critical", "unreachable":
-		return "danger"
-	default:
-		return "secondary"
-	}
-}
-
 // formatBytes converts bytes to human readable format
 func formatBytes(bytes int64) string {
 	if bytes == 0 {
@@ -95,18 +95,6 @@ func NewCluster() *Cluster {
 	}
 }

-func (cluster *Cluster) getGroupMembers(filerGroup FilerGroupName, nodeType string, createIfNotFound bool) *GroupMembers {
-	switch nodeType {
-	case FilerType:
-		return cluster.filerGroups.getGroupMembers(filerGroup, createIfNotFound)
-	case BrokerType:
-		return cluster.brokerGroups.getGroupMembers(filerGroup, createIfNotFound)
-	case S3Type:
-		return cluster.s3Groups.getGroupMembers(filerGroup, createIfNotFound)
-	}
-	return nil
-}
-
 func (cluster *Cluster) AddClusterNode(ns, nodeType string, dataCenter DataCenter, rack Rack, address pb.ServerAddress, version string) []*master_pb.KeepConnectedResponse {
 	filerGroup := FilerGroupName(ns)
 	switch nodeType {
@@ -30,6 +30,7 @@ import (
 	"github.com/seaweedfs/seaweedfs/weed/pb"
 	"github.com/seaweedfs/seaweedfs/weed/security"
 	"github.com/seaweedfs/seaweedfs/weed/util"
+	"github.com/seaweedfs/seaweedfs/weed/util/grace"
 )

 var (
@@ -48,6 +49,10 @@ type AdminOptions struct {
 	dataDir          *string
 	icebergPort      *int
 	urlPrefix        *string
+	debug            *bool
+	debugPort        *int
+	cpuProfile       *string
+	memProfile       *string
 }

 func init() {
@@ -64,6 +69,10 @@ func init() {
 	a.readOnlyPassword = cmdAdmin.Flag.String("readOnlyPassword", "", "read-only user password (optional, for view-only access; requires adminPassword to be set)")
 	a.icebergPort = cmdAdmin.Flag.Int("iceberg.port", 8181, "Iceberg REST Catalog port (0 to hide in UI)")
 	a.urlPrefix = cmdAdmin.Flag.String("urlPrefix", "", "URL path prefix when running behind a reverse proxy under a subdirectory (e.g. /seaweedfs)")
+	a.debug = cmdAdmin.Flag.Bool("debug", false, "serves runtime profiling data via pprof on the port specified by -debug.port")
+	a.debugPort = cmdAdmin.Flag.Int("debug.port", 6060, "http port for debugging")
+	a.cpuProfile = cmdAdmin.Flag.String("cpuprofile", "", "cpu profile output file")
+	a.memProfile = cmdAdmin.Flag.String("memprofile", "", "memory profile output file")
 }

 var cmdAdmin = &Command{
@@ -140,6 +149,16 @@ var cmdAdmin = &Command{
    - All static assets, API endpoints, and navigation links will use the prefix
    - Session cookies are scoped to the prefix path

+  Debugging and Profiling:
+    - Use -debug to start a pprof HTTP server for live profiling (localhost only)
+    - Set -debug.port to choose the pprof port (default 6060)
+    - Profiles are accessible at http://127.0.0.1:<debug.port>/debug/pprof/
+    - Use -cpuprofile and -memprofile to write profiles to files on shutdown
+    - WARNING: -debug exposes runtime internals; use only in trusted environments
+    - Examples:
+      weed admin -debug -debug.port=6060 -master="localhost:9333"
+      weed admin -cpuprofile=cpu.prof -memprofile=mem.prof -master="localhost:9333"
+
  Configuration File:
    - The security.toml file is read from ".", "$HOME/.seaweedfs/",
      "/usr/local/etc/seaweedfs/", or "/etc/seaweedfs/", in that order
@@ -149,6 +168,12 @@ var cmdAdmin = &Command{
 }

 func runAdmin(cmd *Command, args []string) bool {
+	if *a.debug {
+		grace.StartDebugServer(*a.debugPort)
+	}
+
+	grace.SetupProfiling(*a.cpuProfile, *a.memProfile)
+
 	// Load security configuration
 	util.LoadSecurityConfiguration()

@@ -511,11 +536,6 @@ func recoveryMiddleware(next http.Handler) http.Handler {
 	})
 }

-// GetAdminOptions returns the admin command options for testing
-func GetAdminOptions() *AdminOptions {
-	return &AdminOptions{}
-}
-
 // loadOrGenerateSessionKeys loads or creates authentication/encryption keys for session cookies.
 func loadOrGenerateSessionKeys(dataDir string) ([]byte, []byte, error) {
 	const keyLen = 32
@@ -132,16 +132,3 @@ func fetchContent(masterFn operation.GetMasterFn, grpcDialOption grpc.DialOption
 	content, e = io.ReadAll(rc.Body)
 	return
 }
-
-func WriteFile(filename string, data []byte, perm os.FileMode) error {
-	f, err := os.OpenFile(filename, os.O_WRONLY|os.O_CREATE|os.O_TRUNC, perm)
-	if err != nil {
-		return err
-	}
-	n, err := f.Write(data)
-	f.Close()
-	if err == nil && n < len(data) {
-		err = io.ErrShortWrite
-	}
-	return err
-}
@@ -81,6 +81,7 @@ type FilerOptions struct {
 	exposeDirectoryData       *bool
 	tusBasePath               *string
 	certProvider              certprovider.Provider
+	s3ConfigFile              *string // optional path to static S3 identity config
 }

 func init() {
@@ -146,6 +147,7 @@ func init() {
 	filerS3Options.iamReadOnly = cmdFiler.Flag.Bool("s3.iam.readOnly", true, "disable IAM write operations on this server")
 	filerS3Options.portIceberg = cmdFiler.Flag.Int("s3.port.iceberg", 8181, "Iceberg REST Catalog server listen port (0 to disable)")
 	filerS3Options.externalUrl = cmdFiler.Flag.String("s3.externalUrl", "", "the external URL clients use to connect (e.g. https://api.example.com:9000). Used for S3 signature verification behind a reverse proxy. Falls back to S3_EXTERNAL_URL env var.")
+	filerS3Options.defaultFileMode = cmdFiler.Flag.String("s3.defaultFileMode", "", "default file mode for S3 uploaded objects, e.g. 0660, 0644, 0666")

 	// start webdav on filer
 	filerStartWebDav = cmdFiler.Flag.Bool("webdav", false, "whether to start webdav gateway")
@@ -342,6 +344,15 @@ func (fo *FilerOptions) startFiler() {
 		glog.V(0).Infof("Initialized credential manager: %s", credentialManager.GetStoreName())
 	}

+	// Load static S3 identities from config file if specified
+	if fo.s3ConfigFile != nil && *fo.s3ConfigFile != "" {
+		if credentialManager != nil {
+			if err := credentialManager.LoadS3ConfigFile(*fo.s3ConfigFile); err != nil {
+				glog.Warningf("Failed to load S3 config file for static identities: %v", err)
+			}
+		}
+	}
+
 	fs, nfs_err := weed_server.NewFilerServer(defaultMux, publicVolumeMux, &weed_server.FilerOption{
 		Masters:                   fo.masters,
 		FilerGroup:                *fo.filerGroup,
@@ -53,6 +53,8 @@ type SyncOptions struct {
 	chunkConcurrency *int
 	aDoDeleteFiles   *bool
 	bDoDeleteFiles   *bool
+	aSecurity        *string
+	bSecurity        *string
 	clientId         int32
 	clientEpoch      atomic.Int32
 	debug            *bool
@@ -113,6 +115,8 @@ func init() {
 	syncOptions.metricsHttpPort = cmdFilerSynchronize.Flag.Int("metricsPort", 0, "metrics listen port")
 	syncOptions.aDoDeleteFiles = cmdFilerSynchronize.Flag.Bool("a.doDeleteFiles", true, "delete and update files when synchronizing on filer A")
 	syncOptions.bDoDeleteFiles = cmdFilerSynchronize.Flag.Bool("b.doDeleteFiles", true, "delete and update files when synchronizing on filer B")
+	syncOptions.aSecurity = cmdFilerSynchronize.Flag.String("a.security", "", "security.toml file for filer A when clusters use different certificates")
+	syncOptions.bSecurity = cmdFilerSynchronize.Flag.String("b.security", "", "security.toml file for filer B when clusters use different certificates")
 	syncOptions.debug = cmdFilerSynchronize.Flag.Bool("debug", false, "serves runtime profiling data via pprof on the port specified by -debug.port")
 	syncOptions.debugPort = cmdFilerSynchronize.Flag.Int("debug.port", 6060, "http port for debugging")
 	syncOptions.clientId = util.RandomInt32()
@@ -144,6 +148,22 @@ func runFilerSynchronize(cmd *Command, args []string) bool {
 	util.LoadSecurityConfiguration()
 	grpcDialOption := security.LoadClientTLS(util.GetViper(), "grpc.client")

+	// per-filer TLS when clusters use different certificates
+	grpcDialOptionA := grpcDialOption
+	grpcDialOptionB := grpcDialOption
+	if *syncOptions.aSecurity != "" {
+		var err error
+		if grpcDialOptionA, err = security.LoadClientTLSFromFile(*syncOptions.aSecurity, "grpc.client"); err != nil {
+			glog.Fatalf("load security config for filer A: %v", err)
+		}
+	}
+	if *syncOptions.bSecurity != "" {
+		var err error
+		if grpcDialOptionB, err = security.LoadClientTLSFromFile(*syncOptions.bSecurity, "grpc.client"); err != nil {
+			glog.Fatalf("load security config for filer B: %v", err)
+		}
+	}
+
 	grace.SetupProfiling(*syncCpuProfile, *syncMemProfile)

 	filerA := pb.ServerAddress(*syncOptions.filerA)
@@ -153,13 +173,13 @@ func runFilerSynchronize(cmd *Command, args []string) bool {
 	go statsCollect.StartMetricsServer(*syncOptions.metricsHttpIp, *syncOptions.metricsHttpPort)

 	// read a filer signature
-	aFilerSignature, aFilerErr := replication.ReadFilerSignature(grpcDialOption, filerA)
+	aFilerSignature, aFilerErr := replication.ReadFilerSignature(grpcDialOptionA, filerA)
 	if aFilerErr != nil {
 		glog.Errorf("get filer 'a' signature %d error from %s to %s: %v", aFilerSignature, *syncOptions.filerA, *syncOptions.filerB, aFilerErr)
 		return true
 	}
 	// read b filer signature
-	bFilerSignature, bFilerErr := replication.ReadFilerSignature(grpcDialOption, filerB)
+	bFilerSignature, bFilerErr := replication.ReadFilerSignature(grpcDialOptionB, filerB)
 	if bFilerErr != nil {
 		glog.Errorf("get filer 'b' signature %d error from %s to %s: %v", bFilerSignature, *syncOptions.filerA, *syncOptions.filerB, bFilerErr)
 		return true
@@ -189,9 +209,9 @@ func runFilerSynchronize(cmd *Command, args []string) bool {
 	go func() {
 		// a->b
 		// set synchronization start timestamp to offset
-		initOffsetError := initOffsetFromTsMs(grpcDialOption, filerB, aFilerSignature, *syncOptions.bFromTsMs, getSignaturePrefixByPath(*syncOptions.aPath))
+		initOffsetError := initOffsetFromTsMs(grpcDialOptionB, filerB, aFilerSignature, *syncOptions.aFromTsMs, getSignaturePrefixByPath(*syncOptions.aPath))
 		if initOffsetError != nil {
-			glog.Errorf("init offset from timestamp %d error from %s to %s: %v", *syncOptions.bFromTsMs, *syncOptions.filerA, *syncOptions.filerB, initOffsetError)
+			glog.Errorf("init offset from timestamp %d error from %s to %s: %v", *syncOptions.aFromTsMs, *syncOptions.filerA, *syncOptions.filerB, initOffsetError)
 			os.Exit(2)
 		}
 		for {
@@ -199,11 +219,12 @@ func runFilerSynchronize(cmd *Command, args []string) bool {
 			err := doSubscribeFilerMetaChanges(
 				syncOptions.clientId,
 				syncOptions.clientEpoch.Load(),
-				grpcDialOption,
+				grpcDialOptionA,
 				filerA,
 				*syncOptions.aPath,
 				util.StringSplit(*syncOptions.aExcludePaths, ","),
 				*syncOptions.aProxyByFiler,
+				grpcDialOptionB,
 				filerB,
 				*syncOptions.bPath,
 				*syncOptions.bReplication,
@@ -228,9 +249,9 @@ func runFilerSynchronize(cmd *Command, args []string) bool {
 	if !*syncOptions.isActivePassive {
 		// b->a
 		// set synchronization start timestamp to offset
-		initOffsetError := initOffsetFromTsMs(grpcDialOption, filerA, bFilerSignature, *syncOptions.aFromTsMs, getSignaturePrefixByPath(*syncOptions.bPath))
+		initOffsetError := initOffsetFromTsMs(grpcDialOptionA, filerA, bFilerSignature, *syncOptions.bFromTsMs, getSignaturePrefixByPath(*syncOptions.bPath))
 		if initOffsetError != nil {
-			glog.Errorf("init offset from timestamp %d error from %s to %s: %v", *syncOptions.aFromTsMs, *syncOptions.filerB, *syncOptions.filerA, initOffsetError)
+			glog.Errorf("init offset from timestamp %d error from %s to %s: %v", *syncOptions.bFromTsMs, *syncOptions.filerB, *syncOptions.filerA, initOffsetError)
 			os.Exit(2)
 		}
 		go func() {
@@ -239,11 +260,12 @@ func runFilerSynchronize(cmd *Command, args []string) bool {
 				err := doSubscribeFilerMetaChanges(
 					syncOptions.clientId,
 					syncOptions.clientEpoch.Load(),
-					grpcDialOption,
+					grpcDialOptionB,
 					filerB,
 					*syncOptions.bPath,
 					util.StringSplit(*syncOptions.bExcludePaths, ","),
 					*syncOptions.bProxyByFiler,
+					grpcDialOptionA,
 					filerA,
 					*syncOptions.aPath,
 					*syncOptions.aReplication,
@@ -285,12 +307,12 @@ func initOffsetFromTsMs(grpcDialOption grpc.DialOption, targetFiler pb.ServerAdd
 	return nil
 }

-func doSubscribeFilerMetaChanges(clientId int32, clientEpoch int32, grpcDialOption grpc.DialOption, sourceFiler pb.ServerAddress, sourcePath string, sourceExcludePaths []string, sourceReadChunkFromFiler bool, targetFiler pb.ServerAddress, targetPath string,
+func doSubscribeFilerMetaChanges(clientId int32, clientEpoch int32, sourceGrpcDialOption grpc.DialOption, sourceFiler pb.ServerAddress, sourcePath string, sourceExcludePaths []string, sourceReadChunkFromFiler bool, targetGrpcDialOption grpc.DialOption, targetFiler pb.ServerAddress, targetPath string,
 	replicationStr, collection string, ttlSec int, sinkWriteChunkByFiler bool, diskType string, debug bool, concurrency int, chunkConcurrency int, doDeleteFiles bool, sourceFilerSignature int32, targetFilerSignature int32, statePtr *atomic.Pointer[syncState]) error {

 	// if first time, start from now
 	// if has previously synced, resume from that point of time
-	sourceFilerOffsetTsNs, err := getOffset(grpcDialOption, targetFiler, getSignaturePrefixByPath(sourcePath), sourceFilerSignature)
+	sourceFilerOffsetTsNs, err := getOffset(targetGrpcDialOption, targetFiler, getSignaturePrefixByPath(sourcePath), sourceFilerSignature)
 	if err != nil {
 		return err
 	}
@@ -300,8 +322,9 @@ func doSubscribeFilerMetaChanges(clientId int32, clientEpoch int32, grpcDialOpti
 	// create filer sink
 	filerSource := &source.FilerSource{}
 	filerSource.DoInitialize(sourceFiler.ToHttpAddress(), sourceFiler.ToGrpcAddress(), sourcePath, sourceReadChunkFromFiler)
+	filerSource.SetGrpcDialOption(sourceGrpcDialOption)
 	filerSink := &filersink.FilerSink{}
-	filerSink.DoInitialize(targetFiler.ToHttpAddress(), targetFiler.ToGrpcAddress(), targetPath, replicationStr, collection, ttlSec, diskType, grpcDialOption, sinkWriteChunkByFiler)
+	filerSink.DoInitialize(targetFiler.ToHttpAddress(), targetFiler.ToGrpcAddress(), targetPath, replicationStr, collection, ttlSec, diskType, targetGrpcDialOption, sinkWriteChunkByFiler)
 	filerSink.SetChunkConcurrency(chunkConcurrency)
 	filerSink.SetSourceFiler(filerSource)

@@ -328,7 +351,7 @@ func doSubscribeFilerMetaChanges(clientId int32, clientEpoch int32, grpcDialOpti
 	if statePtr != nil {
 		statePtr.Store(&syncState{
 			processor:            processor,
-			grpcDialOption:       grpcDialOption,
+			grpcDialOption:       targetGrpcDialOption,
 			targetFiler:          targetFiler,
 			sourcePath:           sourcePath,
 			sourceFilerSignature: sourceFilerSignature,
@@ -336,6 +359,7 @@ func doSubscribeFilerMetaChanges(clientId int32, clientEpoch int32, grpcDialOpti
 	}

 	var lastLogTsNs = time.Now().UnixNano()
+	var lastProgressedTsNs int64
 	var clientName = fmt.Sprintf("syncFrom_%s_To_%s", string(sourceFiler), string(targetFiler))
 	processEventFnWithOffset := pb.AddOffsetFunc(func(resp *filer_pb.SubscribeMetadataResponse) error {
 		processor.AddSyncJob(resp)
@@ -349,9 +373,21 @@ func doSubscribeFilerMetaChanges(clientId int32, clientEpoch int32, grpcDialOpti
 		now := time.Now().UnixNano()
 		glog.V(0).Infof("sync %s to %s progressed to %v %0.2f/sec", sourceFiler, targetFiler, time.Unix(0, offsetTsNs), float64(counter)/(float64(now-lastLogTsNs)/1e9))
 		lastLogTsNs = now
+		if offsetTsNs == lastProgressedTsNs {
+			for _, t := range filerSink.ActiveTransfers() {
+				if t.LastErr != "" {
+					glog.V(0).Infof("  %s %s: %d bytes received, %s, last error: %s",
+						t.ChunkFileId, t.Path, t.BytesReceived, t.Status, t.LastErr)
+				} else {
+					glog.V(0).Infof("  %s %s: %d bytes received, %s",
+						t.ChunkFileId, t.Path, t.BytesReceived, t.Status)
+				}
+			}
+		}
+		lastProgressedTsNs = offsetTsNs
 		// collect synchronous offset
 		statsCollect.FilerSyncOffsetGauge.WithLabelValues(sourceFiler.String(), targetFiler.String(), clientName, sourcePath).Set(float64(offsetTsNs))
-		return setOffset(grpcDialOption, targetFiler, getSignaturePrefixByPath(sourcePath), sourceFilerSignature, offsetTsNs)
+		return setOffset(targetGrpcDialOption, targetFiler, getSignaturePrefixByPath(sourcePath), sourceFilerSignature, offsetTsNs)
 	})

 	prefix := sourcePath
@@ -372,7 +408,7 @@ func doSubscribeFilerMetaChanges(clientId int32, clientEpoch int32, grpcDialOpti
 		EventErrorType:         pb.RetryForeverOnError,
 	}

-	return pb.FollowMetadata(sourceFiler, grpcDialOption, metadataFollowOption, processEventFnWithOffset)
+	return pb.FollowMetadata(sourceFiler, sourceGrpcDialOption, metadataFollowOption, processEventFnWithOffset)

 }

@@ -250,6 +250,7 @@ func initMiniS3Flags() {
 	miniS3Options.auditLogConfig = cmdMini.Flag.String("s3.auditLogConfig", "", "path to the audit log config file")
 	miniS3Options.allowDeleteBucketNotEmpty = miniS3AllowDeleteBucketNotEmpty
 	miniS3Options.externalUrl = cmdMini.Flag.String("s3.externalUrl", "", "the external URL clients use to connect (e.g. https://api.example.com:9000). Used for S3 signature verification behind a reverse proxy. Falls back to S3_EXTERNAL_URL env var.")
+	miniS3Options.defaultFileMode = cmdMini.Flag.String("s3.defaultFileMode", "", "default file mode for S3 uploaded objects, e.g. 0660, 0644, 0666")
 	// In mini mode, S3 uses the shared debug server started at line 681, not its own separate debug server
 	miniS3Options.debug = new(bool) // explicitly false
 	miniS3Options.debugPort = cmdMini.Flag.Int("s3.debug.port", 6060, "http port for debugging (unused in mini mode)")
@@ -819,6 +820,10 @@ func runMini(cmd *Command, args []string) bool {
 	miniFilerOptions.disableHttp = miniDisableHttp
 	miniMasterOptions.disableHttp = miniDisableHttp

+	// Share the S3 static identity config file with the filer so its
+	// credential manager can also serve static users.
+	miniFilerOptions.s3ConfigFile = miniS3Config
+
 	filerAddress := string(pb.NewServerAddress(*miniIp, *miniFilerOptions.port, *miniFilerOptions.portGrpc))
 	miniS3Options.filer = &filerAddress
 	miniWebDavOptions.filer = &filerAddress
@@ -10,6 +10,7 @@ import (
 	"net/http"
 	"os"
 	"runtime"
+	"strconv"
 	"strings"
 	"time"

@@ -36,6 +37,9 @@ var (
 	s3StandaloneOptions S3Options
 )

+// S3Options holds CLI flags for the S3 gateway.
+// Flags are registered in multiple commands: s3.go (standalone), server.go, filer.go, and mini.go.
+// When adding a new field, update all four flag registration sites.
 type S3Options struct {
 	filer                     *string
 	bindIp                    *string
@@ -68,6 +72,7 @@ type S3Options struct {
 	debugPort                 *int
 	cipher                    *bool
 	externalUrl               *string
+	defaultFileMode           *string
 }

 func init() {
@@ -103,6 +108,7 @@ func init() {
 	s3StandaloneOptions.debugPort = cmdS3.Flag.Int("debug.port", 6060, "http port for debugging")
 	s3StandaloneOptions.cipher = cmdS3.Flag.Bool("encryptVolumeData", false, "encrypt data on volume servers")
 	s3StandaloneOptions.externalUrl = cmdS3.Flag.String("externalUrl", "", "the external URL clients use to connect (e.g. https://api.example.com:9000). Used for S3 signature verification behind a reverse proxy. Falls back to S3_EXTERNAL_URL env var.")
+	s3StandaloneOptions.defaultFileMode = cmdS3.Flag.String("defaultFileMode", "", "default file mode for S3 uploaded objects, e.g. 0660, 0644, 0666")
 }

 var cmdS3 = &Command{
@@ -232,6 +238,17 @@ func (s3opt *S3Options) resolveExternalUrl() string {
 	return os.Getenv("S3_EXTERNAL_URL")
 }

+func (s3opt *S3Options) parseDefaultFileMode() (uint32, error) {
+	if s3opt.defaultFileMode == nil || *s3opt.defaultFileMode == "" {
+		return 0, nil
+	}
+	mode, err := strconv.ParseUint(*s3opt.defaultFileMode, 8, 32)
+	if err != nil {
+		return 0, fmt.Errorf("invalid defaultFileMode %q: %v", *s3opt.defaultFileMode, err)
+	}
+	return uint32(mode), nil
+}
+
 func (s3opt *S3Options) startS3Server() bool {

 	filerAddresses := pb.ServerAddresses(*s3opt.filer).ToAddresses()
@@ -298,6 +315,11 @@ func (s3opt *S3Options) startS3Server() bool {
 		*s3opt.bindIp = "0.0.0.0"
 	}

+	defaultFileMode, fileModeErr := s3opt.parseDefaultFileMode()
+	if fileModeErr != nil {
+		glog.Fatalf("S3 API Server startup error: %v", fileModeErr)
+	}
+
 	s3ApiServer, s3ApiServer_err = s3api.NewS3ApiServer(router, &s3api.S3ApiServerOption{
 		Filers:                    filerAddresses,
 		Masters:                   masterAddresses,
@@ -320,6 +342,7 @@ func (s3opt *S3Options) startS3Server() bool {
 		BindIp:                    *s3opt.bindIp,
 		GrpcPort:                  *s3opt.portGrpc,
 		ExternalUrl:               s3opt.resolveExternalUrl(),
+		DefaultFileMode:           defaultFileMode,
 	})
 	if s3ApiServer_err != nil {
 		glog.Fatalf("S3 API Server startup error: %v", s3ApiServer_err)
@@ -179,6 +179,19 @@ password = ""
 user = ""
 password = ""

+# SSE-S3 server-side encryption key management
+# These settings configure the Key Encryption Key (KEK) for S3 SSE-S3 encryption.
+# Set exactly one of kek or key. If neither is set, SSE-S3 is disabled.
+# Can also be set via env vars: WEED_S3_SSE_KEK, WEED_S3_SSE_KEY
+[s3.sse]
+# hex-encoded 256-bit key, same format as the legacy /etc/s3/sse_kek filer file.
+# Use this to migrate from a filer-stored KEK: copy the value from /etc/s3/sse_kek.
+# Generate a new one with: openssl rand -hex 32
+kek = ""
+# any secret string; a 256-bit key is derived automatically via HKDF-SHA256.
+# Cannot be used while /etc/s3/sse_kek exists on the filer — delete it first.
+key = ""
+
 # white list. It's checking request ip address.
 [guard]
 white_list = ""
@@ -180,6 +180,7 @@ func init() {
 	s3Options.iamReadOnly = cmdServer.Flag.Bool("s3.iam.readOnly", true, "disable IAM write operations on this server")
 	s3Options.cipher = cmdServer.Flag.Bool("s3.encryptVolumeData", false, "encrypt data on volume servers for S3 uploads")
 	s3Options.externalUrl = cmdServer.Flag.String("s3.externalUrl", "", "the external URL clients use to connect (e.g. https://api.example.com:9000). Used for S3 signature verification behind a reverse proxy. Falls back to S3_EXTERNAL_URL env var.")
+	s3Options.defaultFileMode = cmdServer.Flag.String("s3.defaultFileMode", "", "default file mode for S3 uploaded objects, e.g. 0660, 0644, 0666")

 	sftpOptions.port = cmdServer.Flag.Int("sftp.port", 2022, "SFTP server listen port")
 	sftpOptions.sshPrivateKey = cmdServer.Flag.String("sftp.sshPrivateKey", "", "path to the SSH private key file for host authentication")
@@ -371,6 +372,8 @@ func runServer(cmd *Command, args []string) bool {
 		} else if *serverIamConfig != "" && *s3Options.iamConfig != *serverIamConfig {
 			glog.V(0).Infof("both -s3.iam.config(%s) and -iam.config(%s) provided; using -s3.iam.config", *s3Options.iamConfig, *serverIamConfig)
 		}
+		// Share the S3 static identity config file with the filer
+		filerOptions.s3ConfigFile = s3Options.config
 		go func() {
 			time.Sleep(2 * time.Second)
 			s3Options.localFilerSocket = filerOptions.localSocket
@@ -57,42 +57,6 @@ func LoadCredentialConfiguration() (*CredentialConfig, error) {
 	}, nil
 }

-// GetCredentialStoreConfig extracts credential store configuration from command line flags
-// This is used when credential store is configured via command line instead of credential.toml
-func GetCredentialStoreConfig(store string, config util.Configuration, prefix string) *CredentialConfig {
-	if store == "" {
-		return nil
-	}
-
-	return &CredentialConfig{
-		Store:  store,
-		Config: config,
-		Prefix: prefix,
-	}
-}
-
-// MergeCredentialConfig merges command line credential config with credential.toml config
-// Command line flags take priority over credential.toml
-func MergeCredentialConfig(cmdLineStore string, cmdLineConfig util.Configuration, cmdLinePrefix string) (*CredentialConfig, error) {
-	// If command line credential store is specified, use it
-	if cmdLineStore != "" {
-		glog.V(0).Infof("Using command line credential configuration: store=%s", cmdLineStore)
-		return GetCredentialStoreConfig(cmdLineStore, cmdLineConfig, cmdLinePrefix), nil
-	}
-
-	// Otherwise, try to load from credential.toml
-	config, err := LoadCredentialConfiguration()
-	if err != nil {
-		return nil, err
-	}
-
-	if config == nil {
-		glog.V(1).Info("No credential store configured")
-	}
-
-	return config, nil
-}
-
 // NewCredentialManagerWithDefaults creates a credential manager with fallback to defaults
 // If explicitStore is provided, it will be used regardless of credential.toml
 // If explicitStore is empty, it tries credential.toml first, then defaults to "filer_etc"
@@ -3,14 +3,18 @@ package credential
 import (
 	"context"
 	"fmt"
+	"os"
 	"strings"
+	"sync"

+	"github.com/seaweedfs/seaweedfs/weed/glog"
 	"github.com/seaweedfs/seaweedfs/weed/pb"
 	"github.com/seaweedfs/seaweedfs/weed/pb/iam_pb"
 	"github.com/seaweedfs/seaweedfs/weed/s3api/policy_engine"
 	"github.com/seaweedfs/seaweedfs/weed/util"
 	"github.com/seaweedfs/seaweedfs/weed/wdclient"
 	"google.golang.org/grpc"
+	"google.golang.org/protobuf/encoding/protojson"
 )

 // FilerAddressSetter is an interface for credential stores that need a dynamic filer address
@@ -21,6 +25,15 @@ type FilerAddressSetter interface {
 // CredentialManager manages user credentials using a configurable store
 type CredentialManager struct {
 	Store CredentialStore
+	// staticMu protects staticIdentities and staticNames, which are written
+	// by SetStaticIdentities (startup + config reload) and read concurrently
+	// by LoadConfiguration, SaveConfiguration, GetStaticUsernames, and IsStaticIdentity.
+	staticMu sync.RWMutex
+	// staticIdentities holds identities loaded from a static config file (-s3.config).
+	// These are included in LoadConfiguration so that listing operations
+	// return all configured identities, not just dynamic ones from the store.
+	staticIdentities []*iam_pb.Identity
+	staticNames      map[string]bool
 }

 // NewCredentialManager creates a new credential manager with the specified store
@@ -74,13 +87,97 @@ func (cm *CredentialManager) GetStoreName() string {
 	return ""
 }

-// LoadConfiguration loads the S3 API configuration
-func (cm *CredentialManager) LoadConfiguration(ctx context.Context) (*iam_pb.S3ApiConfiguration, error) {
-	return cm.Store.LoadConfiguration(ctx)
+// SetStaticIdentities registers identities loaded from a static config file.
+// These identities are included in LoadConfiguration and ListUsers results
+// but are never persisted to the dynamic store.
+func (cm *CredentialManager) SetStaticIdentities(identities []*iam_pb.Identity) {
+	filtered := make([]*iam_pb.Identity, 0, len(identities))
+	names := make(map[string]bool, len(identities))
+	for _, ident := range identities {
+		if ident != nil {
+			filtered = append(filtered, ident)
+			names[ident.Name] = true
+		}
+	}
+	cm.staticMu.Lock()
+	cm.staticIdentities = filtered
+	cm.staticNames = names
+	cm.staticMu.Unlock()
 }

-// SaveConfiguration saves the S3 API configuration
+// IsStaticIdentity returns true if the named identity was loaded from static config.
+func (cm *CredentialManager) IsStaticIdentity(name string) bool {
+	cm.staticMu.RLock()
+	defer cm.staticMu.RUnlock()
+	return cm.staticNames[name]
+}
+
+// GetStaticIdentity returns the protobuf identity for a static user, or nil.
+func (cm *CredentialManager) GetStaticIdentity(name string) *iam_pb.Identity {
+	cm.staticMu.RLock()
+	defer cm.staticMu.RUnlock()
+	for _, ident := range cm.staticIdentities {
+		if ident.Name == name {
+			return ident
+		}
+	}
+	return nil
+}
+
+// GetStaticUsernames returns the names of all static identities.
+func (cm *CredentialManager) GetStaticUsernames() []string {
+	cm.staticMu.RLock()
+	defer cm.staticMu.RUnlock()
+	names := make([]string, 0, len(cm.staticIdentities))
+	for _, ident := range cm.staticIdentities {
+		names = append(names, ident.Name)
+	}
+	return names
+}
+
+// LoadConfiguration loads the S3 API configuration from the store and merges
+// in any static identities so that listing operations show all users.
+func (cm *CredentialManager) LoadConfiguration(ctx context.Context) (*iam_pb.S3ApiConfiguration, error) {
+	config, err := cm.Store.LoadConfiguration(ctx)
+	if err != nil {
+		return config, err
+	}
+	// Merge static identities that are not already in the dynamic config
+	cm.staticMu.RLock()
+	staticIdents := cm.staticIdentities
+	cm.staticMu.RUnlock()
+	if len(staticIdents) > 0 {
+		dynamicNames := make(map[string]bool, len(config.Identities))
+		for _, ident := range config.Identities {
+			dynamicNames[ident.Name] = true
+		}
+		for _, si := range staticIdents {
+			if !dynamicNames[si.Name] {
+				config.Identities = append(config.Identities, si)
+			}
+		}
+	}
+	return config, nil
+}
+
+// SaveConfiguration saves the S3 API configuration.
+// Static identities are filtered out before saving to the store.
+// The caller's config is not mutated.
 func (cm *CredentialManager) SaveConfiguration(ctx context.Context, config *iam_pb.S3ApiConfiguration) error {
+	cm.staticMu.RLock()
+	staticNames := cm.staticNames
+	cm.staticMu.RUnlock()
+	if len(staticNames) > 0 {
+		var dynamicOnly []*iam_pb.Identity
+		for _, ident := range config.Identities {
+			if !staticNames[ident.Name] {
+				dynamicOnly = append(dynamicOnly, ident)
+			}
+		}
+		configCopy := *config
+		configCopy.Identities = dynamicOnly
+		return cm.Store.SaveConfiguration(ctx, &configCopy)
+	}
 	return cm.Store.SaveConfiguration(ctx, config)
 }

@@ -104,7 +201,12 @@ func (cm *CredentialManager) DeleteUser(ctx context.Context, username string) er
 	return cm.Store.DeleteUser(ctx, username)
 }

-// ListUsers returns all usernames
+// ListUsers returns usernames from the dynamic store via cm.Store.ListUsers.
+// On store error the error is returned directly without merging static entries.
+// Static identities (cm.staticIdentities) are NOT included here because
+// internal callers (e.g. DeletePolicy) look up each user in the store and
+// would fail on non-existent static entries. External callers that need the
+// full list should merge GetStaticUsernames separately.
 func (cm *CredentialManager) ListUsers(ctx context.Context) ([]string, error) {
 	return cm.Store.ListUsers(ctx)
 }
@@ -169,6 +271,26 @@ func (cm *CredentialManager) UpdatePolicy(ctx context.Context, name string, docu
 	return cm.Store.PutPolicy(ctx, name, document)
 }

+// LoadS3ConfigFile reads a static S3 identity config file and registers
+// the identities so they appear in LoadConfiguration and listing results.
+func (cm *CredentialManager) LoadS3ConfigFile(path string) error {
+	content, err := os.ReadFile(path)
+	if err != nil {
+		return fmt.Errorf("read %s: %w", path, err)
+	}
+	config := &iam_pb.S3ApiConfiguration{}
+	opts := protojson.UnmarshalOptions{DiscardUnknown: true, AllowPartial: true}
+	if err := opts.Unmarshal(content, config); err != nil {
+		return fmt.Errorf("parse %s: %w", path, err)
+	}
+	for _, ident := range config.Identities {
+		ident.IsStatic = true
+	}
+	cm.SetStaticIdentities(config.Identities)
+	glog.V(1).Infof("Loaded %d static identities from %s", len(config.Identities), path)
+	return nil
+}
+
 // Shutdown performs cleanup
 func (cm *CredentialManager) Shutdown() {
 	if cm.Store != nil {
@@ -207,32 +207,6 @@ func (store *FilerEtcStore) loadPoliciesFromMultiFile(ctx context.Context, polic
 	})
 }

-func (store *FilerEtcStore) migratePoliciesToMultiFile(ctx context.Context, policies map[string]policy_engine.PolicyDocument) error {
-	glog.Infof("Migrating IAM policies to multi-file layout...")
-
-	// 1. Save all policies to individual files
-	for name, policy := range policies {
-		if err := store.savePolicy(ctx, name, policy); err != nil {
-			return err
-		}
-	}
-
-	// 2. Rename legacy file
-	return store.withFilerClient(func(client filer_pb.SeaweedFilerClient) error {
-		_, err := client.AtomicRenameEntry(ctx, &filer_pb.AtomicRenameEntryRequest{
-			OldDirectory: filer.IamConfigDirectory,
-			OldName:      filer.IamPoliciesFile,
-			NewDirectory: filer.IamConfigDirectory,
-			NewName:      IamLegacyPoliciesOldFile,
-		})
-		if err != nil {
-			glog.Errorf("Failed to rename legacy IAM policies file %s/%s to %s: %v",
-				filer.IamConfigDirectory, filer.IamPoliciesFile, IamLegacyPoliciesOldFile, err)
-		}
-		return err
-	})
-}
-
 func (store *FilerEtcStore) savePolicy(ctx context.Context, name string, document policy_engine.PolicyDocument) error {
 	if err := validatePolicyName(name); err != nil {
 		return err
@@ -1,221 +0,0 @@
-package credential
-
-import (
-	"context"
-	"fmt"
-
-	"github.com/seaweedfs/seaweedfs/weed/glog"
-	"github.com/seaweedfs/seaweedfs/weed/pb/iam_pb"
-	"github.com/seaweedfs/seaweedfs/weed/util"
-)
-
-// MigrateCredentials migrates credentials from one store to another
-func MigrateCredentials(fromStoreName, toStoreName CredentialStoreTypeName, configuration util.Configuration, fromPrefix, toPrefix string) error {
-	ctx := context.Background()
-
-	// Create source credential manager
-	fromCM, err := NewCredentialManager(fromStoreName, configuration, fromPrefix)
-	if err != nil {
-		return fmt.Errorf("failed to create source credential manager (%s): %v", fromStoreName, err)
-	}
-	defer fromCM.Shutdown()
-
-	// Create destination credential manager
-	toCM, err := NewCredentialManager(toStoreName, configuration, toPrefix)
-	if err != nil {
-		return fmt.Errorf("failed to create destination credential manager (%s): %v", toStoreName, err)
-	}
-	defer toCM.Shutdown()
-
-	// Load configuration from source
-	glog.Infof("Loading configuration from %s store...", fromStoreName)
-	config, err := fromCM.LoadConfiguration(ctx)
-	if err != nil {
-		return fmt.Errorf("failed to load configuration from source store: %w", err)
-	}
-
-	if config == nil || len(config.Identities) == 0 {
-		glog.Info("No identities found in source store")
-		return nil
-	}
-
-	glog.Infof("Found %d identities in source store", len(config.Identities))
-
-	// Migrate each identity
-	var migrated, failed int
-	for _, identity := range config.Identities {
-		glog.V(1).Infof("Migrating user: %s", identity.Name)
-
-		// Check if user already exists in destination
-		existingUser, err := toCM.GetUser(ctx, identity.Name)
-		if err != nil && err != ErrUserNotFound {
-			glog.Errorf("Failed to check if user %s exists in destination: %v", identity.Name, err)
-			failed++
-			continue
-		}
-
-		if existingUser != nil {
-			glog.Warningf("User %s already exists in destination store, skipping", identity.Name)
-			continue
-		}
-
-		// Create user in destination
-		err = toCM.CreateUser(ctx, identity)
-		if err != nil {
-			glog.Errorf("Failed to create user %s in destination store: %v", identity.Name, err)
-			failed++
-			continue
-		}
-
-		migrated++
-		glog.V(1).Infof("Successfully migrated user: %s", identity.Name)
-	}
-
-	glog.Infof("Migration completed: %d migrated, %d failed", migrated, failed)
-
-	if failed > 0 {
-		return fmt.Errorf("migration completed with %d failures", failed)
-	}
-
-	return nil
-}
-
-// ExportCredentials exports credentials from a store to a configuration
-func ExportCredentials(storeName CredentialStoreTypeName, configuration util.Configuration, prefix string) (*iam_pb.S3ApiConfiguration, error) {
-	ctx := context.Background()
-
-	// Create credential manager
-	cm, err := NewCredentialManager(storeName, configuration, prefix)
-	if err != nil {
-		return nil, fmt.Errorf("failed to create credential manager (%s): %v", storeName, err)
-	}
-	defer cm.Shutdown()
-
-	// Load configuration
-	config, err := cm.LoadConfiguration(ctx)
-	if err != nil {
-		return nil, fmt.Errorf("failed to load configuration: %w", err)
-	}
-
-	return config, nil
-}
-
-// ImportCredentials imports credentials from a configuration to a store
-func ImportCredentials(storeName CredentialStoreTypeName, configuration util.Configuration, prefix string, config *iam_pb.S3ApiConfiguration) error {
-	ctx := context.Background()
-
-	// Create credential manager
-	cm, err := NewCredentialManager(storeName, configuration, prefix)
-	if err != nil {
-		return fmt.Errorf("failed to create credential manager (%s): %v", storeName, err)
-	}
-	defer cm.Shutdown()
-
-	// Import each identity
-	var imported, failed int
-	for _, identity := range config.Identities {
-		glog.V(1).Infof("Importing user: %s", identity.Name)
-
-		// Check if user already exists
-		existingUser, err := cm.GetUser(ctx, identity.Name)
-		if err != nil && err != ErrUserNotFound {
-			glog.Errorf("Failed to check if user %s exists: %v", identity.Name, err)
-			failed++
-			continue
-		}
-
-		if existingUser != nil {
-			glog.Warningf("User %s already exists, skipping", identity.Name)
-			continue
-		}
-
-		// Create user
-		err = cm.CreateUser(ctx, identity)
-		if err != nil {
-			glog.Errorf("Failed to create user %s: %v", identity.Name, err)
-			failed++
-			continue
-		}
-
-		imported++
-		glog.V(1).Infof("Successfully imported user: %s", identity.Name)
-	}
-
-	glog.Infof("Import completed: %d imported, %d failed", imported, failed)
-
-	if failed > 0 {
-		return fmt.Errorf("import completed with %d failures", failed)
-	}
-
-	return nil
-}
-
-// ValidateCredentials validates that all credentials in a store are accessible
-func ValidateCredentials(storeName CredentialStoreTypeName, configuration util.Configuration, prefix string) error {
-	ctx := context.Background()
-
-	// Create credential manager
-	cm, err := NewCredentialManager(storeName, configuration, prefix)
-	if err != nil {
-		return fmt.Errorf("failed to create credential manager (%s): %v", storeName, err)
-	}
-	defer cm.Shutdown()
-
-	// Load configuration
-	config, err := cm.LoadConfiguration(ctx)
-	if err != nil {
-		return fmt.Errorf("failed to load configuration: %w", err)
-	}
-
-	if config == nil || len(config.Identities) == 0 {
-		glog.Info("No identities found in store")
-		return nil
-	}
-
-	glog.Infof("Validating %d identities...", len(config.Identities))
-
-	// Validate each identity
-	var validated, failed int
-	for _, identity := range config.Identities {
-		// Check if user can be retrieved
-		user, err := cm.GetUser(ctx, identity.Name)
-		if err != nil {
-			glog.Errorf("Failed to retrieve user %s: %v", identity.Name, err)
-			failed++
-			continue
-		}
-
-		if user == nil {
-			glog.Errorf("User %s not found", identity.Name)
-			failed++
-			continue
-		}
-
-		// Validate access keys
-		for _, credential := range identity.Credentials {
-			accessKeyUser, err := cm.GetUserByAccessKey(ctx, credential.AccessKey)
-			if err != nil {
-				glog.Errorf("Failed to retrieve user by access key %s: %v", credential.AccessKey, err)
-				failed++
-				continue
-			}
-
-			if accessKeyUser == nil || accessKeyUser.Name != identity.Name {
-				glog.Errorf("Access key %s does not map to correct user %s", credential.AccessKey, identity.Name)
-				failed++
-				continue
-			}
-		}
-
-		validated++
-		glog.V(1).Infof("Successfully validated user: %s", identity.Name)
-	}
-
-	glog.Infof("Validation completed: %d validated, %d failed", validated, failed)
-
-	if failed > 0 {
-		return fmt.Errorf("validation completed with %d failures", failed)
-	}
-
-	return nil
-}
@@ -246,10 +246,6 @@ func NewLogFileEntryCollector(f *Filer, startPosition log_buffer.MessagePosition
 	}
 }

-func (c *LogFileEntryCollector) hasMore() bool {
-	return c.dayEntryQueue.Len() > 0
-}
-
 func (c *LogFileEntryCollector) collectMore(v *OrderedLogVisitor) (err error) {
 	dayEntry := c.dayEntryQueue.Dequeue()
 	if dayEntry == nil {
@@ -128,6 +128,9 @@ func (fsw *FilerStoreWrapper) Initialize(configuration util.Configuration, prefi
 }

 func (fsw *FilerStoreWrapper) InsertEntry(ctx context.Context, entry *Entry) error {
+	if err := ctx.Err(); err != nil {
+		return err
+	}
 	ctx = context.WithoutCancel(ctx)
 	actualStore := fsw.getActualStore(entry.FullPath)
 	stats.FilerStoreCounter.WithLabelValues(actualStore.GetName(), "insert").Inc()
@@ -155,6 +158,9 @@ func (fsw *FilerStoreWrapper) InsertEntry(ctx context.Context, entry *Entry) err
 // InsertEntryKnownAbsent skips the pre-insert FindEntry path when the caller has
 // already established that the target path does not exist.
 func (fsw *FilerStoreWrapper) InsertEntryKnownAbsent(ctx context.Context, entry *Entry) error {
+	if err := ctx.Err(); err != nil {
+		return err
+	}
 	ctx = context.WithoutCancel(ctx)
 	actualStore := fsw.getActualStore(entry.FullPath)
 	stats.FilerStoreCounter.WithLabelValues(actualStore.GetName(), "insert").Inc()
@@ -178,6 +184,9 @@ func (fsw *FilerStoreWrapper) InsertEntryKnownAbsent(ctx context.Context, entry
 }

 func (fsw *FilerStoreWrapper) UpdateEntry(ctx context.Context, entry *Entry) error {
+	if err := ctx.Err(); err != nil {
+		return err
+	}
 	ctx = context.WithoutCancel(ctx)
 	actualStore := fsw.getActualStore(entry.FullPath)
 	stats.FilerStoreCounter.WithLabelValues(actualStore.GetName(), "update").Inc()
@@ -236,6 +245,9 @@ func (fsw *FilerStoreWrapper) FindEntry(ctx context.Context, fp util.FullPath) (
 }

 func (fsw *FilerStoreWrapper) DeleteEntry(ctx context.Context, fp util.FullPath) (err error) {
+	if err := ctx.Err(); err != nil {
+		return err
+	}
 	ctx = context.WithoutCancel(ctx)
 	actualStore := fsw.getActualStore(fp)
 	stats.FilerStoreCounter.WithLabelValues(actualStore.GetName(), "delete").Inc()
@@ -264,6 +276,9 @@ func (fsw *FilerStoreWrapper) DeleteEntry(ctx context.Context, fp util.FullPath)
 }

 func (fsw *FilerStoreWrapper) DeleteOneEntry(ctx context.Context, existingEntry *Entry) (err error) {
+	if err := ctx.Err(); err != nil {
+		return err
+	}
 	ctx = context.WithoutCancel(ctx)
 	actualStore := fsw.getActualStore(existingEntry.FullPath)
 	stats.FilerStoreCounter.WithLabelValues(actualStore.GetName(), "delete").Inc()
@@ -288,6 +303,9 @@ func (fsw *FilerStoreWrapper) DeleteOneEntry(ctx context.Context, existingEntry
 }

 func (fsw *FilerStoreWrapper) DeleteFolderChildren(ctx context.Context, fp util.FullPath) (err error) {
+	if err := ctx.Err(); err != nil {
+		return err
+	}
 	ctx = context.WithoutCancel(ctx)
 	actualStore := fsw.getActualStore(fp + "/")
 	stats.FilerStoreCounter.WithLabelValues(actualStore.GetName(), "deleteFolderChildren").Inc()
@@ -394,11 +412,17 @@ func (fsw *FilerStoreWrapper) prefixFilterEntries(ctx context.Context, dirPath u
 }

 func (fsw *FilerStoreWrapper) BeginTransaction(ctx context.Context) (context.Context, error) {
+	if err := ctx.Err(); err != nil {
+		return nil, err
+	}
 	ctx = context.WithoutCancel(ctx)
 	return fsw.getDefaultStore().BeginTransaction(ctx)
 }

 func (fsw *FilerStoreWrapper) CommitTransaction(ctx context.Context) error {
+	if err := ctx.Err(); err != nil {
+		return err
+	}
 	ctx = context.WithoutCancel(ctx)
 	return fsw.getDefaultStore().CommitTransaction(ctx)
 }
@@ -413,6 +437,9 @@ func (fsw *FilerStoreWrapper) Shutdown() {
 }

 func (fsw *FilerStoreWrapper) KvPut(ctx context.Context, key []byte, value []byte) (err error) {
+	if err := ctx.Err(); err != nil {
+		return err
+	}
 	ctx = context.WithoutCancel(ctx)
 	return fsw.getDefaultStore().KvPut(ctx, key, value)
 }
@@ -421,6 +448,9 @@ func (fsw *FilerStoreWrapper) KvGet(ctx context.Context, key []byte) (value []by
 	return fsw.getDefaultStore().KvGet(ctx, key)
 }
 func (fsw *FilerStoreWrapper) KvDelete(ctx context.Context, key []byte) (err error) {
+	if err := ctx.Err(); err != nil {
+		return err
+	}
 	ctx = context.WithoutCancel(ctx)
 	return fsw.getDefaultStore().KvDelete(ctx, key)
 }
@@ -2,8 +2,10 @@ package filer

 import (
 	"context"
+	"errors"
 	"os"
 	"testing"
+	"time"

 	"github.com/seaweedfs/seaweedfs/weed/util"
 	"github.com/stretchr/testify/assert"
@@ -69,3 +71,137 @@ func TestFilerStoreWrapperMimeNormalization(t *testing.T) {
 		}
 	}
 }
+
+// cancelledCtx returns a context that is already cancelled.
+func cancelledCtx() context.Context {
+	ctx, cancel := context.WithCancel(context.Background())
+	cancel()
+	return ctx
+}
+
+// expiredCtx returns a context whose deadline has already passed.
+func expiredCtx() context.Context {
+	ctx, cancel := context.WithDeadline(context.Background(), time.Now().Add(-time.Second))
+	cancel() // release resources immediately as it's already expired
+	return ctx
+}
+
+func TestFilerStoreWrapperWriteOpsRejectCancelledContext(t *testing.T) {
+	newEntry := func(path string) *Entry {
+		return &Entry{
+			FullPath: util.FullPath(path),
+			Attr:     Attr{Mode: 0o660, Mime: "application/octet-stream"},
+		}
+	}
+
+	// Each write operation that should be guarded.
+	writeOps := []struct {
+		name string
+		run  func(*FilerStoreWrapper, context.Context) error
+	}{
+		{"InsertEntry", func(fsw *FilerStoreWrapper, ctx context.Context) error {
+			return fsw.InsertEntry(ctx, newEntry("/test/a"))
+		}},
+		{"InsertEntryKnownAbsent", func(fsw *FilerStoreWrapper, ctx context.Context) error {
+			return fsw.InsertEntryKnownAbsent(ctx, newEntry("/test/b"))
+		}},
+		{"UpdateEntry", func(fsw *FilerStoreWrapper, ctx context.Context) error {
+			_ = fsw.InsertEntry(context.Background(), newEntry("/test/c"))
+			return fsw.UpdateEntry(ctx, newEntry("/test/c"))
+		}},
+		{"DeleteEntry", func(fsw *FilerStoreWrapper, ctx context.Context) error {
+			_ = fsw.InsertEntry(context.Background(), newEntry("/test/d"))
+			return fsw.DeleteEntry(ctx, "/test/d")
+		}},
+		{"DeleteOneEntry", func(fsw *FilerStoreWrapper, ctx context.Context) error {
+			e := newEntry("/test/e")
+			_ = fsw.InsertEntry(context.Background(), e)
+			return fsw.DeleteOneEntry(ctx, e)
+		}},
+		{"DeleteFolderChildren", func(fsw *FilerStoreWrapper, ctx context.Context) error {
+			_ = fsw.InsertEntry(context.Background(), newEntry("/test/folder/child"))
+			return fsw.DeleteFolderChildren(ctx, "/test/folder")
+		}},
+		{"BeginTransaction", func(fsw *FilerStoreWrapper, ctx context.Context) error {
+			_, err := fsw.BeginTransaction(ctx)
+			return err
+		}},
+		{"CommitTransaction", func(fsw *FilerStoreWrapper, ctx context.Context) error {
+			return fsw.CommitTransaction(ctx)
+		}},
+		{"KvPut", func(fsw *FilerStoreWrapper, ctx context.Context) error {
+			return fsw.KvPut(ctx, []byte("k"), []byte("v"))
+		}},
+		{"KvDelete", func(fsw *FilerStoreWrapper, ctx context.Context) error {
+			_ = fsw.KvPut(context.Background(), []byte("k"), []byte("v"))
+			return fsw.KvDelete(ctx, []byte("k"))
+		}},
+	}
+
+	badContexts := []struct {
+		name      string
+		ctx       context.Context
+		wantError error
+	}{
+		{"cancelled", cancelledCtx(), context.Canceled},
+		{"deadline exceeded", expiredCtx(), context.DeadlineExceeded},
+	}
+
+	for _, op := range writeOps {
+		for _, bc := range badContexts {
+			t.Run(op.name+"/"+bc.name, func(t *testing.T) {
+				wrapper := NewFilerStoreWrapper(newStubFilerStore())
+				err := op.run(wrapper, bc.ctx)
+				require.Error(t, err)
+				assert.True(t, errors.Is(err, bc.wantError), "got %v, want %v", err, bc.wantError)
+			})
+		}
+	}
+}
+
+func TestFilerStoreWrapperWriteOpsSucceedWithActiveContext(t *testing.T) {
+	wrapper := NewFilerStoreWrapper(newStubFilerStore())
+	ctx := context.Background()
+	entry := &Entry{
+		FullPath: util.FullPath("/test/obj"),
+		Attr:     Attr{Mode: 0o660},
+	}
+
+	require.NoError(t, wrapper.InsertEntry(ctx, entry))
+	require.NoError(t, wrapper.UpdateEntry(ctx, entry))
+	require.NoError(t, wrapper.DeleteOneEntry(ctx, entry))
+	require.NoError(t, wrapper.InsertEntryKnownAbsent(ctx, entry))
+	require.NoError(t, wrapper.DeleteEntry(ctx, entry.FullPath))
+	require.NoError(t, wrapper.KvPut(ctx, []byte("k"), []byte("v")))
+	require.NoError(t, wrapper.KvDelete(ctx, []byte("k")))
+
+	txCtx, err := wrapper.BeginTransaction(ctx)
+	require.NoError(t, err)
+	require.NoError(t, wrapper.CommitTransaction(txCtx))
+}
+
+func TestFilerStoreWrapperReadOpsSucceedWithCancelledContext(t *testing.T) {
+	wrapper := NewFilerStoreWrapper(newStubFilerStore())
+	entry := &Entry{
+		FullPath: util.FullPath("/test/readable"),
+		Attr:     Attr{Mode: 0o660},
+	}
+	require.NoError(t, wrapper.InsertEntry(context.Background(), entry))
+	require.NoError(t, wrapper.KvPut(context.Background(), []byte("rk"), []byte("rv")))
+
+	ctx := cancelledCtx()
+
+	_, err := wrapper.FindEntry(ctx, entry.FullPath)
+	assert.NoError(t, err)
+
+	_, err = wrapper.KvGet(ctx, []byte("rk"))
+	assert.NoError(t, err)
+}
+
+// RollbackTransaction must succeed even when the context is cancelled or
+// expired, because it is a cleanup operation called after failures.
+func TestFilerStoreWrapperRollbackSucceedsWithCancelledContext(t *testing.T) {
+	wrapper := NewFilerStoreWrapper(newStubFilerStore())
+	assert.NoError(t, wrapper.RollbackTransaction(cancelledCtx()))
+	assert.NoError(t, wrapper.RollbackTransaction(expiredCtx()))
+}
@@ -2,7 +2,6 @@ package filer

 import (
 	"context"
-	"sync"

 	"github.com/seaweedfs/seaweedfs/weed/glog"
 	"github.com/seaweedfs/seaweedfs/weed/pb/filer_pb"
@@ -36,39 +35,3 @@ func Replay(filerStore FilerStore, resp *filer_pb.SubscribeMetadataResponse) err

 	return nil
 }
-
-// ParallelProcessDirectoryStructure processes each entry in parallel, and also ensure parent directories are processed first.
-// This also assumes the parent directories are in the entryChan already.
-func ParallelProcessDirectoryStructure(entryChan chan *Entry, concurrency int, eachEntryFn func(entry *Entry) error) (firstErr error) {
-
-	executors := util.NewLimitedConcurrentExecutor(concurrency)
-
-	var wg sync.WaitGroup
-	for entry := range entryChan {
-		wg.Add(1)
-		if entry.IsDirectory() {
-			func() {
-				defer wg.Done()
-				if err := eachEntryFn(entry); err != nil {
-					if firstErr == nil {
-						firstErr = err
-					}
-				}
-			}()
-		} else {
-			executors.Execute(func() {
-				defer wg.Done()
-				if err := eachEntryFn(entry); err != nil {
-					if firstErr == nil {
-						firstErr = err
-					}
-				}
-			})
-		}
-		if firstErr != nil {
-			break
-		}
-	}
-	wg.Wait()
-	return
-}
@@ -173,14 +173,16 @@ func (store *UniversalRedisStore) ListDirectoryEntries(ctx context.Context, dirP
 		members = members[:limit]
 	}

+	var entry *filer.Entry
 	// fetch entry meta
 	for _, fileName := range members {
 		path := util.NewFullPath(string(dirPath), fileName)
-		entry, err := store.FindEntry(ctx, path)
+		entry, err = store.FindEntry(ctx, path)
 		lastFileName = fileName
 		if err != nil {
 			glog.V(0).InfofCtx(ctx, "list %s : %v", path, err)
 			if err == filer_pb.ErrNotFound {
+				err = nil
 				continue
 			}
 		} else {
@@ -16,15 +16,6 @@ type ItemList struct {
 	prefix    string
 }

-func newItemList(client redis.UniversalClient, prefix string, store skiplist.ListStore, batchSize int) *ItemList {
-	return &ItemList{
-		skipList:  skiplist.New(store),
-		batchSize: batchSize,
-		client:    client,
-		prefix:    prefix,
-	}
-}
-
 /*
 Be reluctant to create new nodes. Try to fit into either previous node or next node.
 Prefer to add to previous node.
@@ -1,48 +0,0 @@
-package redis_lua
-
-import (
-	"github.com/redis/go-redis/v9"
-	"github.com/seaweedfs/seaweedfs/weed/filer"
-	"github.com/seaweedfs/seaweedfs/weed/util"
-)
-
-func init() {
-	filer.Stores = append(filer.Stores, &RedisLuaClusterStore{})
-}
-
-type RedisLuaClusterStore struct {
-	UniversalRedisLuaStore
-}
-
-func (store *RedisLuaClusterStore) GetName() string {
-	return "redis_lua_cluster"
-}
-
-func (store *RedisLuaClusterStore) Initialize(configuration util.Configuration, prefix string) (err error) {
-
-	configuration.SetDefault(prefix+"useReadOnly", false)
-	configuration.SetDefault(prefix+"routeByLatency", false)
-
-	return store.initialize(
-		configuration.GetStringSlice(prefix+"addresses"),
-		configuration.GetString(prefix+"username"),
-		configuration.GetString(prefix+"password"),
-		configuration.GetString(prefix+"keyPrefix"),
-		configuration.GetBool(prefix+"useReadOnly"),
-		configuration.GetBool(prefix+"routeByLatency"),
-		configuration.GetStringSlice(prefix+"superLargeDirectories"),
-	)
-}
-
-func (store *RedisLuaClusterStore) initialize(addresses []string, username string, password string, keyPrefix string, readOnly, routeByLatency bool, superLargeDirectories []string) (err error) {
-	store.Client = redis.NewClusterClient(&redis.ClusterOptions{
-		Addrs:          addresses,
-		Username:       username,
-		Password:       password,
-		ReadOnly:       readOnly,
-		RouteByLatency: routeByLatency,
-	})
-	store.keyPrefix = keyPrefix
-	store.loadSuperLargeDirectories(superLargeDirectories)
-	return
-}
@@ -1,48 +0,0 @@
-package redis_lua
-
-import (
-	"time"
-
-	"github.com/redis/go-redis/v9"
-	"github.com/seaweedfs/seaweedfs/weed/filer"
-	"github.com/seaweedfs/seaweedfs/weed/util"
-)
-
-func init() {
-	filer.Stores = append(filer.Stores, &RedisLuaSentinelStore{})
-}
-
-type RedisLuaSentinelStore struct {
-	UniversalRedisLuaStore
-}
-
-func (store *RedisLuaSentinelStore) GetName() string {
-	return "redis_lua_sentinel"
-}
-
-func (store *RedisLuaSentinelStore) Initialize(configuration util.Configuration, prefix string) (err error) {
-	return store.initialize(
-		configuration.GetStringSlice(prefix+"addresses"),
-		configuration.GetString(prefix+"masterName"),
-		configuration.GetString(prefix+"username"),
-		configuration.GetString(prefix+"password"),
-		configuration.GetInt(prefix+"database"),
-		configuration.GetString(prefix+"keyPrefix"),
-	)
-}
-
-func (store *RedisLuaSentinelStore) initialize(addresses []string, masterName string, username string, password string, database int, keyPrefix string) (err error) {
-	store.Client = redis.NewFailoverClient(&redis.FailoverOptions{
-		MasterName:      masterName,
-		SentinelAddrs:   addresses,
-		Username:        username,
-		Password:        password,
-		DB:              database,
-		MinRetryBackoff: time.Millisecond * 100,
-		MaxRetryBackoff: time.Minute * 1,
-		ReadTimeout:     time.Second * 30,
-		WriteTimeout:    time.Second * 5,
-	})
-	store.keyPrefix = keyPrefix
-	return
-}
@@ -1,42 +0,0 @@
-package redis_lua
-
-import (
-	"github.com/redis/go-redis/v9"
-	"github.com/seaweedfs/seaweedfs/weed/filer"
-	"github.com/seaweedfs/seaweedfs/weed/util"
-)
-
-func init() {
-	filer.Stores = append(filer.Stores, &RedisLuaStore{})
-}
-
-type RedisLuaStore struct {
-	UniversalRedisLuaStore
-}
-
-func (store *RedisLuaStore) GetName() string {
-	return "redis_lua"
-}
-
-func (store *RedisLuaStore) Initialize(configuration util.Configuration, prefix string) (err error) {
-	return store.initialize(
-		configuration.GetString(prefix+"address"),
-		configuration.GetString(prefix+"username"),
-		configuration.GetString(prefix+"password"),
-		configuration.GetInt(prefix+"database"),
-		configuration.GetString(prefix+"keyPrefix"),
-		configuration.GetStringSlice(prefix+"superLargeDirectories"),
-	)
-}
-
-func (store *RedisLuaStore) initialize(hostPort string, username string, password string, database int, keyPrefix string, superLargeDirectories []string) (err error) {
-	store.Client = redis.NewClient(&redis.Options{
-		Addr:     hostPort,
-		Username: username,
-		Password: password,
-		DB:       database,
-	})
-	store.keyPrefix = keyPrefix
-	store.loadSuperLargeDirectories(superLargeDirectories)
-	return
-}
@@ -1,19 +0,0 @@
-- KEYS[1]: full path of entry
-local fullpath = KEYS[1]
-- KEYS[2]: full path of entry
-local fullpath_list_key = KEYS[2]
-- KEYS[3]: dir of the entry
-local dir_list_key = KEYS[3]
-
-- ARGV[1]: isSuperLargeDirectory
-local isSuperLargeDirectory = ARGV[1] == "1"
-- ARGV[2]: name of the entry
-local name = ARGV[2]
-
-redis.call("DEL", fullpath, fullpath_list_key)
-
-if not isSuperLargeDirectory and name ~= "" then
-    redis.call("ZREM", dir_list_key, name)
-end
-
-return 0
@@ -1,15 +0,0 @@
-- KEYS[1]: full path of entry
-local fullpath = KEYS[1]
-
-if fullpath ~= "" and string.sub(fullpath, -1) == "/" then
-    fullpath = string.sub(fullpath, 0, -2)
-end
-
-local files = redis.call("ZRANGE", fullpath .. "\0", "0", "-1")
-
-for _, name in ipairs(files) do
-    local file_path = fullpath .. "/" .. name
-    redis.call("DEL", file_path, file_path .. "\0")
-end
-
-return 0
@@ -1,25 +0,0 @@
-package stored_procedure
-
-import (
-	_ "embed"
-
-	"github.com/redis/go-redis/v9"
-)
-
-func init() {
-	InsertEntryScript = redis.NewScript(insertEntry)
-	DeleteEntryScript = redis.NewScript(deleteEntry)
-	DeleteFolderChildrenScript = redis.NewScript(deleteFolderChildren)
-}
-
-//go:embed insert_entry.lua
-var insertEntry string
-var InsertEntryScript *redis.Script
-
-//go:embed delete_entry.lua
-var deleteEntry string
-var DeleteEntryScript *redis.Script
-
-//go:embed delete_folder_children.lua
-var deleteFolderChildren string
-var DeleteFolderChildrenScript *redis.Script
@@ -1,27 +0,0 @@
-- KEYS[1]: full path of entry
-local full_path = KEYS[1]
-- KEYS[2]: dir of the entry
-local dir_list_key = KEYS[2]
-
-- ARGV[1]: content of the entry
-local entry = ARGV[1]
-- ARGV[2]: TTL of the entry
-local ttlSec = tonumber(ARGV[2])
-- ARGV[3]: isSuperLargeDirectory
-local isSuperLargeDirectory = ARGV[3] == "1"
-- ARGV[4]: zscore of the entry in zset
-local zscore = tonumber(ARGV[4])
-- ARGV[5]: name of the entry
-local name = ARGV[5]
-
-if ttlSec > 0 then
-    redis.call("SET", full_path, entry, "EX", ttlSec)
-else
-    redis.call("SET", full_path, entry)
-end
-
-if not isSuperLargeDirectory and name ~= "" then
-    redis.call("ZADD", dir_list_key, "NX", zscore, name)
-end
-
-return 0
@@ -1,206 +0,0 @@
-package redis_lua
-
-import (
-	"context"
-	"fmt"
-	"time"
-
-	"github.com/redis/go-redis/v9"
-
-	"github.com/seaweedfs/seaweedfs/weed/filer"
-	"github.com/seaweedfs/seaweedfs/weed/filer/redis_lua/stored_procedure"
-	"github.com/seaweedfs/seaweedfs/weed/glog"
-	"github.com/seaweedfs/seaweedfs/weed/pb/filer_pb"
-	"github.com/seaweedfs/seaweedfs/weed/util"
-)
-
-const (
-	DIR_LIST_MARKER = "\x00"
-)
-
-type UniversalRedisLuaStore struct {
-	Client                  redis.UniversalClient
-	keyPrefix               string
-	superLargeDirectoryHash map[string]bool
-}
-
-func (store *UniversalRedisLuaStore) isSuperLargeDirectory(dir string) (isSuperLargeDirectory bool) {
-	_, isSuperLargeDirectory = store.superLargeDirectoryHash[dir]
-	return
-}
-
-func (store *UniversalRedisLuaStore) loadSuperLargeDirectories(superLargeDirectories []string) {
-	// set directory hash
-	store.superLargeDirectoryHash = make(map[string]bool)
-	for _, dir := range superLargeDirectories {
-		store.superLargeDirectoryHash[dir] = true
-	}
-}
-
-func (store *UniversalRedisLuaStore) getKey(key string) string {
-	if store.keyPrefix == "" {
-		return key
-	}
-	return store.keyPrefix + key
-}
-
-func (store *UniversalRedisLuaStore) BeginTransaction(ctx context.Context) (context.Context, error) {
-	return ctx, nil
-}
-func (store *UniversalRedisLuaStore) CommitTransaction(ctx context.Context) error {
-	return nil
-}
-func (store *UniversalRedisLuaStore) RollbackTransaction(ctx context.Context) error {
-	return nil
-}
-
-func (store *UniversalRedisLuaStore) InsertEntry(ctx context.Context, entry *filer.Entry) (err error) {
-
-	value, err := entry.EncodeAttributesAndChunks()
-	if err != nil {
-		return fmt.Errorf("encoding %s %+v: %v", entry.FullPath, entry.Attr, err)
-	}
-
-	if len(entry.GetChunks()) > filer.CountEntryChunksForGzip {
-		value = util.MaybeGzipData(value)
-	}
-
-	dir, name := entry.FullPath.DirAndName()
-
-	err = stored_procedure.InsertEntryScript.Run(ctx, store.Client,
-		[]string{store.getKey(string(entry.FullPath)), store.getKey(genDirectoryListKey(dir))},
-		value, entry.TtlSec,
-		store.isSuperLargeDirectory(dir), 0, name).Err()
-
-	if err != nil {
-		return fmt.Errorf("persisting %s : %v", entry.FullPath, err)
-	}
-
-	return nil
-}
-
-func (store *UniversalRedisLuaStore) UpdateEntry(ctx context.Context, entry *filer.Entry) (err error) {
-
-	return store.InsertEntry(ctx, entry)
-}
-
-func (store *UniversalRedisLuaStore) FindEntry(ctx context.Context, fullpath util.FullPath) (entry *filer.Entry, err error) {
-
-	data, err := store.Client.Get(ctx, store.getKey(string(fullpath))).Result()
-	if err == redis.Nil {
-		return nil, filer_pb.ErrNotFound
-	}
-
-	if err != nil {
-		return nil, fmt.Errorf("get %s : %v", fullpath, err)
-	}
-
-	entry = &filer.Entry{
-		FullPath: fullpath,
-	}
-	err = entry.DecodeAttributesAndChunks(util.MaybeDecompressData([]byte(data)))
-	if err != nil {
-		return entry, fmt.Errorf("decode %s : %v", entry.FullPath, err)
-	}
-
-	return entry, nil
-}
-
-func (store *UniversalRedisLuaStore) DeleteEntry(ctx context.Context, fullpath util.FullPath) (err error) {
-
-	dir, name := fullpath.DirAndName()
-
-	err = stored_procedure.DeleteEntryScript.Run(ctx, store.Client,
-		[]string{store.getKey(string(fullpath)), store.getKey(genDirectoryListKey(string(fullpath))), store.getKey(genDirectoryListKey(dir))},
-		store.isSuperLargeDirectory(dir), name).Err()
-
-	if err != nil {
-		return fmt.Errorf("DeleteEntry %s : %v", fullpath, err)
-	}
-
-	return nil
-}
-
-func (store *UniversalRedisLuaStore) DeleteFolderChildren(ctx context.Context, fullpath util.FullPath) (err error) {
-
-	if store.isSuperLargeDirectory(string(fullpath)) {
-		return nil
-	}
-
-	err = stored_procedure.DeleteFolderChildrenScript.Run(ctx, store.Client,
-		[]string{store.getKey(string(fullpath))}).Err()
-
-	if err != nil {
-		return fmt.Errorf("DeleteFolderChildren %s : %v", fullpath, err)
-	}
-
-	return nil
-}
-
-func (store *UniversalRedisLuaStore) ListDirectoryPrefixedEntries(ctx context.Context, dirPath util.FullPath, startFileName string, includeStartFile bool, limit int64, prefix string, eachEntryFunc filer.ListEachEntryFunc) (lastFileName string, err error) {
-	return lastFileName, filer.ErrUnsupportedListDirectoryPrefixed
-}
-
-func (store *UniversalRedisLuaStore) ListDirectoryEntries(ctx context.Context, dirPath util.FullPath, startFileName string, includeStartFile bool, limit int64, eachEntryFunc filer.ListEachEntryFunc) (lastFileName string, err error) {
-
-	dirListKey := store.getKey(genDirectoryListKey(string(dirPath)))
-
-	min := "-"
-	if startFileName != "" {
-		if includeStartFile {
-			min = "[" + startFileName
-		} else {
-			min = "(" + startFileName
-		}
-	}
-
-	members, err := store.Client.ZRangeByLex(ctx, dirListKey, &redis.ZRangeBy{
-		Min:    min,
-		Max:    "+",
-		Offset: 0,
-		Count:  limit,
-	}).Result()
-	if err != nil {
-		return lastFileName, fmt.Errorf("list %s : %v", dirPath, err)
-	}
-
-	// fetch entry meta
-	for _, fileName := range members {
-		path := util.NewFullPath(string(dirPath), fileName)
-		entry, err := store.FindEntry(ctx, path)
-		lastFileName = fileName
-		if err != nil {
-			glog.V(0).InfofCtx(ctx, "list %s : %v", path, err)
-			if err == filer_pb.ErrNotFound {
-				continue
-			}
-		} else {
-			if entry.TtlSec > 0 {
-				if entry.Attr.Crtime.Add(time.Duration(entry.TtlSec) * time.Second).Before(time.Now()) {
-					store.DeleteEntry(ctx, path)
-					continue
-				}
-			}
-
-			resEachEntryFunc, resEachEntryFuncErr := eachEntryFunc(entry)
-			if resEachEntryFuncErr != nil {
-				err = fmt.Errorf("failed to process eachEntryFunc: %w", resEachEntryFuncErr)
-				break
-			}
-
-			if !resEachEntryFunc {
-				break
-			}
-		}
-	}
-
-	return lastFileName, err
-}
-
-func genDirectoryListKey(dir string) (dirList string) {
-	return dir + DIR_LIST_MARKER
-}
-
-func (store *UniversalRedisLuaStore) Shutdown() {
-	store.Client.Close()
-}
@@ -1,42 +0,0 @@
-package redis_lua
-
-import (
-	"context"
-	"fmt"
-
-	"github.com/redis/go-redis/v9"
-	"github.com/seaweedfs/seaweedfs/weed/filer"
-)
-
-func (store *UniversalRedisLuaStore) KvPut(ctx context.Context, key []byte, value []byte) (err error) {
-
-	_, err = store.Client.Set(ctx, string(key), value, 0).Result()
-
-	if err != nil {
-		return fmt.Errorf("kv put: %w", err)
-	}
-
-	return nil
-}
-
-func (store *UniversalRedisLuaStore) KvGet(ctx context.Context, key []byte) (value []byte, err error) {
-
-	data, err := store.Client.Get(ctx, string(key)).Result()
-
-	if err == redis.Nil {
-		return nil, filer.ErrKvNotFound
-	}
-
-	return []byte(data), err
-}
-
-func (store *UniversalRedisLuaStore) KvDelete(ctx context.Context, key []byte) (err error) {
-
-	_, err = store.Client.Del(ctx, string(key)).Result()
-
-	if err != nil {
-		return fmt.Errorf("kv delete: %w", err)
-	}
-
-	return nil
-}
@@ -102,10 +102,6 @@ func PrepareStreamContent(masterClient wdclient.HasLookupFileIdFunction, jwtFunc

 type VolumeServerJwtFunction func(fileId string) string

-func noJwtFunc(string) string {
-	return ""
-}
-
 type CacheInvalidator interface {
 	InvalidateCache(fileId string)
 }
@@ -249,6 +245,66 @@ func PrepareStreamContentWithThrottler(ctx context.Context, masterClient wdclien
 	}, nil
 }

+// PrepareStreamContentWithPrefetch is like PrepareStreamContentWithThrottler but uses
+// concurrent chunk prefetching to overlap network I/O. When prefetchAhead > 1, fetch
+// goroutines establish HTTP connections to volume servers ahead of time, streaming data
+// through io.Pipe with minimal memory overhead.
+//
+// prefetchAhead controls the number of chunks fetched concurrently:
+//   - 0 or 1: falls back to sequential fetching (same as PrepareStreamContentWithThrottler)
+//   - 2+: uses pipe-based prefetch pipeline with that many concurrent fetches
+func PrepareStreamContentWithPrefetch(ctx context.Context, masterClient wdclient.HasLookupFileIdFunction, jwtFunc VolumeServerJwtFunction, chunks []*filer_pb.FileChunk, offset int64, size int64, downloadMaxBytesPs int64, prefetchAhead int) (DoStreamContent, error) {
+	if prefetchAhead <= 1 {
+		return PrepareStreamContentWithThrottler(ctx, masterClient, jwtFunc, chunks, offset, size, downloadMaxBytesPs)
+	}
+
+	glog.V(4).InfofCtx(ctx, "prepare to stream content with prefetch=%d for chunks: %d", prefetchAhead, len(chunks))
+	chunkViews := ViewFromChunks(ctx, masterClient.GetLookupFileIdFunction(), chunks, offset, size)
+
+	fileId2Url := make(map[string][]string)
+
+	for x := chunkViews.Front(); x != nil; x = x.Next {
+		chunkView := x.Value
+		var urlStrings []string
+		var err error
+		for _, backoff := range getLookupFileIdBackoffSchedule {
+			if err := ctx.Err(); err != nil {
+				return nil, err
+			}
+			urlStrings, err = masterClient.GetLookupFileIdFunction()(ctx, chunkView.FileId)
+			if err == nil && len(urlStrings) > 0 {
+				break
+			}
+			if err := ctx.Err(); err != nil {
+				return nil, err
+			}
+			glog.V(4).InfofCtx(ctx, "waiting for chunk: %s", chunkView.FileId)
+			timer := time.NewTimer(backoff)
+			select {
+			case <-ctx.Done():
+				if !timer.Stop() {
+					<-timer.C
+				}
+				return nil, ctx.Err()
+			case <-timer.C:
+			}
+		}
+		if err != nil {
+			glog.V(1).InfofCtx(ctx, "operation LookupFileId %s failed, err: %v", chunkView.FileId, err)
+			return nil, err
+		} else if len(urlStrings) == 0 {
+			errUrlNotFound := fmt.Errorf("operation LookupFileId %s failed, err: urls not found", chunkView.FileId)
+			glog.ErrorCtx(ctx, errUrlNotFound)
+			return nil, errUrlNotFound
+		}
+		fileId2Url[chunkView.FileId] = urlStrings
+	}
+
+	return func(writer io.Writer) error {
+		return streamChunksPrefetched(ctx, writer, chunkViews, fileId2Url, jwtFunc, masterClient, offset, size, downloadMaxBytesPs, prefetchAhead)
+	}, nil
+}
+
 func StreamContent(masterClient wdclient.HasLookupFileIdFunction, writer io.Writer, chunks []*filer_pb.FileChunk, offset int64, size int64) error {
 	streamFn, err := PrepareStreamContent(masterClient, JwtForVolumeServer, chunks, offset, size)
 	if err != nil {
@@ -276,33 +332,6 @@ func writeZero(w io.Writer, size int64) (err error) {
 	return
 }

-func ReadAll(ctx context.Context, buffer []byte, masterClient *wdclient.MasterClient, chunks []*filer_pb.FileChunk) error {
-
-	lookupFileIdFn := func(ctx context.Context, fileId string) (targetUrls []string, err error) {
-		return masterClient.LookupFileId(ctx, fileId)
-	}
-
-	chunkViews := ViewFromChunks(ctx, lookupFileIdFn, chunks, 0, int64(len(buffer)))
-
-	idx := 0
-
-	for x := chunkViews.Front(); x != nil; x = x.Next {
-		chunkView := x.Value
-		urlStrings, err := lookupFileIdFn(ctx, chunkView.FileId)
-		if err != nil {
-			glog.V(1).InfofCtx(ctx, "operation LookupFileId %s failed, err: %v", chunkView.FileId, err)
-			return err
-		}
-
-		n, err := util_http.RetriedFetchChunkData(ctx, buffer[idx:idx+int(chunkView.ViewSize)], urlStrings, chunkView.CipherKey, chunkView.IsGzipped, chunkView.IsFullChunk(), chunkView.OffsetInChunk, chunkView.FileId)
-		if err != nil {
-			return err
-		}
-		idx += n
-	}
-	return nil
-}
-
 // ----------------  ChunkStreamReader ----------------------------------
 type ChunkStreamReader struct {
 	head         *Interval[*ChunkView]
@@ -0,0 +1,317 @@
+package filer
+
+import (
+	"bytes"
+	"context"
+	"fmt"
+	"io"
+	"math/rand"
+	"net/http"
+	"net/http/httptest"
+	"os"
+	"strings"
+	"sync"
+	"testing"
+	"time"
+
+	util_http "github.com/seaweedfs/seaweedfs/weed/util/http"
+	"github.com/seaweedfs/seaweedfs/weed/pb/filer_pb"
+	"github.com/seaweedfs/seaweedfs/weed/wdclient"
+)
+
+func TestMain(m *testing.M) {
+	// Initialize the global HTTP client required by ReadUrlAsStream
+	util_http.InitGlobalHttpClient()
+	os.Exit(m.Run())
+}
+
+// mockMasterClientForBenchmark implements HasLookupFileIdFunction and CacheInvalidator
+type mockMasterClientForBenchmark struct {
+	urls map[string][]string
+}
+
+func (m *mockMasterClientForBenchmark) GetLookupFileIdFunction() wdclient.LookupFileIdFunctionType {
+	return func(ctx context.Context, fileId string) ([]string, error) {
+		if urls, ok := m.urls[fileId]; ok {
+			return urls, nil
+		}
+		return nil, fmt.Errorf("fileId %s not found", fileId)
+	}
+}
+
+func (m *mockMasterClientForBenchmark) InvalidateCache(fileId string) {}
+
+// noopJwtFunc returns empty JWT for testing
+func noopJwtFunc(fileId string) string {
+	return ""
+}
+
+// createMockVolumeServer creates an httptest server that serves chunk data
+// with configurable per-request latency to simulate network conditions.
+// The latency is applied once per request (simulating RTT), not per byte.
+func createMockVolumeServer(chunkData map[string][]byte, latency time.Duration) *httptest.Server {
+	var mu sync.RWMutex
+	return httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		// Simulate network latency (connection setup + RTT)
+		if latency > 0 {
+			time.Sleep(latency)
+		}
+
+		// Extract fileId from path (e.g., "/1,abc123")
+		path := r.URL.Path
+		if strings.HasPrefix(path, "/") {
+			path = path[1:]
+		}
+
+		mu.RLock()
+		data, ok := chunkData[path]
+		mu.RUnlock()
+		if !ok {
+			http.Error(w, "not found", http.StatusNotFound)
+			return
+		}
+
+		// Handle Range header
+		rangeHeader := r.Header.Get("Range")
+		if rangeHeader != "" {
+			var start, end int64
+			fmt.Sscanf(rangeHeader, "bytes=%d-%d", &start, &end)
+			if start >= 0 && end < int64(len(data)) && start <= end {
+				w.Header().Set("Content-Length", fmt.Sprintf("%d", end-start+1))
+				w.WriteHeader(http.StatusPartialContent)
+				w.Write(data[start : end+1])
+				return
+			}
+		}
+
+		w.Header().Set("Content-Length", fmt.Sprintf("%d", len(data)))
+		w.WriteHeader(http.StatusOK)
+		w.Write(data)
+	}))
+}
+
+// benchmarkConfig holds parameters for a single benchmark scenario
+type benchmarkConfig struct {
+	numChunks  int
+	chunkSize  int
+	latency    time.Duration
+	prefetch   int // 0 = sequential
+}
+
+func (c benchmarkConfig) name() string {
+	name := fmt.Sprintf("chunks=%d/size=%dKB/latency=%dms",
+		c.numChunks, c.chunkSize/1024, c.latency.Milliseconds())
+	if c.prefetch > 0 {
+		name += fmt.Sprintf("/prefetch=%d", c.prefetch)
+	}
+	return name
+}
+
+// setupBenchmark creates mock infrastructure and returns chunks, master client, and cleanup func
+func setupBenchmark(b *testing.B, cfg benchmarkConfig) ([]*filer_pb.FileChunk, *mockMasterClientForBenchmark, func()) {
+	b.Helper()
+
+	// Generate random chunk data
+	chunkData := make(map[string][]byte, cfg.numChunks)
+	chunks := make([]*filer_pb.FileChunk, cfg.numChunks)
+
+	for i := 0; i < cfg.numChunks; i++ {
+		fileId := fmt.Sprintf("1,%x", i)
+		data := make([]byte, cfg.chunkSize)
+		rand.Read(data)
+		chunkData[fileId] = data
+
+		chunks[i] = &filer_pb.FileChunk{
+			FileId:       fileId,
+			Offset:       int64(i * cfg.chunkSize),
+			Size:         uint64(cfg.chunkSize),
+			ModifiedTsNs: int64(i),
+			Fid:          &filer_pb.FileId{FileKey: uint64(i)},
+		}
+	}
+
+	// Start mock volume server
+	server := createMockVolumeServer(chunkData, cfg.latency)
+
+	// Build URL map
+	urls := make(map[string][]string, cfg.numChunks)
+	for i := 0; i < cfg.numChunks; i++ {
+		fileId := fmt.Sprintf("1,%x", i)
+		urls[fileId] = []string{server.URL + "/" + fileId}
+	}
+
+	masterClient := &mockMasterClientForBenchmark{urls: urls}
+	cleanup := func() { server.Close() }
+
+	return chunks, masterClient, cleanup
+}
+
+// runSequentialBenchmark runs the current sequential streaming path
+func runSequentialBenchmark(b *testing.B, cfg benchmarkConfig) {
+	chunks, masterClient, cleanup := setupBenchmark(b, cfg)
+	defer cleanup()
+
+	totalSize := int64(cfg.numChunks * cfg.chunkSize)
+
+	b.ResetTimer()
+	b.SetBytes(totalSize)
+
+	for i := 0; i < b.N; i++ {
+		streamFn, err := PrepareStreamContentWithThrottler(
+			context.Background(),
+			masterClient,
+			noopJwtFunc,
+			chunks,
+			0,
+			totalSize,
+			0, // no throttle
+		)
+		if err != nil {
+			b.Fatal(err)
+		}
+		if err := streamFn(io.Discard); err != nil {
+			b.Fatal(err)
+		}
+	}
+}
+
+// BenchmarkStreamSequential benchmarks the current sequential streaming path.
+// This provides the BEFORE baseline for comparison.
+func BenchmarkStreamSequential(b *testing.B) {
+	configs := []benchmarkConfig{
+		// Pure throughput (no latency)
+		{numChunks: 16, chunkSize: 64 * 1024, latency: 0},
+		{numChunks: 64, chunkSize: 64 * 1024, latency: 0},
+		// Moderate latency — shows RTT gap overhead
+		{numChunks: 16, chunkSize: 64 * 1024, latency: 5 * time.Millisecond},
+		{numChunks: 64, chunkSize: 64 * 1024, latency: 5 * time.Millisecond},
+		// High latency — significant RTT overhead
+		{numChunks: 16, chunkSize: 64 * 1024, latency: 20 * time.Millisecond},
+		{numChunks: 64, chunkSize: 64 * 1024, latency: 10 * time.Millisecond},
+	}
+
+	for _, cfg := range configs {
+		b.Run(cfg.name(), func(b *testing.B) {
+			runSequentialBenchmark(b, cfg)
+		})
+	}
+}
+
+// BenchmarkStreamSequentialVerify is a quick functional test that the benchmark
+// infrastructure works correctly — ensures data integrity through the pipeline.
+func BenchmarkStreamSequentialVerify(b *testing.B) {
+	cfg := benchmarkConfig{numChunks: 4, chunkSize: 1024, latency: 0}
+	chunks, masterClient, cleanup := setupBenchmark(b, cfg)
+	defer cleanup()
+
+	totalSize := int64(cfg.numChunks * cfg.chunkSize)
+
+	streamFn, err := PrepareStreamContentWithThrottler(
+		context.Background(),
+		masterClient,
+		noopJwtFunc,
+		chunks,
+		0,
+		totalSize,
+		0,
+	)
+	if err != nil {
+		b.Fatal(err)
+	}
+
+	var buf bytes.Buffer
+	if err := streamFn(&buf); err != nil {
+		b.Fatal(err)
+	}
+
+	if buf.Len() != int(totalSize) {
+		b.Fatalf("expected %d bytes, got %d", totalSize, buf.Len())
+	}
+}
+
+// runPrefetchBenchmark runs the new prefetch streaming path
+func runPrefetchBenchmark(b *testing.B, cfg benchmarkConfig) {
+	chunks, masterClient, cleanup := setupBenchmark(b, cfg)
+	defer cleanup()
+
+	totalSize := int64(cfg.numChunks * cfg.chunkSize)
+
+	b.ResetTimer()
+	b.SetBytes(totalSize)
+
+	for i := 0; i < b.N; i++ {
+		streamFn, err := PrepareStreamContentWithPrefetch(
+			context.Background(),
+			masterClient,
+			noopJwtFunc,
+			chunks,
+			0,
+			totalSize,
+			0, // no throttle
+			cfg.prefetch,
+		)
+		if err != nil {
+			b.Fatal(err)
+		}
+		if err := streamFn(io.Discard); err != nil {
+			b.Fatal(err)
+		}
+	}
+}
+
+// BenchmarkStreamPrefetch benchmarks the new prefetch streaming path.
+// Compare against BenchmarkStreamSequential for the AFTER measurement.
+func BenchmarkStreamPrefetch(b *testing.B) {
+	configs := []benchmarkConfig{
+		// Pure throughput (no latency) — should be similar to sequential
+		{numChunks: 16, chunkSize: 64 * 1024, latency: 0, prefetch: 4},
+		{numChunks: 64, chunkSize: 64 * 1024, latency: 0, prefetch: 4},
+		// Moderate latency — prefetch should eliminate most RTT overhead
+		{numChunks: 16, chunkSize: 64 * 1024, latency: 5 * time.Millisecond, prefetch: 4},
+		{numChunks: 64, chunkSize: 64 * 1024, latency: 5 * time.Millisecond, prefetch: 4},
+		// High latency — most benefit from prefetch
+		{numChunks: 16, chunkSize: 64 * 1024, latency: 20 * time.Millisecond, prefetch: 4},
+		{numChunks: 64, chunkSize: 64 * 1024, latency: 10 * time.Millisecond, prefetch: 4},
+		// Vary prefetch count with moderate latency
+		{numChunks: 64, chunkSize: 64 * 1024, latency: 5 * time.Millisecond, prefetch: 2},
+		{numChunks: 64, chunkSize: 64 * 1024, latency: 5 * time.Millisecond, prefetch: 8},
+	}
+
+	for _, cfg := range configs {
+		b.Run(cfg.name(), func(b *testing.B) {
+			runPrefetchBenchmark(b, cfg)
+		})
+	}
+}
+
+// BenchmarkStreamPrefetchVerify verifies data integrity through the prefetch pipeline.
+func BenchmarkStreamPrefetchVerify(b *testing.B) {
+	cfg := benchmarkConfig{numChunks: 4, chunkSize: 1024, latency: 0, prefetch: 4}
+	chunks, masterClient, cleanup := setupBenchmark(b, cfg)
+	defer cleanup()
+
+	totalSize := int64(cfg.numChunks * cfg.chunkSize)
+
+	streamFn, err := PrepareStreamContentWithPrefetch(
+		context.Background(),
+		masterClient,
+		noopJwtFunc,
+		chunks,
+		0,
+		totalSize,
+		0,
+		cfg.prefetch,
+	)
+	if err != nil {
+		b.Fatal(err)
+	}
+
+	var buf bytes.Buffer
+	if err := streamFn(&buf); err != nil {
+		b.Fatal(err)
+	}
+
+	if buf.Len() != int(totalSize) {
+		b.Fatalf("expected %d bytes, got %d", totalSize, buf.Len())
+	}
+}
@@ -1,281 +0,0 @@
-package filer
-
-import (
-	"bytes"
-	"context"
-	"errors"
-	"testing"
-	"time"
-
-	"github.com/seaweedfs/seaweedfs/weed/pb/filer_pb"
-	"github.com/seaweedfs/seaweedfs/weed/wdclient"
-)
-
-// mockMasterClient implements HasLookupFileIdFunction and CacheInvalidator
-type mockMasterClient struct {
-	lookupFunc         func(ctx context.Context, fileId string) ([]string, error)
-	invalidatedFileIds []string
-}
-
-func (m *mockMasterClient) GetLookupFileIdFunction() wdclient.LookupFileIdFunctionType {
-	return m.lookupFunc
-}
-
-func (m *mockMasterClient) InvalidateCache(fileId string) {
-	m.invalidatedFileIds = append(m.invalidatedFileIds, fileId)
-}
-
-// Test urlSlicesEqual helper function
-func TestUrlSlicesEqual(t *testing.T) {
-	tests := []struct {
-		name     string
-		a        []string
-		b        []string
-		expected bool
-	}{
-		{
-			name:     "identical slices",
-			a:        []string{"http://server1", "http://server2"},
-			b:        []string{"http://server1", "http://server2"},
-			expected: true,
-		},
-		{
-			name:     "same URLs different order",
-			a:        []string{"http://server1", "http://server2"},
-			b:        []string{"http://server2", "http://server1"},
-			expected: true,
-		},
-		{
-			name:     "different URLs",
-			a:        []string{"http://server1", "http://server2"},
-			b:        []string{"http://server1", "http://server3"},
-			expected: false,
-		},
-		{
-			name:     "different lengths",
-			a:        []string{"http://server1"},
-			b:        []string{"http://server1", "http://server2"},
-			expected: false,
-		},
-		{
-			name:     "empty slices",
-			a:        []string{},
-			b:        []string{},
-			expected: true,
-		},
-		{
-			name:     "duplicates in both",
-			a:        []string{"http://server1", "http://server1"},
-			b:        []string{"http://server1", "http://server1"},
-			expected: true,
-		},
-		{
-			name:     "different duplicate counts",
-			a:        []string{"http://server1", "http://server1"},
-			b:        []string{"http://server1", "http://server2"},
-			expected: false,
-		},
-	}
-
-	for _, tt := range tests {
-		t.Run(tt.name, func(t *testing.T) {
-			result := urlSlicesEqual(tt.a, tt.b)
-			if result != tt.expected {
-				t.Errorf("urlSlicesEqual(%v, %v) = %v; want %v", tt.a, tt.b, result, tt.expected)
-			}
-		})
-	}
-}
-
-// Test cache invalidation when read fails
-func TestStreamContentWithCacheInvalidation(t *testing.T) {
-	ctx := context.Background()
-	fileId := "3,01234567890"
-
-	callCount := 0
-	oldUrls := []string{"http://failed-server:8080"}
-	newUrls := []string{"http://working-server:8080"}
-
-	mock := &mockMasterClient{
-		lookupFunc: func(ctx context.Context, fid string) ([]string, error) {
-			callCount++
-			if callCount == 1 {
-				// First call returns failing server
-				return oldUrls, nil
-			}
-			// After invalidation, return working server
-			return newUrls, nil
-		},
-	}
-
-	// Create a simple chunk
-	chunks := []*filer_pb.FileChunk{
-		{
-			FileId: fileId,
-			Offset: 0,
-			Size:   10,
-		},
-	}
-
-	streamFn, err := PrepareStreamContentWithThrottler(ctx, mock, noJwtFunc, chunks, 0, 10, 0)
-	if err != nil {
-		t.Fatalf("PrepareStreamContentWithThrottler failed: %v", err)
-	}
-
-	// Note: This test can't fully execute streamFn because it would require actual HTTP servers
-	// However, we can verify the setup was created correctly
-	if streamFn == nil {
-		t.Fatal("Expected non-nil stream function")
-	}
-
-	// Verify the lookup was called
-	if callCount != 1 {
-		t.Errorf("Expected 1 lookup call, got %d", callCount)
-	}
-}
-
-// Test that InvalidateCache is called on read failure
-func TestCacheInvalidationInterface(t *testing.T) {
-	mock := &mockMasterClient{
-		lookupFunc: func(ctx context.Context, fileId string) ([]string, error) {
-			return []string{"http://server:8080"}, nil
-		},
-	}
-
-	fileId := "3,test123"
-
-	// Simulate invalidation
-	if invalidator, ok := interface{}(mock).(CacheInvalidator); ok {
-		invalidator.InvalidateCache(fileId)
-	} else {
-		t.Fatal("mockMasterClient should implement CacheInvalidator")
-	}
-
-	// Check that the file ID was recorded as invalidated
-	if len(mock.invalidatedFileIds) != 1 {
-		t.Fatalf("Expected 1 invalidated file ID, got %d", len(mock.invalidatedFileIds))
-	}
-	if mock.invalidatedFileIds[0] != fileId {
-		t.Errorf("Expected invalidated file ID %s, got %s", fileId, mock.invalidatedFileIds[0])
-	}
-}
-
-// Test retry logic doesn't retry with same URLs
-func TestRetryLogicSkipsSameUrls(t *testing.T) {
-	// This test verifies that the urlSlicesEqual check prevents infinite retries
-	sameUrls := []string{"http://server1:8080", "http://server2:8080"}
-	differentUrls := []string{"http://server3:8080", "http://server4:8080"}
-
-	// Same URLs should return true (and thus skip retry)
-	if !urlSlicesEqual(sameUrls, sameUrls) {
-		t.Error("Expected same URLs to be equal")
-	}
-
-	// Different URLs should return false (and thus allow retry)
-	if urlSlicesEqual(sameUrls, differentUrls) {
-		t.Error("Expected different URLs to not be equal")
-	}
-}
-
-func TestCanceledStreamSkipsCacheInvalidation(t *testing.T) {
-	ctx, cancel := context.WithCancel(context.Background())
-	fileId := "3,canceled"
-
-	mock := &mockMasterClient{
-		lookupFunc: func(ctx context.Context, fid string) ([]string, error) {
-			return []string{"http://server:8080"}, nil
-		},
-	}
-
-	chunks := []*filer_pb.FileChunk{
-		{
-			FileId: fileId,
-			Offset: 0,
-			Size:   10,
-		},
-	}
-
-	streamFn, err := PrepareStreamContentWithThrottler(ctx, mock, noJwtFunc, chunks, 0, 10, 0)
-	if err != nil {
-		t.Fatalf("PrepareStreamContentWithThrottler failed: %v", err)
-	}
-
-	cancel()
-
-	err = streamFn(&bytes.Buffer{})
-	if err != context.Canceled {
-		t.Fatalf("expected context.Canceled, got %v", err)
-	}
-	if len(mock.invalidatedFileIds) != 0 {
-		t.Fatalf("expected no cache invalidation on cancellation, got %v", mock.invalidatedFileIds)
-	}
-}
-
-func TestPrepareStreamContentSkipsLookupWhenContextAlreadyCanceled(t *testing.T) {
-	oldSchedule := getLookupFileIdBackoffSchedule
-	getLookupFileIdBackoffSchedule = []time.Duration{time.Millisecond}
-	t.Cleanup(func() {
-		getLookupFileIdBackoffSchedule = oldSchedule
-	})
-
-	ctx, cancel := context.WithCancel(context.Background())
-	cancel()
-
-	lookupCalls := 0
-	mock := &mockMasterClient{
-		lookupFunc: func(ctx context.Context, fileId string) ([]string, error) {
-			lookupCalls++
-			return nil, errors.New("lookup should not run")
-		},
-	}
-
-	chunks := []*filer_pb.FileChunk{
-		{
-			FileId: "3,precanceled",
-			Offset: 0,
-			Size:   10,
-		},
-	}
-
-	_, err := PrepareStreamContentWithThrottler(ctx, mock, noJwtFunc, chunks, 0, 10, 0)
-	if !errors.Is(err, context.Canceled) {
-		t.Fatalf("expected context.Canceled, got %v", err)
-	}
-	if lookupCalls != 0 {
-		t.Fatalf("expected no lookup calls after cancellation, got %d", lookupCalls)
-	}
-}
-
-func TestPrepareStreamContentStopsLookupRetriesAfterContextCancellation(t *testing.T) {
-	oldSchedule := getLookupFileIdBackoffSchedule
-	getLookupFileIdBackoffSchedule = []time.Duration{time.Millisecond, time.Millisecond, time.Millisecond}
-	t.Cleanup(func() {
-		getLookupFileIdBackoffSchedule = oldSchedule
-	})
-
-	ctx, cancel := context.WithCancel(context.Background())
-	lookupCalls := 0
-	mock := &mockMasterClient{
-		lookupFunc: func(ctx context.Context, fileId string) ([]string, error) {
-			lookupCalls++
-			cancel()
-			return nil, context.Canceled
-		},
-	}
-
-	chunks := []*filer_pb.FileChunk{
-		{
-			FileId: "3,cancel-during-lookup",
-			Offset: 0,
-			Size:   10,
-		},
-	}
-
-	_, err := PrepareStreamContentWithThrottler(ctx, mock, noJwtFunc, chunks, 0, 10, 0)
-	if !errors.Is(err, context.Canceled) {
-		t.Fatalf("expected context.Canceled, got %v", err)
-	}
-	if lookupCalls != 1 {
-		t.Fatalf("expected lookup retries to stop after cancellation, got %d calls", lookupCalls)
-	}
-}
@@ -0,0 +1,274 @@
+package filer
+
+import (
+	"context"
+	"fmt"
+	"io"
+	"sync"
+	"time"
+
+	"github.com/seaweedfs/seaweedfs/weed/glog"
+	"github.com/seaweedfs/seaweedfs/weed/stats"
+	"github.com/seaweedfs/seaweedfs/weed/util"
+	"github.com/seaweedfs/seaweedfs/weed/util/mem"
+	"github.com/seaweedfs/seaweedfs/weed/wdclient"
+)
+
+// chunkPipeResult represents a prefetched chunk streaming through a pipe.
+// The fetch goroutine writes data into the pipeWriter; the consumer reads from pipeReader.
+type chunkPipeResult struct {
+	chunkView  *ChunkView
+	reader     *io.PipeReader
+	fetchErr   error         // final error from fetch goroutine
+	written    int64         // bytes written by fetch goroutine
+	done       chan struct{} // closed when fetch goroutine finishes
+	urlStrings []string      // snapshot of URLs at dispatch time (for retry logic)
+}
+
+// streamChunksPrefetched streams chunks with concurrent prefetch using io.Pipe.
+//
+// For each chunk in file order, a goroutine is launched (bounded by a semaphore)
+// that establishes an HTTP connection to the volume server and streams data through
+// an io.Pipe. The consumer reads from pipes in order, writing to the response.
+//
+// Memory usage is minimal: pipes are synchronous (no buffering), and only one
+// reusable copy buffer is allocated for the consumer.
+func streamChunksPrefetched(
+	ctx context.Context,
+	writer io.Writer,
+	chunkViews *IntervalList[*ChunkView],
+	fileId2Url map[string][]string,
+	jwtFunc VolumeServerJwtFunction,
+	masterClient wdclient.HasLookupFileIdFunction,
+	offset int64,
+	size int64,
+	downloadMaxBytesPs int64,
+	prefetchAhead int,
+) error {
+	downloadThrottler := util.NewWriteThrottler(downloadMaxBytesPs)
+
+	// Create a local cancellable context so the consumer can stop the producer
+	// and all in-flight fetch goroutines on error (e.g., client disconnect).
+	localCtx, localCancel := context.WithCancel(ctx)
+	defer localCancel()
+
+	// Ordered channel: one entry per chunk, in file order.
+	// Capacity = prefetchAhead so the producer can run ahead.
+	// Uses pointer to avoid copying the struct while fetch goroutines write to it.
+	results := make(chan *chunkPipeResult, prefetchAhead)
+
+	// Semaphore to limit concurrent fetch goroutines (and thus HTTP connections).
+	sem := make(chan struct{}, prefetchAhead)
+
+	// Producer: walks chunk list, launches fetch goroutines, sends results in order.
+	// The producer only reads from fileId2Url (populated before streaming starts),
+	// so there is no concurrent map access — the consumer never writes to it.
+	var producerWg sync.WaitGroup
+	producerWg.Add(1)
+	go func() {
+		defer producerWg.Done()
+		defer close(results)
+
+		for x := chunkViews.Front(); x != nil; x = x.Next {
+			chunkView := x.Value
+
+			// Check context before starting new fetch
+			select {
+			case <-localCtx.Done():
+				return
+			default:
+			}
+
+			// Acquire semaphore slot (bounds concurrent HTTP connections)
+			select {
+			case sem <- struct{}{}:
+			case <-localCtx.Done():
+				return
+			}
+
+			pr, pw := io.Pipe()
+			urlStrings := fileId2Url[chunkView.FileId]
+			jwt := jwtFunc(chunkView.FileId)
+
+			result := &chunkPipeResult{
+				chunkView:  chunkView,
+				reader:     pr,
+				done:       make(chan struct{}),
+				urlStrings: urlStrings,
+			}
+
+			// Launch fetch goroutine
+			go func(cv *ChunkView, urls []string, jwt string, pw *io.PipeWriter, res *chunkPipeResult) {
+				defer func() { <-sem }() // release semaphore
+				defer close(res.done)
+
+				written, err := retriedStreamFetchChunkData(
+					localCtx, pw, urls, jwt,
+					cv.CipherKey, cv.IsGzipped, cv.IsFullChunk(),
+					cv.OffsetInChunk, int(cv.ViewSize),
+				)
+				res.written = written
+				res.fetchErr = err
+
+				if err != nil {
+					pw.CloseWithError(err)
+				} else {
+					pw.Close()
+				}
+			}(chunkView, urlStrings, jwt, pw, result)
+
+			// Send result to consumer (blocks if channel full, back-pressuring producer)
+			select {
+			case results <- result:
+			case <-localCtx.Done():
+				// Consumer gone; close the pipe and wait for the fetch goroutine
+				// to finish so we don't leak it (this result was never sent to
+				// the channel, so the drain loop won't handle it).
+				pr.Close()
+				<-result.done
+				return
+			}
+		}
+	}()
+
+	// Consumer: reads from results channel in order, writes to response writer.
+	// Use the SeaweedFS memory pool for the copy buffer to reduce GC pressure.
+	copyBuf := mem.Allocate(256 * 1024)
+	defer mem.Free(copyBuf)
+	remaining := size
+
+	var consumeErr error
+	for result := range results {
+		chunkView := result.chunkView
+
+		// Handle gap before this chunk (zero-fill)
+		if offset < chunkView.ViewOffset {
+			gap := chunkView.ViewOffset - offset
+			remaining -= gap
+			glog.V(4).InfofCtx(ctx, "prefetch zero [%d,%d)", offset, chunkView.ViewOffset)
+			if err := writeZero(writer, gap); err != nil {
+				consumeErr = fmt.Errorf("write zero [%d,%d): %w", offset, chunkView.ViewOffset, err)
+				result.reader.Close()
+				break
+			}
+			offset = chunkView.ViewOffset
+		}
+
+		// Stream chunk data from pipe to response
+		start := time.Now()
+		_, copyErr := io.CopyBuffer(writer, result.reader, copyBuf)
+		result.reader.Close()
+
+		// Wait for fetch goroutine to finish to get final error
+		<-result.done
+
+		// Determine the effective error
+		err := copyErr
+		if err == nil && result.fetchErr != nil && result.written == 0 {
+			err = result.fetchErr
+		}
+
+		// If the fetcher itself failed before writing any data, try cache invalidation
+		// + re-fetch (same as sequential path stream.go:197). We check result.fetchErr
+		// and result.written (not copied) to avoid wrongly retrying when the fetch
+		// succeeded but the response writer failed on the first write.
+		if result.fetchErr != nil && result.written == 0 {
+			if err := localCtx.Err(); err != nil {
+				consumeErr = err
+				break
+			}
+			retryErr := retryWithCacheInvalidation(localCtx, writer, chunkView, result.urlStrings, jwtFunc, masterClient)
+			if retryErr != nil {
+				stats.FilerHandlerCounter.WithLabelValues("chunkDownloadError").Inc()
+				consumeErr = fmt.Errorf("read chunk: %w", retryErr)
+				break
+			}
+			// Retry succeeded
+			err = nil
+		} else if err != nil {
+			if localCtx.Err() != nil {
+				consumeErr = localCtx.Err()
+			} else {
+				stats.FilerHandlerCounter.WithLabelValues("chunkDownloadError").Inc()
+				consumeErr = fmt.Errorf("read chunk: %w", err)
+			}
+			break
+		}
+
+		offset += int64(chunkView.ViewSize)
+		remaining -= int64(chunkView.ViewSize)
+		stats.FilerRequestHistogram.WithLabelValues("chunkDownload").Observe(time.Since(start).Seconds())
+		stats.FilerHandlerCounter.WithLabelValues("chunkDownload").Inc()
+		downloadThrottler.MaybeSlowdown(int64(chunkView.ViewSize))
+	}
+
+	// Cancel the local context to stop the producer and any in-flight fetchers early.
+	// This ensures goroutines don't linger after the consumer exits (e.g., on write error).
+	localCancel()
+
+	// Drain remaining results to close pipes and unblock fetch goroutines
+	for result := range results {
+		result.reader.Close()
+		<-result.done
+	}
+
+	// Wait for producer to finish
+	producerWg.Wait()
+
+	if consumeErr != nil {
+		return consumeErr
+	}
+
+	// Handle trailing zero-fill
+	if remaining > 0 {
+		glog.V(4).InfofCtx(ctx, "prefetch zero [%d,%d)", offset, offset+remaining)
+		if err := writeZero(writer, remaining); err != nil {
+			return fmt.Errorf("write zero [%d,%d): %w", offset, offset+remaining, err)
+		}
+	}
+
+	return nil
+}
+
+// retryWithCacheInvalidation attempts to re-fetch a chunk after invalidating the URL cache.
+// This mirrors the retry logic in PrepareStreamContentWithThrottler's sequential path.
+func retryWithCacheInvalidation(
+	ctx context.Context,
+	writer io.Writer,
+	chunkView *ChunkView,
+	oldUrlStrings []string,
+	jwtFunc VolumeServerJwtFunction,
+	masterClient wdclient.HasLookupFileIdFunction,
+) error {
+	invalidator, ok := masterClient.(CacheInvalidator)
+	if !ok {
+		return fmt.Errorf("read chunk %s failed and no cache invalidator available", chunkView.FileId)
+	}
+
+	glog.V(0).InfofCtx(ctx, "prefetch read chunk %s failed, invalidating cache and retrying", chunkView.FileId)
+	invalidator.InvalidateCache(chunkView.FileId)
+
+	newUrlStrings, lookupErr := masterClient.GetLookupFileIdFunction()(ctx, chunkView.FileId)
+	if lookupErr != nil {
+		glog.WarningfCtx(ctx, "failed to re-lookup chunk %s after cache invalidation: %v", chunkView.FileId, lookupErr)
+		return fmt.Errorf("re-lookup chunk %s: %w", chunkView.FileId, lookupErr)
+	}
+	if len(newUrlStrings) == 0 {
+		glog.WarningfCtx(ctx, "re-lookup for chunk %s returned no locations, skipping retry", chunkView.FileId)
+		return fmt.Errorf("re-lookup chunk %s: no locations", chunkView.FileId)
+	}
+
+	if urlSlicesEqual(oldUrlStrings, newUrlStrings) {
+		glog.V(0).InfofCtx(ctx, "re-lookup returned same locations for chunk %s, skipping retry", chunkView.FileId)
+		return fmt.Errorf("read chunk %s failed, same locations after cache invalidation", chunkView.FileId)
+	}
+
+	glog.V(0).InfofCtx(ctx, "retrying read chunk %s with new locations: %v", chunkView.FileId, newUrlStrings)
+	jwt := jwtFunc(chunkView.FileId)
+	_, err := retriedStreamFetchChunkData(
+		ctx, writer, newUrlStrings, jwt,
+		chunkView.CipherKey, chunkView.IsGzipped, chunkView.IsFullChunk(),
+		chunkView.OffsetInChunk, int(chunkView.ViewSize),
+	)
+	return err
+}
@@ -0,0 +1,365 @@
+package filer
+
+import (
+	"bytes"
+	"context"
+	"fmt"
+	"io"
+	"math/rand"
+	"net/http"
+	"net/http/httptest"
+	"strings"
+	"sync"
+	"sync/atomic"
+	"testing"
+	"time"
+
+	"github.com/seaweedfs/seaweedfs/weed/pb/filer_pb"
+	"github.com/seaweedfs/seaweedfs/weed/wdclient"
+)
+
+// testMasterClient implements HasLookupFileIdFunction and CacheInvalidator for tests
+type testMasterClient struct {
+	urls             map[string][]string
+	invalidatedCount int32
+}
+
+func (m *testMasterClient) GetLookupFileIdFunction() wdclient.LookupFileIdFunctionType {
+	return func(ctx context.Context, fileId string) ([]string, error) {
+		if urls, ok := m.urls[fileId]; ok {
+			return urls, nil
+		}
+		return nil, fmt.Errorf("fileId %s not found", fileId)
+	}
+}
+
+func (m *testMasterClient) InvalidateCache(fileId string) {
+	atomic.AddInt32(&m.invalidatedCount, 1)
+}
+
+func noopJwt(fileId string) string { return "" }
+
+// createTestServer creates a mock volume server that serves chunk data.
+// Supports Range header for partial chunk reads (exercising OffsetInChunk paths).
+func createTestServer(chunkData map[string][]byte) *httptest.Server {
+	var mu sync.RWMutex
+	return httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		path := r.URL.Path
+		if strings.HasPrefix(path, "/") {
+			path = path[1:]
+		}
+		mu.RLock()
+		data, ok := chunkData[path]
+		mu.RUnlock()
+		if !ok {
+			http.Error(w, "not found", http.StatusNotFound)
+			return
+		}
+
+		// Handle Range header for partial chunk reads
+		rangeHeader := r.Header.Get("Range")
+		if rangeHeader != "" {
+			var start, end int64
+			if _, err := fmt.Sscanf(rangeHeader, "bytes=%d-%d", &start, &end); err == nil {
+				if start < 0 || end >= int64(len(data)) || start > end {
+					w.Header().Set("Content-Range", fmt.Sprintf("bytes */%d", len(data)))
+					w.WriteHeader(http.StatusRequestedRangeNotSatisfiable)
+					return
+				}
+				rangeData := data[start : end+1]
+				w.Header().Set("Content-Range", fmt.Sprintf("bytes %d-%d/%d", start, end, len(data)))
+				w.Header().Set("Content-Length", fmt.Sprintf("%d", len(rangeData)))
+				w.WriteHeader(http.StatusPartialContent)
+				w.Write(rangeData)
+				return
+			}
+		}
+
+		w.Header().Set("Content-Length", fmt.Sprintf("%d", len(data)))
+		w.WriteHeader(http.StatusOK)
+		w.Write(data)
+	}))
+}
+
+// makeChunksAndServer creates N chunks of given size, a mock server, and a master client
+func makeChunksAndServer(t *testing.T, numChunks, chunkSize int) ([]*filer_pb.FileChunk, *testMasterClient, map[string][]byte, func()) {
+	t.Helper()
+
+	chunkData := make(map[string][]byte, numChunks)
+	chunks := make([]*filer_pb.FileChunk, numChunks)
+
+	for i := 0; i < numChunks; i++ {
+		fileId := fmt.Sprintf("1,%x", i)
+		data := make([]byte, chunkSize)
+		rand.Read(data)
+		chunkData[fileId] = data
+
+		chunks[i] = &filer_pb.FileChunk{
+			FileId:       fileId,
+			Offset:       int64(i * chunkSize),
+			Size:         uint64(chunkSize),
+			ModifiedTsNs: int64(i),
+			Fid:          &filer_pb.FileId{FileKey: uint64(i)},
+		}
+	}
+
+	server := createTestServer(chunkData)
+	urls := make(map[string][]string, numChunks)
+	for i := 0; i < numChunks; i++ {
+		fileId := fmt.Sprintf("1,%x", i)
+		urls[fileId] = []string{server.URL + "/" + fileId}
+	}
+
+	masterClient := &testMasterClient{urls: urls}
+	return chunks, masterClient, chunkData, func() { server.Close() }
+}
+
+// TestPrefetchInOrderDelivery verifies chunks are written to the output in correct file order
+func TestPrefetchInOrderDelivery(t *testing.T) {
+	chunks, masterClient, chunkData, cleanup := makeChunksAndServer(t, 8, 4096)
+	defer cleanup()
+
+	totalSize := int64(8 * 4096)
+
+	streamFn, err := PrepareStreamContentWithPrefetch(
+		context.Background(), masterClient, noopJwt,
+		chunks, 0, totalSize, 0, 4,
+	)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	var buf bytes.Buffer
+	if err := streamFn(&buf); err != nil {
+		t.Fatal(err)
+	}
+
+	// Verify total size
+	if buf.Len() != int(totalSize) {
+		t.Fatalf("expected %d bytes, got %d", totalSize, buf.Len())
+	}
+
+	// Verify data matches chunk-by-chunk in order
+	result := buf.Bytes()
+	for i := 0; i < 8; i++ {
+		fileId := fmt.Sprintf("1,%x", i)
+		expected := chunkData[fileId]
+		got := result[i*4096 : (i+1)*4096]
+		if !bytes.Equal(expected, got) {
+			t.Fatalf("chunk %d (%s) data mismatch at offset %d", i, fileId, i*4096)
+		}
+	}
+}
+
+// TestPrefetchSingleChunk verifies the pipeline works with just one chunk
+func TestPrefetchSingleChunk(t *testing.T) {
+	chunks, masterClient, chunkData, cleanup := makeChunksAndServer(t, 1, 8192)
+	defer cleanup()
+
+	streamFn, err := PrepareStreamContentWithPrefetch(
+		context.Background(), masterClient, noopJwt,
+		chunks, 0, 8192, 0, 4,
+	)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	var buf bytes.Buffer
+	if err := streamFn(&buf); err != nil {
+		t.Fatal(err)
+	}
+
+	expected := chunkData["1,0"]
+	if !bytes.Equal(expected, buf.Bytes()) {
+		t.Fatal("single chunk data mismatch")
+	}
+}
+
+// TestPrefetchFallbackToSequential verifies prefetch=1 falls back to sequential path
+func TestPrefetchFallbackToSequential(t *testing.T) {
+	chunks, masterClient, chunkData, cleanup := makeChunksAndServer(t, 4, 1024)
+	defer cleanup()
+
+	totalSize := int64(4 * 1024)
+
+	streamFn, err := PrepareStreamContentWithPrefetch(
+		context.Background(), masterClient, noopJwt,
+		chunks, 0, totalSize, 0, 1, // prefetch=1 -> sequential
+	)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	var buf bytes.Buffer
+	if err := streamFn(&buf); err != nil {
+		t.Fatal(err)
+	}
+
+	if buf.Len() != int(totalSize) {
+		t.Fatalf("expected %d bytes, got %d", totalSize, buf.Len())
+	}
+
+	// Verify data order
+	result := buf.Bytes()
+	for i := 0; i < 4; i++ {
+		fileId := fmt.Sprintf("1,%x", i)
+		expected := chunkData[fileId]
+		got := result[i*1024 : (i+1)*1024]
+		if !bytes.Equal(expected, got) {
+			t.Fatalf("chunk %d data mismatch", i)
+		}
+	}
+}
+
+// TestPrefetchContextCancellation verifies all goroutines clean up on cancellation
+func TestPrefetchContextCancellation(t *testing.T) {
+	// Use a slow server so cancellation happens mid-stream
+	var requestCount int32
+	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		atomic.AddInt32(&requestCount, 1)
+		// Slow response
+		time.Sleep(100 * time.Millisecond)
+		w.Header().Set("Content-Length", "1024")
+		w.WriteHeader(http.StatusOK)
+		w.Write(make([]byte, 1024))
+	}))
+	defer server.Close()
+
+	numChunks := 16
+	chunks := make([]*filer_pb.FileChunk, numChunks)
+	urls := make(map[string][]string, numChunks)
+	for i := 0; i < numChunks; i++ {
+		fileId := fmt.Sprintf("1,%x", i)
+		chunks[i] = &filer_pb.FileChunk{
+			FileId: fileId, Offset: int64(i * 1024), Size: 1024,
+			ModifiedTsNs: int64(i), Fid: &filer_pb.FileId{FileKey: uint64(i)},
+		}
+		urls[fileId] = []string{server.URL + "/" + fileId}
+	}
+	masterClient := &testMasterClient{urls: urls}
+
+	// Cancel after a short time
+	ctx, cancel := context.WithTimeout(context.Background(), 50*time.Millisecond)
+	defer cancel()
+
+	streamFn, err := PrepareStreamContentWithPrefetch(
+		ctx, masterClient, noopJwt,
+		chunks, 0, int64(numChunks*1024), 0, 4,
+	)
+	if err != nil {
+		// URL resolution may fail due to cancellation — that's expected
+		return
+	}
+
+	err = streamFn(io.Discard)
+	if err == nil {
+		t.Fatal("expected error from cancelled context")
+	}
+
+	// Verify not all chunks were requested (cancellation stopped early)
+	reqs := atomic.LoadInt32(&requestCount)
+	if reqs >= int32(numChunks) {
+		t.Logf("warning: all %d chunks were requested despite cancellation (got %d)", numChunks, reqs)
+	}
+}
+
+// TestPrefetchRangeRequest verifies prefetch works with offset/size subset
+func TestPrefetchRangeRequest(t *testing.T) {
+	chunks, masterClient, chunkData, cleanup := makeChunksAndServer(t, 8, 4096)
+	defer cleanup()
+
+	// Request only chunks 2-5 (offset=8192, size=16384)
+	offset := int64(2 * 4096)
+	size := int64(4 * 4096)
+
+	streamFn, err := PrepareStreamContentWithPrefetch(
+		context.Background(), masterClient, noopJwt,
+		chunks, offset, size, 0, 4,
+	)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	var buf bytes.Buffer
+	if err := streamFn(&buf); err != nil {
+		t.Fatal(err)
+	}
+
+	if buf.Len() != int(size) {
+		t.Fatalf("expected %d bytes, got %d", size, buf.Len())
+	}
+
+	// Verify data matches chunks 2-5
+	result := buf.Bytes()
+	for i := 2; i < 6; i++ {
+		fileId := fmt.Sprintf("1,%x", i)
+		expected := chunkData[fileId]
+		start := (i - 2) * 4096
+		got := result[start : start+4096]
+		if !bytes.Equal(expected, got) {
+			t.Fatalf("chunk %d data mismatch in range request", i)
+		}
+	}
+}
+
+// TestPrefetchLargePrefetchCount verifies prefetch > numChunks is handled gracefully
+func TestPrefetchLargePrefetchCount(t *testing.T) {
+	chunks, masterClient, _, cleanup := makeChunksAndServer(t, 3, 1024)
+	defer cleanup()
+
+	totalSize := int64(3 * 1024)
+
+	// prefetch=10 but only 3 chunks — should work fine
+	streamFn, err := PrepareStreamContentWithPrefetch(
+		context.Background(), masterClient, noopJwt,
+		chunks, 0, totalSize, 0, 10,
+	)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	var buf bytes.Buffer
+	if err := streamFn(&buf); err != nil {
+		t.Fatal(err)
+	}
+
+	if buf.Len() != int(totalSize) {
+		t.Fatalf("expected %d bytes, got %d", totalSize, buf.Len())
+	}
+}
+
+// TestPrefetchConcurrentDownloads verifies multiple concurrent prefetch streams
+func TestPrefetchConcurrentDownloads(t *testing.T) {
+	chunks, masterClient, _, cleanup := makeChunksAndServer(t, 8, 2048)
+	defer cleanup()
+
+	totalSize := int64(8 * 2048)
+
+	var wg sync.WaitGroup
+	errors := make(chan error, 4)
+
+	for i := 0; i < 4; i++ {
+		wg.Add(1)
+		go func() {
+			defer wg.Done()
+			streamFn, err := PrepareStreamContentWithPrefetch(
+				context.Background(), masterClient, noopJwt,
+				chunks, 0, totalSize, 0, 4,
+			)
+			if err != nil {
+				errors <- err
+				return
+			}
+			if err := streamFn(io.Discard); err != nil {
+				errors <- err
+			}
+		}()
+	}
+
+	wg.Wait()
+	close(errors)
+
+	for err := range errors {
+		t.Fatalf("concurrent download error: %v", err)
+	}
+}
@@ -37,11 +37,6 @@ func GenerateRandomString(length int, charset string) (string, error) {
 	return string(b), nil
 }

-// GenerateAccessKeyId generates a new access key ID.
-func GenerateAccessKeyId() (string, error) {
-	return GenerateRandomString(AccessKeyIdLength, CharsetUpper)
-}
-
 // GenerateSecretAccessKey generates a new secret access key.
 func GenerateSecretAccessKey() (string, error) {
 	return GenerateRandomString(SecretAccessKeyLength, Charset)
@@ -179,11 +174,3 @@ func MapToIdentitiesAction(action string) string {
 		return ""
 	}
 }
-
-// MaskAccessKey masks an access key for logging, showing only the first 4 characters.
-func MaskAccessKey(accessKeyId string) string {
-	if len(accessKeyId) > 4 {
-		return accessKeyId[:4] + "***"
-	}
-	return accessKeyId
-}
--- a/Show More
+++ b/Show More
				`@@ -0,0 +1 @@`
				`{"sessionId":"d6574c47-eafc-4a94-9dce-f9ffea22b53c","pid":10111,"acquiredAt":1775248373916}`