filer.remote.sync: classify refused stamps by gRPC status only

A FailedPrecondition substring in an unrelated error would have been swallowed as a skipped stamp; status.FromError already unwraps.
filer: compare IF_ENTRY_EQUAL against the normalized expected entry
2026-10-09 08:05:51 +00:00 · 2026-09-23 17:07:02 -07:00 · 2026-09-23 17:07:02 -07:00 · 2026-09-23 16:34:38 -07:00 · 2026-09-23 16:34:29 -07:00 · 2026-09-23 16:11:48 -07:00
9 changed files with 363 additions and 114 deletions
@@ -20,6 +20,8 @@ import (
 	"github.com/seaweedfs/seaweedfs/weed/replication/source"
 	"github.com/seaweedfs/seaweedfs/weed/util"
 	"google.golang.org/grpc"
+	"google.golang.org/grpc/codes"
+	"google.golang.org/grpc/status"
 	"google.golang.org/protobuf/proto"
 )

@@ -417,16 +419,49 @@ func shouldSendToRemote(entry *filer_pb.Entry) bool {
 	return false
 }

+// updateLocalEntry stamps the entry an event described with its RemoteEntry.
+// The write carries IF_ENTRY_EQUAL over the event's entry: the filer deletes
+// every stored chunk absent from an updated entry, so a snapshot older than
+// the live entry (the file was rewritten while its upload was in flight, or
+// the event is a replay) would delete the live chunks. A failed precondition
+// means the filer moved past this event; the event that superseded it follows
+// in the log and stamps the current entry, so the stale stamp is skipped the
+// same way a superseded upload is.
 func updateLocalEntry(filerClient filer_pb.FilerClient, dir string, entry *filer_pb.Entry, remoteEntry *filer_pb.RemoteEntry) error {
 	remoteEntry.LastLocalSyncTsNs = time.Now().UnixNano()
+	expected := proto.Clone(entry).(*filer_pb.Entry)
 	entry.RemoteEntry = remoteEntry
-	return filerClient.WithFilerClient(false, func(client filer_pb.SeaweedFilerClient) error {
+	err := filerClient.WithFilerClient(false, func(client filer_pb.SeaweedFilerClient) error {
 		_, err := client.UpdateEntry(context.Background(), &filer_pb.UpdateEntryRequest{
 			Directory: dir,
 			Entry:     entry,
+			Condition: ifEntryEqual(expected),
 		})
 		return err
 	})
+	if isFailedPrecondition(err) {
+		glog.Errorf("skipping stale stamp of %s: %v", util.NewFullPath(dir, entry.Name), err)
+		return nil
+	}
+	return err
+}
+
+// ifEntryEqual builds the precondition that the stored entry still equals the
+// one the event described: chunk fids, inline content, and metadata alike.
+func ifEntryEqual(entry *filer_pb.Entry) *filer_pb.WriteCondition {
+	return &filer_pb.WriteCondition{
+		Clauses: []*filer_pb.WriteCondition_Clause{{Kind: filer_pb.WriteCondition_IF_ENTRY_EQUAL, ExpectedEntry: entry}},
+	}
+}
+
+// isFailedPrecondition reports a write condition the filer refused, through
+// any wrapping WithFilerClient added.
+func isFailedPrecondition(err error) bool {
+	if err == nil {
+		return false
+	}
+	st, ok := status.FromError(err)
+	return ok && st.Code() == codes.FailedPrecondition
 }

 func isMultipartUploadFile(dir string, name string) bool {
@@ -16,6 +16,8 @@ import (
 	"github.com/seaweedfs/seaweedfs/weed/s3api/s3_constants"
 	"github.com/seaweedfs/seaweedfs/weed/util"
 	"google.golang.org/grpc"
+	"google.golang.org/grpc/codes"
+	"google.golang.org/grpc/status"
 	"google.golang.org/protobuf/proto"
 )

@@ -808,3 +810,41 @@ func TestRenameDeleteOldKeyNotFoundStillWrites(t *testing.T) {
 		t.Errorf("writes = %+v, want the new key %s written", remote.writes, remote_storage.FormatLocation(wantWrite))
 	}
 }
+
+func TestIfEntryEqualCarriesTheEventEntry(t *testing.T) {
+	entry := &filer_pb.Entry{
+		Name:    "f",
+		Content: []byte("inline"),
+		Chunks: []*filer_pb.FileChunk{
+			{FileId: "3,01a"},
+			{Fid: &filer_pb.FileId{VolumeId: 4, FileKey: 0x2b, Cookie: 0x0c}},
+		},
+	}
+	cond := ifEntryEqual(entry)
+	if len(cond.Clauses) != 1 || cond.Clauses[0].Kind != filer_pb.WriteCondition_IF_ENTRY_EQUAL {
+		t.Fatalf("condition = %v, want one IF_ENTRY_EQUAL clause", cond)
+	}
+	if got := cond.Clauses[0].ExpectedEntry; !proto.Equal(got, entry) {
+		t.Fatalf("expected_entry = %v, want %v", got, entry)
+	}
+}
+
+func TestIsFailedPrecondition(t *testing.T) {
+	refused := status.Error(codes.FailedPrecondition, "precondition failed: /buckets/b/f")
+	cases := []struct {
+		name string
+		err  error
+		want bool
+	}{
+		{"nil", nil, false},
+		{"status", refused, true},
+		{"wrapped status", fmt.Errorf("update entry: %w", refused), true},
+		{"message only", errors.New("rpc error: code = FailedPrecondition desc = precondition failed: /f"), false},
+		{"other", status.Error(codes.Unavailable, "filer down"), false},
+	}
+	for _, c := range cases {
+		if got := isFailedPrecondition(c.err); got != c.want {
+			t.Errorf("%s: isFailedPrecondition = %v, want %v", c.name, got, c.want)
+		}
+	}
+}
@@ -281,6 +281,7 @@ message WriteCondition {
        IF_EXTENDED_NOT_EQUAL = 7; // fail if present and extended[ext_key] == ext_value
        IF_EXTENDED_TIME_ELAPSED = 8; // fail if present and extended[ext_key] (unix seconds) is in the future
        IF_CHUNKS_EQUAL       = 9; // fail unless the stored chunk fid multiset equals fids (absent entry = no chunks)
+        IF_ENTRY_EQUAL        = 10; // fail unless the stored entry equals expected_entry
    }
    // Clause is one primitive comparison. IF_ETAG_MATCH holds when the current
    // entry's ETag equals any value in etags; IF_ETAG_NOT_MATCH holds when it
@@ -309,6 +310,7 @@ message WriteCondition {
        string gate_key = 7;       // IF_EXTENDED_TIME_ELAPSED: only enforce when extended[gate_key] == gate_value
        string gate_value = 8;     // gate value (e.g. retention mode COMPLIANCE for governance bypass)
        repeated string fids = 9;  // chunk fid strings for IF_CHUNKS_EQUAL
+        Entry expected_entry = 10; // whole-entry comparison for IF_ENTRY_EQUAL
    }
    repeated Clause clauses = 1;   // all must hold (logical AND)
 }
@@ -477,6 +479,7 @@ message UpdateEntryRequest {
    // the write, under the filer's per-path lock. The caller must route the
    // key's writes to this entry's owner filer for the check to be authoritative.
    WriteCondition condition = 6;
+    bool is_moved = 7;             // set on a forwarded request so the owner applies it locally
 }
 message UpdateEntryResponse {
    SubscribeMetadataResponse metadata_event = 1;
@@ -1,7 +1,7 @@
 // Code generated by protoc-gen-go. DO NOT EDIT.
 // versions:
 // 	protoc-gen-go v1.36.6
-// 	protoc        v7.35.0
+// 	protoc        v3.21.12
 // source: filer.proto

 package filer_pb
@@ -197,31 +197,33 @@ func (PosixLockOp) EnumDescriptor() ([]byte, []int) {
 type WriteCondition_Kind int32

 const (
-	WriteCondition_NONE                     WriteCondition_Kind = 0 // unconditional
-	WriteCondition_IF_NOT_EXISTS            WriteCondition_Kind = 1 // fail if the entry exists (If-None-Match: *)
-	WriteCondition_IF_EXISTS                WriteCondition_Kind = 2 // fail if the entry is absent (If-Match: *)
-	WriteCondition_IF_ETAG_MATCH            WriteCondition_Kind = 3 // fail if absent or etag matches none of the set (If-Match)
-	WriteCondition_IF_ETAG_NOT_MATCH        WriteCondition_Kind = 4 // fail if present and etag matches any of the set (If-None-Match)
-	WriteCondition_IF_UNMODIFIED_SINCE      WriteCondition_Kind = 5 // fail if present and mtime > unix_time
-	WriteCondition_IF_MODIFIED_SINCE        WriteCondition_Kind = 6 // fail if present and mtime <= unix_time
-	WriteCondition_IF_EXTENDED_NOT_EQUAL    WriteCondition_Kind = 7 // fail if present and extended[ext_key] == ext_value
-	WriteCondition_IF_EXTENDED_TIME_ELAPSED WriteCondition_Kind = 8 // fail if present and extended[ext_key] (unix seconds) is in the future
-	WriteCondition_IF_CHUNKS_EQUAL          WriteCondition_Kind = 9 // fail unless the stored chunk fid multiset equals fids (absent entry = no chunks)
+	WriteCondition_NONE                     WriteCondition_Kind = 0  // unconditional
+	WriteCondition_IF_NOT_EXISTS            WriteCondition_Kind = 1  // fail if the entry exists (If-None-Match: *)
+	WriteCondition_IF_EXISTS                WriteCondition_Kind = 2  // fail if the entry is absent (If-Match: *)
+	WriteCondition_IF_ETAG_MATCH            WriteCondition_Kind = 3  // fail if absent or etag matches none of the set (If-Match)
+	WriteCondition_IF_ETAG_NOT_MATCH        WriteCondition_Kind = 4  // fail if present and etag matches any of the set (If-None-Match)
+	WriteCondition_IF_UNMODIFIED_SINCE      WriteCondition_Kind = 5  // fail if present and mtime > unix_time
+	WriteCondition_IF_MODIFIED_SINCE        WriteCondition_Kind = 6  // fail if present and mtime <= unix_time
+	WriteCondition_IF_EXTENDED_NOT_EQUAL    WriteCondition_Kind = 7  // fail if present and extended[ext_key] == ext_value
+	WriteCondition_IF_EXTENDED_TIME_ELAPSED WriteCondition_Kind = 8  // fail if present and extended[ext_key] (unix seconds) is in the future
+	WriteCondition_IF_CHUNKS_EQUAL          WriteCondition_Kind = 9  // fail unless the stored chunk fid multiset equals fids (absent entry = no chunks)
+	WriteCondition_IF_ENTRY_EQUAL           WriteCondition_Kind = 10 // fail unless the stored entry equals expected_entry
 )

 // Enum value maps for WriteCondition_Kind.
 var (
 	WriteCondition_Kind_name = map[int32]string{
-		0: "NONE",
-		1: "IF_NOT_EXISTS",
-		2: "IF_EXISTS",
-		3: "IF_ETAG_MATCH",
-		4: "IF_ETAG_NOT_MATCH",
-		5: "IF_UNMODIFIED_SINCE",
-		6: "IF_MODIFIED_SINCE",
-		7: "IF_EXTENDED_NOT_EQUAL",
-		8: "IF_EXTENDED_TIME_ELAPSED",
-		9: "IF_CHUNKS_EQUAL",
+		0:  "NONE",
+		1:  "IF_NOT_EXISTS",
+		2:  "IF_EXISTS",
+		3:  "IF_ETAG_MATCH",
+		4:  "IF_ETAG_NOT_MATCH",
+		5:  "IF_UNMODIFIED_SINCE",
+		6:  "IF_MODIFIED_SINCE",
+		7:  "IF_EXTENDED_NOT_EQUAL",
+		8:  "IF_EXTENDED_TIME_ELAPSED",
+		9:  "IF_CHUNKS_EQUAL",
+		10: "IF_ENTRY_EQUAL",
 	}
 	WriteCondition_Kind_value = map[string]int32{
 		"NONE":                     0,
@@ -234,6 +236,7 @@ var (
 		"IF_EXTENDED_NOT_EQUAL":    7,
 		"IF_EXTENDED_TIME_ELAPSED": 8,
 		"IF_CHUNKS_EQUAL":          9,
+		"IF_ENTRY_EQUAL":           10,
 	}
 )

@@ -2375,6 +2378,7 @@ type UpdateEntryRequest struct {
 	// the write, under the filer's per-path lock. The caller must route the
 	// key's writes to this entry's owner filer for the check to be authoritative.
 	Condition     *WriteCondition `protobuf:"bytes,6,opt,name=condition,proto3" json:"condition,omitempty"`
+	IsMoved       bool            `protobuf:"varint,7,opt,name=is_moved,json=isMoved,proto3" json:"is_moved,omitempty"` // set on a forwarded request so the owner applies it locally
 	unknownFields protoimpl.UnknownFields
 	sizeCache     protoimpl.SizeCache
 }
@@ -2451,6 +2455,13 @@ func (x *UpdateEntryRequest) GetCondition() *WriteCondition {
 	return nil
 }

+func (x *UpdateEntryRequest) GetIsMoved() bool {
+	if x != nil {
+		return x.IsMoved
+	}
+	return false
+}
+
 type UpdateEntryResponse struct {
 	state         protoimpl.MessageState     `protogen:"open.v1"`
 	MetadataEvent *SubscribeMetadataResponse `protobuf:"bytes,1,opt,name=metadata_event,json=metadataEvent,proto3" json:"metadata_event,omitempty"`
@@ -6856,14 +6867,15 @@ func (x *LookupDirectoryEntriesResponse) GetReadAuth() map[string]string {
 type WriteCondition_Clause struct {
 	state         protoimpl.MessageState `protogen:"open.v1"`
 	Kind          WriteCondition_Kind    `protobuf:"varint,1,opt,name=kind,proto3,enum=filer_pb.WriteCondition_Kind" json:"kind,omitempty"`
-	Etags         []string               `protobuf:"bytes,2,rep,name=etags,proto3" json:"etags,omitempty"`                           // ETag set for IF_ETAG_* kinds
-	UnixTime      int64                  `protobuf:"varint,3,opt,name=unix_time,json=unixTime,proto3" json:"unix_time,omitempty"`    // bound (unix seconds) for IF_*_SINCE kinds
-	AllowWeak     bool                   `protobuf:"varint,4,opt,name=allow_weak,json=allowWeak,proto3" json:"allow_weak,omitempty"` // compare ETags ignoring the weak (W/) marker
-	ExtKey        string                 `protobuf:"bytes,5,opt,name=ext_key,json=extKey,proto3" json:"ext_key,omitempty"`           // extended attribute name for IF_EXTENDED_* kinds
-	ExtValue      string                 `protobuf:"bytes,6,opt,name=ext_value,json=extValue,proto3" json:"ext_value,omitempty"`     // blocking value for IF_EXTENDED_NOT_EQUAL
-	GateKey       string                 `protobuf:"bytes,7,opt,name=gate_key,json=gateKey,proto3" json:"gate_key,omitempty"`        // IF_EXTENDED_TIME_ELAPSED: only enforce when extended[gate_key] == gate_value
-	GateValue     string                 `protobuf:"bytes,8,opt,name=gate_value,json=gateValue,proto3" json:"gate_value,omitempty"`  // gate value (e.g. retention mode COMPLIANCE for governance bypass)
-	Fids          []string               `protobuf:"bytes,9,rep,name=fids,proto3" json:"fids,omitempty"`                             // chunk fid strings for IF_CHUNKS_EQUAL
+	Etags         []string               `protobuf:"bytes,2,rep,name=etags,proto3" json:"etags,omitempty"`                                       // ETag set for IF_ETAG_* kinds
+	UnixTime      int64                  `protobuf:"varint,3,opt,name=unix_time,json=unixTime,proto3" json:"unix_time,omitempty"`                // bound (unix seconds) for IF_*_SINCE kinds
+	AllowWeak     bool                   `protobuf:"varint,4,opt,name=allow_weak,json=allowWeak,proto3" json:"allow_weak,omitempty"`             // compare ETags ignoring the weak (W/) marker
+	ExtKey        string                 `protobuf:"bytes,5,opt,name=ext_key,json=extKey,proto3" json:"ext_key,omitempty"`                       // extended attribute name for IF_EXTENDED_* kinds
+	ExtValue      string                 `protobuf:"bytes,6,opt,name=ext_value,json=extValue,proto3" json:"ext_value,omitempty"`                 // blocking value for IF_EXTENDED_NOT_EQUAL
+	GateKey       string                 `protobuf:"bytes,7,opt,name=gate_key,json=gateKey,proto3" json:"gate_key,omitempty"`                    // IF_EXTENDED_TIME_ELAPSED: only enforce when extended[gate_key] == gate_value
+	GateValue     string                 `protobuf:"bytes,8,opt,name=gate_value,json=gateValue,proto3" json:"gate_value,omitempty"`              // gate value (e.g. retention mode COMPLIANCE for governance bypass)
+	Fids          []string               `protobuf:"bytes,9,rep,name=fids,proto3" json:"fids,omitempty"`                                         // chunk fid strings for IF_CHUNKS_EQUAL
+	ExpectedEntry *Entry                 `protobuf:"bytes,10,opt,name=expected_entry,json=expectedEntry,proto3" json:"expected_entry,omitempty"` // whole-entry comparison for IF_ENTRY_EQUAL
 	unknownFields protoimpl.UnknownFields
 	sizeCache     protoimpl.SizeCache
 }
@@ -6961,6 +6973,13 @@ func (x *WriteCondition_Clause) GetFids() []string {
 	return nil
 }

+func (x *WriteCondition_Clause) GetExpectedEntry() *Entry {
+	if x != nil {
+		return x.ExpectedEntry
+	}
+	return nil
+}
+
 // if found, send the exact address
 // if not found, send the full list of existing brokers
 type LocateBrokerResponse_Resource struct {
@@ -7301,9 +7320,9 @@ const file_filer_proto_rawDesc = "" +
 	"signatures\x12=\n" +
 	"\x1bskip_check_parent_directory\x18\x06 \x01(\bR\x18skipCheckParentDirectory\x126\n" +
 	"\tcondition\x18\a \x01(\v2\x18.filer_pb.WriteConditionR\tcondition\x12\x19\n" +
-	"\bis_moved\x18\b \x01(\bR\aisMoved\"\xbc\x04\n" +
+	"\bis_moved\x18\b \x01(\bR\aisMoved\"\x88\x05\n" +
 	"\x0eWriteCondition\x129\n" +
-	"\aclauses\x18\x01 \x03(\v2\x1f.filer_pb.WriteCondition.ClauseR\aclauses\x1a\x91\x02\n" +
+	"\aclauses\x18\x01 \x03(\v2\x1f.filer_pb.WriteCondition.ClauseR\aclauses\x1a\xc9\x02\n" +
 	"\x06Clause\x121\n" +
 	"\x04kind\x18\x01 \x01(\x0e2\x1d.filer_pb.WriteCondition.KindR\x04kind\x12\x14\n" +
 	"\x05etags\x18\x02 \x03(\tR\x05etags\x12\x1b\n" +
@@ -7315,7 +7334,9 @@ const file_filer_proto_rawDesc = "" +
 	"\bgate_key\x18\a \x01(\tR\agateKey\x12\x1d\n" +
 	"\n" +
 	"gate_value\x18\b \x01(\tR\tgateValue\x12\x12\n" +
-	"\x04fids\x18\t \x03(\tR\x04fids\"\xda\x01\n" +
+	"\x04fids\x18\t \x03(\tR\x04fids\x126\n" +
+	"\x0eexpected_entry\x18\n" +
+	" \x01(\v2\x0f.filer_pb.EntryR\rexpectedEntry\"\xee\x01\n" +
 	"\x04Kind\x12\b\n" +
 	"\x04NONE\x10\x00\x12\x11\n" +
 	"\rIF_NOT_EXISTS\x10\x01\x12\r\n" +
@@ -7326,7 +7347,9 @@ const file_filer_proto_rawDesc = "" +
 	"\x11IF_MODIFIED_SINCE\x10\x06\x12\x19\n" +
 	"\x15IF_EXTENDED_NOT_EQUAL\x10\a\x12\x1c\n" +
 	"\x18IF_EXTENDED_TIME_ELAPSED\x10\b\x12\x13\n" +
-	"\x0fIF_CHUNKS_EQUAL\x10\t\"\xa2\x05\n" +
+	"\x0fIF_CHUNKS_EQUAL\x10\t\x12\x12\n" +
+	"\x0eIF_ENTRY_EQUAL\x10\n" +
+	"\"\xa2\x05\n" +
 	"\x0eObjectMutation\x121\n" +
 	"\x04type\x18\x01 \x01(\x0e2\x1d.filer_pb.ObjectMutation.TypeR\x04type\x12\x1c\n" +
 	"\tdirectory\x18\x02 \x01(\tR\tdirectory\x12\x12\n" +
@@ -7414,7 +7437,7 @@ const file_filer_proto_rawDesc = "" +
 	"\n" +
 	"error_code\x18\x03 \x01(\x0e2\x14.filer_pb.FilerErrorR\terrorCode\x12\x1a\n" +
 	"\tlog_ts_ns\x18\x04 \x01(\x03R\alogTsNs\x12#\n" +
-	"\rlog_signature\x18\x05 \x01(\x05R\flogSignature\"\x8a\x03\n" +
+	"\rlog_signature\x18\x05 \x01(\x05R\flogSignature\"\xa5\x03\n" +
 	"\x12UpdateEntryRequest\x12\x1c\n" +
 	"\tdirectory\x18\x01 \x01(\tR\tdirectory\x12%\n" +
 	"\x05entry\x18\x02 \x01(\v2\x0f.filer_pb.EntryR\x05entry\x121\n" +
@@ -7423,7 +7446,8 @@ const file_filer_proto_rawDesc = "" +
 	"signatures\x18\x04 \x03(\x05R\n" +
 	"signatures\x12_\n" +
 	"\x11expected_extended\x18\x05 \x03(\v22.filer_pb.UpdateEntryRequest.ExpectedExtendedEntryR\x10expectedExtended\x126\n" +
-	"\tcondition\x18\x06 \x01(\v2\x18.filer_pb.WriteConditionR\tcondition\x1aC\n" +
+	"\tcondition\x18\x06 \x01(\v2\x18.filer_pb.WriteConditionR\tcondition\x12\x19\n" +
+	"\bis_moved\x18\a \x01(\bR\aisMoved\x1aC\n" +
 	"\x15ExpectedExtendedEntry\x12\x10\n" +
 	"\x03key\x18\x01 \x01(\tR\x03key\x12\x14\n" +
 	"\x05value\x18\x02 \x01(\fR\x05value:\x028\x01\"\xa2\x01\n" +
@@ -8072,83 +8096,84 @@ var file_filer_proto_depIdxs = []int32{
 	107, // 67: filer_pb.LookupDirectoryEntriesResponse.locations_map:type_name -> filer_pb.LookupDirectoryEntriesResponse.LocationsMapEntry
 	108, // 68: filer_pb.LookupDirectoryEntriesResponse.read_auth:type_name -> filer_pb.LookupDirectoryEntriesResponse.ReadAuthEntry
 	3,   // 69: filer_pb.WriteCondition.Clause.kind:type_name -> filer_pb.WriteCondition.Kind
-	44,  // 70: filer_pb.LookupVolumeResponse.LocationsMapEntry.value:type_name -> filer_pb.Locations
-	44,  // 71: filer_pb.LookupDirectoryEntriesResponse.LocationsMapEntry.value:type_name -> filer_pb.Locations
-	5,   // 72: filer_pb.SeaweedFiler.LookupDirectoryEntry:input_type -> filer_pb.LookupDirectoryEntryRequest
-	96,  // 73: filer_pb.SeaweedFiler.LookupDirectoryEntries:input_type -> filer_pb.LookupDirectoryEntriesRequest
-	7,   // 74: filer_pb.SeaweedFiler.ListEntries:input_type -> filer_pb.ListEntriesRequest
-	17,  // 75: filer_pb.SeaweedFiler.CreateEntry:input_type -> filer_pb.CreateEntryRequest
-	29,  // 76: filer_pb.SeaweedFiler.UpdateEntry:input_type -> filer_pb.UpdateEntryRequest
-	31,  // 77: filer_pb.SeaweedFiler.TouchAccessTime:input_type -> filer_pb.TouchAccessTimeRequest
-	33,  // 78: filer_pb.SeaweedFiler.AppendToEntry:input_type -> filer_pb.AppendToEntryRequest
-	35,  // 79: filer_pb.SeaweedFiler.DeleteEntry:input_type -> filer_pb.DeleteEntryRequest
-	21,  // 80: filer_pb.SeaweedFiler.ObjectTransaction:input_type -> filer_pb.ObjectTransactionRequest
-	26,  // 81: filer_pb.SeaweedFiler.ObjectTransactionBatch:input_type -> filer_pb.ObjectTransactionBatchRequest
-	24,  // 82: filer_pb.SeaweedFiler.PosixLock:input_type -> filer_pb.PosixLockRequest
-	37,  // 83: filer_pb.SeaweedFiler.AtomicRenameEntry:input_type -> filer_pb.AtomicRenameEntryRequest
-	39,  // 84: filer_pb.SeaweedFiler.StreamRenameEntry:input_type -> filer_pb.StreamRenameEntryRequest
-	89,  // 85: filer_pb.SeaweedFiler.StreamMutateEntry:input_type -> filer_pb.StreamMutateEntryRequest
-	41,  // 86: filer_pb.SeaweedFiler.AssignVolume:input_type -> filer_pb.AssignVolumeRequest
-	43,  // 87: filer_pb.SeaweedFiler.LookupVolume:input_type -> filer_pb.LookupVolumeRequest
-	48,  // 88: filer_pb.SeaweedFiler.CollectionList:input_type -> filer_pb.CollectionListRequest
-	50,  // 89: filer_pb.SeaweedFiler.DeleteCollection:input_type -> filer_pb.DeleteCollectionRequest
-	52,  // 90: filer_pb.SeaweedFiler.Statistics:input_type -> filer_pb.StatisticsRequest
-	54,  // 91: filer_pb.SeaweedFiler.Ping:input_type -> filer_pb.PingRequest
-	56,  // 92: filer_pb.SeaweedFiler.GetFilerConfiguration:input_type -> filer_pb.GetFilerConfigurationRequest
-	64,  // 93: filer_pb.SeaweedFiler.TraverseBfsMetadata:input_type -> filer_pb.TraverseBfsMetadataRequest
-	58,  // 94: filer_pb.SeaweedFiler.SubscribeMetadata:input_type -> filer_pb.SubscribeMetadataRequest
-	58,  // 95: filer_pb.SeaweedFiler.SubscribeLocalMetadata:input_type -> filer_pb.SubscribeMetadataRequest
-	60,  // 96: filer_pb.SeaweedFiler.ListMetadataSubscribers:input_type -> filer_pb.ListMetadataSubscribersRequest
-	71,  // 97: filer_pb.SeaweedFiler.KvGet:input_type -> filer_pb.KvGetRequest
-	73,  // 98: filer_pb.SeaweedFiler.KvPut:input_type -> filer_pb.KvPutRequest
-	76,  // 99: filer_pb.SeaweedFiler.CacheRemoteObjectToLocalCluster:input_type -> filer_pb.CacheRemoteObjectToLocalClusterRequest
-	78,  // 100: filer_pb.SeaweedFiler.DistributedLock:input_type -> filer_pb.LockRequest
-	80,  // 101: filer_pb.SeaweedFiler.DistributedUnlock:input_type -> filer_pb.UnlockRequest
-	82,  // 102: filer_pb.SeaweedFiler.FindLockOwner:input_type -> filer_pb.FindLockOwnerRequest
-	85,  // 103: filer_pb.SeaweedFiler.TransferLocks:input_type -> filer_pb.TransferLocksRequest
-	87,  // 104: filer_pb.SeaweedFiler.ReplicateLock:input_type -> filer_pb.ReplicateLockRequest
-	91,  // 105: filer_pb.SeaweedFiler.MountRegister:input_type -> filer_pb.MountRegisterRequest
-	93,  // 106: filer_pb.SeaweedFiler.MountList:input_type -> filer_pb.MountListRequest
-	6,   // 107: filer_pb.SeaweedFiler.LookupDirectoryEntry:output_type -> filer_pb.LookupDirectoryEntryResponse
-	98,  // 108: filer_pb.SeaweedFiler.LookupDirectoryEntries:output_type -> filer_pb.LookupDirectoryEntriesResponse
-	8,   // 109: filer_pb.SeaweedFiler.ListEntries:output_type -> filer_pb.ListEntriesResponse
-	28,  // 110: filer_pb.SeaweedFiler.CreateEntry:output_type -> filer_pb.CreateEntryResponse
-	30,  // 111: filer_pb.SeaweedFiler.UpdateEntry:output_type -> filer_pb.UpdateEntryResponse
-	32,  // 112: filer_pb.SeaweedFiler.TouchAccessTime:output_type -> filer_pb.TouchAccessTimeResponse
-	34,  // 113: filer_pb.SeaweedFiler.AppendToEntry:output_type -> filer_pb.AppendToEntryResponse
-	36,  // 114: filer_pb.SeaweedFiler.DeleteEntry:output_type -> filer_pb.DeleteEntryResponse
-	22,  // 115: filer_pb.SeaweedFiler.ObjectTransaction:output_type -> filer_pb.ObjectTransactionResponse
-	27,  // 116: filer_pb.SeaweedFiler.ObjectTransactionBatch:output_type -> filer_pb.ObjectTransactionBatchResponse
-	25,  // 117: filer_pb.SeaweedFiler.PosixLock:output_type -> filer_pb.PosixLockResponse
-	38,  // 118: filer_pb.SeaweedFiler.AtomicRenameEntry:output_type -> filer_pb.AtomicRenameEntryResponse
-	40,  // 119: filer_pb.SeaweedFiler.StreamRenameEntry:output_type -> filer_pb.StreamRenameEntryResponse
-	90,  // 120: filer_pb.SeaweedFiler.StreamMutateEntry:output_type -> filer_pb.StreamMutateEntryResponse
-	42,  // 121: filer_pb.SeaweedFiler.AssignVolume:output_type -> filer_pb.AssignVolumeResponse
-	46,  // 122: filer_pb.SeaweedFiler.LookupVolume:output_type -> filer_pb.LookupVolumeResponse
-	49,  // 123: filer_pb.SeaweedFiler.CollectionList:output_type -> filer_pb.CollectionListResponse
-	51,  // 124: filer_pb.SeaweedFiler.DeleteCollection:output_type -> filer_pb.DeleteCollectionResponse
-	53,  // 125: filer_pb.SeaweedFiler.Statistics:output_type -> filer_pb.StatisticsResponse
-	55,  // 126: filer_pb.SeaweedFiler.Ping:output_type -> filer_pb.PingResponse
-	57,  // 127: filer_pb.SeaweedFiler.GetFilerConfiguration:output_type -> filer_pb.GetFilerConfigurationResponse
-	65,  // 128: filer_pb.SeaweedFiler.TraverseBfsMetadata:output_type -> filer_pb.TraverseBfsMetadataResponse
-	59,  // 129: filer_pb.SeaweedFiler.SubscribeMetadata:output_type -> filer_pb.SubscribeMetadataResponse
-	59,  // 130: filer_pb.SeaweedFiler.SubscribeLocalMetadata:output_type -> filer_pb.SubscribeMetadataResponse
-	61,  // 131: filer_pb.SeaweedFiler.ListMetadataSubscribers:output_type -> filer_pb.ListMetadataSubscribersResponse
-	72,  // 132: filer_pb.SeaweedFiler.KvGet:output_type -> filer_pb.KvGetResponse
-	74,  // 133: filer_pb.SeaweedFiler.KvPut:output_type -> filer_pb.KvPutResponse
-	77,  // 134: filer_pb.SeaweedFiler.CacheRemoteObjectToLocalCluster:output_type -> filer_pb.CacheRemoteObjectToLocalClusterResponse
-	79,  // 135: filer_pb.SeaweedFiler.DistributedLock:output_type -> filer_pb.LockResponse
-	81,  // 136: filer_pb.SeaweedFiler.DistributedUnlock:output_type -> filer_pb.UnlockResponse
-	83,  // 137: filer_pb.SeaweedFiler.FindLockOwner:output_type -> filer_pb.FindLockOwnerResponse
-	86,  // 138: filer_pb.SeaweedFiler.TransferLocks:output_type -> filer_pb.TransferLocksResponse
-	88,  // 139: filer_pb.SeaweedFiler.ReplicateLock:output_type -> filer_pb.ReplicateLockResponse
-	92,  // 140: filer_pb.SeaweedFiler.MountRegister:output_type -> filer_pb.MountRegisterResponse
-	94,  // 141: filer_pb.SeaweedFiler.MountList:output_type -> filer_pb.MountListResponse
-	107, // [107:142] is the sub-list for method output_type
-	72,  // [72:107] is the sub-list for method input_type
-	72,  // [72:72] is the sub-list for extension type_name
-	72,  // [72:72] is the sub-list for extension extendee
-	0,   // [0:72] is the sub-list for field type_name
+	10,  // 70: filer_pb.WriteCondition.Clause.expected_entry:type_name -> filer_pb.Entry
+	44,  // 71: filer_pb.LookupVolumeResponse.LocationsMapEntry.value:type_name -> filer_pb.Locations
+	44,  // 72: filer_pb.LookupDirectoryEntriesResponse.LocationsMapEntry.value:type_name -> filer_pb.Locations
+	5,   // 73: filer_pb.SeaweedFiler.LookupDirectoryEntry:input_type -> filer_pb.LookupDirectoryEntryRequest
+	96,  // 74: filer_pb.SeaweedFiler.LookupDirectoryEntries:input_type -> filer_pb.LookupDirectoryEntriesRequest
+	7,   // 75: filer_pb.SeaweedFiler.ListEntries:input_type -> filer_pb.ListEntriesRequest
+	17,  // 76: filer_pb.SeaweedFiler.CreateEntry:input_type -> filer_pb.CreateEntryRequest
+	29,  // 77: filer_pb.SeaweedFiler.UpdateEntry:input_type -> filer_pb.UpdateEntryRequest
+	31,  // 78: filer_pb.SeaweedFiler.TouchAccessTime:input_type -> filer_pb.TouchAccessTimeRequest
+	33,  // 79: filer_pb.SeaweedFiler.AppendToEntry:input_type -> filer_pb.AppendToEntryRequest
+	35,  // 80: filer_pb.SeaweedFiler.DeleteEntry:input_type -> filer_pb.DeleteEntryRequest
+	21,  // 81: filer_pb.SeaweedFiler.ObjectTransaction:input_type -> filer_pb.ObjectTransactionRequest
+	26,  // 82: filer_pb.SeaweedFiler.ObjectTransactionBatch:input_type -> filer_pb.ObjectTransactionBatchRequest
+	24,  // 83: filer_pb.SeaweedFiler.PosixLock:input_type -> filer_pb.PosixLockRequest
+	37,  // 84: filer_pb.SeaweedFiler.AtomicRenameEntry:input_type -> filer_pb.AtomicRenameEntryRequest
+	39,  // 85: filer_pb.SeaweedFiler.StreamRenameEntry:input_type -> filer_pb.StreamRenameEntryRequest
+	89,  // 86: filer_pb.SeaweedFiler.StreamMutateEntry:input_type -> filer_pb.StreamMutateEntryRequest
+	41,  // 87: filer_pb.SeaweedFiler.AssignVolume:input_type -> filer_pb.AssignVolumeRequest
+	43,  // 88: filer_pb.SeaweedFiler.LookupVolume:input_type -> filer_pb.LookupVolumeRequest
+	48,  // 89: filer_pb.SeaweedFiler.CollectionList:input_type -> filer_pb.CollectionListRequest
+	50,  // 90: filer_pb.SeaweedFiler.DeleteCollection:input_type -> filer_pb.DeleteCollectionRequest
+	52,  // 91: filer_pb.SeaweedFiler.Statistics:input_type -> filer_pb.StatisticsRequest
+	54,  // 92: filer_pb.SeaweedFiler.Ping:input_type -> filer_pb.PingRequest
+	56,  // 93: filer_pb.SeaweedFiler.GetFilerConfiguration:input_type -> filer_pb.GetFilerConfigurationRequest
+	64,  // 94: filer_pb.SeaweedFiler.TraverseBfsMetadata:input_type -> filer_pb.TraverseBfsMetadataRequest
+	58,  // 95: filer_pb.SeaweedFiler.SubscribeMetadata:input_type -> filer_pb.SubscribeMetadataRequest
+	58,  // 96: filer_pb.SeaweedFiler.SubscribeLocalMetadata:input_type -> filer_pb.SubscribeMetadataRequest
+	60,  // 97: filer_pb.SeaweedFiler.ListMetadataSubscribers:input_type -> filer_pb.ListMetadataSubscribersRequest
+	71,  // 98: filer_pb.SeaweedFiler.KvGet:input_type -> filer_pb.KvGetRequest
+	73,  // 99: filer_pb.SeaweedFiler.KvPut:input_type -> filer_pb.KvPutRequest
+	76,  // 100: filer_pb.SeaweedFiler.CacheRemoteObjectToLocalCluster:input_type -> filer_pb.CacheRemoteObjectToLocalClusterRequest
+	78,  // 101: filer_pb.SeaweedFiler.DistributedLock:input_type -> filer_pb.LockRequest
+	80,  // 102: filer_pb.SeaweedFiler.DistributedUnlock:input_type -> filer_pb.UnlockRequest
+	82,  // 103: filer_pb.SeaweedFiler.FindLockOwner:input_type -> filer_pb.FindLockOwnerRequest
+	85,  // 104: filer_pb.SeaweedFiler.TransferLocks:input_type -> filer_pb.TransferLocksRequest
+	87,  // 105: filer_pb.SeaweedFiler.ReplicateLock:input_type -> filer_pb.ReplicateLockRequest
+	91,  // 106: filer_pb.SeaweedFiler.MountRegister:input_type -> filer_pb.MountRegisterRequest
+	93,  // 107: filer_pb.SeaweedFiler.MountList:input_type -> filer_pb.MountListRequest
+	6,   // 108: filer_pb.SeaweedFiler.LookupDirectoryEntry:output_type -> filer_pb.LookupDirectoryEntryResponse
+	98,  // 109: filer_pb.SeaweedFiler.LookupDirectoryEntries:output_type -> filer_pb.LookupDirectoryEntriesResponse
+	8,   // 110: filer_pb.SeaweedFiler.ListEntries:output_type -> filer_pb.ListEntriesResponse
+	28,  // 111: filer_pb.SeaweedFiler.CreateEntry:output_type -> filer_pb.CreateEntryResponse
+	30,  // 112: filer_pb.SeaweedFiler.UpdateEntry:output_type -> filer_pb.UpdateEntryResponse
+	32,  // 113: filer_pb.SeaweedFiler.TouchAccessTime:output_type -> filer_pb.TouchAccessTimeResponse
+	34,  // 114: filer_pb.SeaweedFiler.AppendToEntry:output_type -> filer_pb.AppendToEntryResponse
+	36,  // 115: filer_pb.SeaweedFiler.DeleteEntry:output_type -> filer_pb.DeleteEntryResponse
+	22,  // 116: filer_pb.SeaweedFiler.ObjectTransaction:output_type -> filer_pb.ObjectTransactionResponse
+	27,  // 117: filer_pb.SeaweedFiler.ObjectTransactionBatch:output_type -> filer_pb.ObjectTransactionBatchResponse
+	25,  // 118: filer_pb.SeaweedFiler.PosixLock:output_type -> filer_pb.PosixLockResponse
+	38,  // 119: filer_pb.SeaweedFiler.AtomicRenameEntry:output_type -> filer_pb.AtomicRenameEntryResponse
+	40,  // 120: filer_pb.SeaweedFiler.StreamRenameEntry:output_type -> filer_pb.StreamRenameEntryResponse
+	90,  // 121: filer_pb.SeaweedFiler.StreamMutateEntry:output_type -> filer_pb.StreamMutateEntryResponse
+	42,  // 122: filer_pb.SeaweedFiler.AssignVolume:output_type -> filer_pb.AssignVolumeResponse
+	46,  // 123: filer_pb.SeaweedFiler.LookupVolume:output_type -> filer_pb.LookupVolumeResponse
+	49,  // 124: filer_pb.SeaweedFiler.CollectionList:output_type -> filer_pb.CollectionListResponse
+	51,  // 125: filer_pb.SeaweedFiler.DeleteCollection:output_type -> filer_pb.DeleteCollectionResponse
+	53,  // 126: filer_pb.SeaweedFiler.Statistics:output_type -> filer_pb.StatisticsResponse
+	55,  // 127: filer_pb.SeaweedFiler.Ping:output_type -> filer_pb.PingResponse
+	57,  // 128: filer_pb.SeaweedFiler.GetFilerConfiguration:output_type -> filer_pb.GetFilerConfigurationResponse
+	65,  // 129: filer_pb.SeaweedFiler.TraverseBfsMetadata:output_type -> filer_pb.TraverseBfsMetadataResponse
+	59,  // 130: filer_pb.SeaweedFiler.SubscribeMetadata:output_type -> filer_pb.SubscribeMetadataResponse
+	59,  // 131: filer_pb.SeaweedFiler.SubscribeLocalMetadata:output_type -> filer_pb.SubscribeMetadataResponse
+	61,  // 132: filer_pb.SeaweedFiler.ListMetadataSubscribers:output_type -> filer_pb.ListMetadataSubscribersResponse
+	72,  // 133: filer_pb.SeaweedFiler.KvGet:output_type -> filer_pb.KvGetResponse
+	74,  // 134: filer_pb.SeaweedFiler.KvPut:output_type -> filer_pb.KvPutResponse
+	77,  // 135: filer_pb.SeaweedFiler.CacheRemoteObjectToLocalCluster:output_type -> filer_pb.CacheRemoteObjectToLocalClusterResponse
+	79,  // 136: filer_pb.SeaweedFiler.DistributedLock:output_type -> filer_pb.LockResponse
+	81,  // 137: filer_pb.SeaweedFiler.DistributedUnlock:output_type -> filer_pb.UnlockResponse
+	83,  // 138: filer_pb.SeaweedFiler.FindLockOwner:output_type -> filer_pb.FindLockOwnerResponse
+	86,  // 139: filer_pb.SeaweedFiler.TransferLocks:output_type -> filer_pb.TransferLocksResponse
+	88,  // 140: filer_pb.SeaweedFiler.ReplicateLock:output_type -> filer_pb.ReplicateLockResponse
+	92,  // 141: filer_pb.SeaweedFiler.MountRegister:output_type -> filer_pb.MountRegisterResponse
+	94,  // 142: filer_pb.SeaweedFiler.MountList:output_type -> filer_pb.MountListResponse
+	108, // [108:143] is the sub-list for method output_type
+	73,  // [73:108] is the sub-list for method input_type
+	73,  // [73:73] is the sub-list for extension type_name
+	73,  // [73:73] is the sub-list for extension extendee
+	0,   // [0:73] is the sub-list for field type_name
 }

 func init() { file_filer_proto_init() }
@@ -1,7 +1,7 @@
 // Code generated by protoc-gen-go-grpc. DO NOT EDIT.
 // versions:
 // - protoc-gen-go-grpc v1.6.2
-// - protoc             v7.35.0
+// - protoc             v3.21.12
 // source: filer.proto

 package filer_pb
@@ -1135,6 +1135,16 @@ func (m *WriteCondition_Clause) MarshalToSizedBufferVT(dAtA []byte) (int, error)
 		i -= len(m.unknownFields)
 		copy(dAtA[i:], m.unknownFields)
 	}
+	if m.ExpectedEntry != nil {
+		size, err := m.ExpectedEntry.MarshalToSizedBufferVT(dAtA[:i])
+		if err != nil {
+			return 0, err
+		}
+		i -= size
+		i = protohelpers.EncodeVarint(dAtA, i, uint64(size))
+		i--
+		dAtA[i] = 0x52
+	}
 	if len(m.Fids) > 0 {
 		for iNdEx := len(m.Fids) - 1; iNdEx >= 0; iNdEx-- {
 			i -= len(m.Fids[iNdEx])
@@ -2087,6 +2097,16 @@ func (m *UpdateEntryRequest) MarshalToSizedBufferVT(dAtA []byte) (int, error) {
 		i -= len(m.unknownFields)
 		copy(dAtA[i:], m.unknownFields)
 	}
+	if m.IsMoved {
+		i--
+		if m.IsMoved {
+			dAtA[i] = 1
+		} else {
+			dAtA[i] = 0
+		}
+		i--
+		dAtA[i] = 0x38
+	}
 	if m.Condition != nil {
 		size, err := m.Condition.MarshalToSizedBufferVT(dAtA[:i])
 		if err != nil {
@@ -6958,6 +6978,10 @@ func (m *WriteCondition_Clause) SizeVT() (n int) {
 			n += 1 + l + protohelpers.SizeOfVarint(uint64(l))
 		}
 	}
+	if m.ExpectedEntry != nil {
+		l = m.ExpectedEntry.SizeVT()
+		n += 1 + l + protohelpers.SizeOfVarint(uint64(l))
+	}
 	n += len(m.unknownFields)
 	return n
 }
@@ -7332,6 +7356,9 @@ func (m *UpdateEntryRequest) SizeVT() (n int) {
 		l = m.Condition.SizeVT()
 		n += 1 + l + protohelpers.SizeOfVarint(uint64(l))
 	}
+	if m.IsMoved {
+		n += 2
+	}
 	n += len(m.unknownFields)
 	return n
 }
@@ -12374,6 +12401,42 @@ func (m *WriteCondition_Clause) UnmarshalVT(dAtA []byte) error {
 			}
 			m.Fids = append(m.Fids, string(dAtA[iNdEx:postIndex]))
 			iNdEx = postIndex
+		case 10:
+			if wireType != 2 {
+				return fmt.Errorf("proto: wrong wireType = %d for field ExpectedEntry", wireType)
+			}
+			var msglen int
+			for shift := uint(0); ; shift += 7 {
+				if shift >= 64 {
+					return protohelpers.ErrIntOverflow
+				}
+				if iNdEx >= l {
+					return io.ErrUnexpectedEOF
+				}
+				b := dAtA[iNdEx]
+				iNdEx++
+				msglen |= int(b&0x7F) << shift
+				if b < 0x80 {
+					break
+				}
+			}
+			if msglen < 0 {
+				return protohelpers.ErrInvalidLength
+			}
+			postIndex := iNdEx + msglen
+			if postIndex < 0 {
+				return protohelpers.ErrInvalidLength
+			}
+			if postIndex > l {
+				return io.ErrUnexpectedEOF
+			}
+			if m.ExpectedEntry == nil {
+				m.ExpectedEntry = &Entry{}
+			}
+			if err := m.ExpectedEntry.UnmarshalVT(dAtA[iNdEx:postIndex]); err != nil {
+				return err
+			}
+			iNdEx = postIndex
 		default:
 			iNdEx = preIndex
 			skippy, err := protohelpers.Skip(dAtA[iNdEx:])
@@ -15067,6 +15130,26 @@ func (m *UpdateEntryRequest) UnmarshalVT(dAtA []byte) error {
 				return err
 			}
 			iNdEx = postIndex
+		case 7:
+			if wireType != 0 {
+				return fmt.Errorf("proto: wrong wireType = %d for field IsMoved", wireType)
+			}
+			var v int
+			for shift := uint(0); ; shift += 7 {
+				if shift >= 64 {
+					return protohelpers.ErrIntOverflow
+				}
+				if iNdEx >= l {
+					return io.ErrUnexpectedEOF
+				}
+				b := dAtA[iNdEx]
+				iNdEx++
+				v |= int(b&0x7F) << shift
+				if b < 0x80 {
+					break
+				}
+			}
+			m.IsMoved = bool(v != 0)
 		default:
 			iNdEx = preIndex
 			skippy, err := protohelpers.Skip(dAtA[iNdEx:])
@@ -662,6 +662,32 @@ func (fs *FilerServer) UpdateEntry(ctx context.Context, req *filer_pb.UpdateEntr

 	fullpath := util.Join(req.Directory, req.Entry.Name)

+	// A conditional or preconditioned update is a read-then-write that the
+	// per-path lock below only makes atomic on this filer. Route it to the
+	// entry's ring owner so one filer's lock arbitrates every writer
+	// cluster-wide; is_moved bounds this to one hop.
+	if !req.IsMoved && (conditionIsSet(req.Condition) || len(req.ExpectedExtended) > 0) {
+		var ownerResp *filer_pb.UpdateEntryResponse
+		handled, forwardErr := fs.forwardToWriteOwner(ctx, entryRouteKey(util.FullPath(fullpath)), func(owner pb.ServerAddress) error {
+			glog.V(2).InfofCtx(ctx, "UpdateEntry %s: forwarding to owner %s", fullpath, owner)
+			req.IsMoved = true
+			return pb.WithFilerClient(false, 0, owner, fs.grpcDialOption, func(client filer_pb.SeaweedFilerClient) error {
+				forwarded, e := client.UpdateEntry(ctx, req)
+				if e != nil {
+					return e
+				}
+				ownerResp = forwarded
+				return nil
+			})
+		})
+		if handled {
+			if forwardErr != nil {
+				return &filer_pb.UpdateEntryResponse{}, forwardErr
+			}
+			return ownerResp, nil
+		}
+	}
+
 	// Serialize concurrent mutations to the same path on this filer so the
 	// read (preconditions, garbage diff) and the write are atomic. Callers
 	// route a key's writes to this owner filer, making this local lock
@@ -8,6 +8,7 @@ import (
 	"github.com/seaweedfs/seaweedfs/weed/filer"
 	"github.com/seaweedfs/seaweedfs/weed/pb/filer_pb"
 	"github.com/seaweedfs/seaweedfs/weed/s3api/s3_constants"
+	"google.golang.org/protobuf/proto"
 )

 // conditionIsSet reports whether a condition asks for any check at all.
@@ -82,6 +83,14 @@ func clauseSatisfied(c *filer_pb.WriteCondition_Clause, current *filer.Entry) bo
 		return deadline <= time.Now().Unix()
 	case filer_pb.WriteCondition_IF_CHUNKS_EQUAL:
 		return chunkFidsEqual(current, c.Fids)
+	case filer_pb.WriteCondition_IF_ENTRY_EQUAL:
+		if !exists || c.ExpectedEntry == nil {
+			return !exists && c.ExpectedEntry == nil
+		}
+		// Normalize the expected entry the way FindEntry normalizes the stored
+		// one (e.g. FileSize grows to the chunk extent), or an unchanged entry
+		// can compare unequal.
+		return proto.Equal(current.ToProtoEntry(), filer.FromPbEntry("", c.ExpectedEntry).ToProtoEntry())
 	default:
 		// An unrecognized clause kind (e.g. from a newer client) must not be
 		// treated as satisfied, which would silently bypass the guard. Fail
@@ -193,6 +193,34 @@ func TestWriteConditionUnknownKindFailsClosed(t *testing.T) {
 	}
 }

+// IF_ENTRY_EQUAL compares the expected entry after the same normalization
+// FindEntry applied to the stored one: a raw event entry whose FileSize is
+// still zero must match a stored entry grown to its chunk extent.
+func TestIfEntryEqualNormalizesExpected(t *testing.T) {
+	raw := &filer_pb.Entry{
+		Name:       "f",
+		Attributes: &filer_pb.FuseAttributes{Mtime: 42},
+		Chunks: []*filer_pb.FileChunk{
+			{Fid: &filer_pb.FileId{VolumeId: 3, FileKey: 1, Cookie: 2}, Size: 100},
+		},
+	}
+	stored := filer.FromPbEntry("/d", raw)
+	if stored.FileSize != 100 {
+		t.Fatalf("stored FileSize = %d, want chunk extent 100", stored.FileSize)
+	}
+	cond := one(&filer_pb.WriteCondition_Clause{
+		Kind:          filer_pb.WriteCondition_IF_ENTRY_EQUAL,
+		ExpectedEntry: raw,
+	})
+	if !writeConditionSatisfied(cond, stored) {
+		t.Error("raw expected entry must equal its normalized stored form")
+	}
+	raw.Attributes.Mtime = 43
+	if writeConditionSatisfied(cond, stored) {
+		t.Error("changed expected entry must not equal the stored entry")
+	}
+}
+
 // storedEntryETag prefers the stored Seaweed ETag attribute and falls back to
 // the Md5-derived ETag, matching the S3 gateway.
 func TestStoredEntryETag(t *testing.T) {
Author	SHA1	Message	Date
Chris Lu	b6dd48403e	filer.remote.sync: classify refused stamps by gRPC status only A FailedPrecondition substring in an unrelated error would have been swallowed as a skipped stamp; status.FromError already unwraps.	2026-09-23 17:07:02 -07:00
Chris Lu	8cc3e299fc	filer: compare IF_ENTRY_EQUAL against the normalized expected entry FindEntry grows FileSize to the chunk extent, so a raw event entry with FileSize still zero failed the condition on an unchanged file and the stamp was skipped, letting a replay upload the object again.	2026-09-23 17:07:02 -07:00
Chris Lu	92bd112cfd	filer: route conditional UpdateEntry to the entry's owner filer Two filers locking the same path locally could still pass a stale condition on the non-owner while the owner's entry had moved on. When a condition or expected_extended precondition is set, forward the request to the entry's owner the same way conditional CreateEntry does, with is_moved bounding the hop.	2026-09-23 16:34:38 -07:00
Chris Lu	4f1178e0ba	filer.remote.sync: stamp entries with IF_ENTRY_EQUAL so stale inline content or metadata cannot be restored The IF_CHUNKS_EQUAL guard compared only the chunk fid multiset, so a rewrite that touched inline content or metadata alone still compared equal and the stale snapshot overwrote the live entry. The new clause compares the whole stored entry against the event's entry under the same path lock.	2026-09-23 16:34:29 -07:00
jsas	071d9d2f3c	filer.remote.sync: stamp entries with IF_CHUNKS_EQUAL so a stale write-back cannot delete live chunks updateLocalEntry records the RemoteEntry stamp after an upload by writing the event's entry back with UpdateEntry. The filer deletes every stored chunk absent from an updated entry, so when the file was rewritten while its upload was in flight (or the event is a replay), the stale snapshot deletes the rewrite's chunks: the entry then points at the new fid with no needle behind it, and the rewrite's own upload fails and is skipped as superseded. The stamp write now carries WriteCondition IF_CHUNKS_EQUAL over the event's chunk fids, evaluated by the filer under the path lock. A refused stamp means the filer moved past this event; the superseding event follows in the log and stamps the current entry, so the refusal is logged and skipped like a superseded upload. Reproduction: weed server -filer plus a weed server -s3 remote, remote.mount, filer.remote.sync; hold the remote (docker pause) so one upload stays in flight, rewrite the file through the filer, unpause. Before: the entry's chunk is 404 on every volume server. After: the stale stamp is refused, the rewrite's chunk stays live and reads back after a vacuum.	2026-09-23 16:11:48 -07:00