Revert "s3: listObjectParts return ErrNoSuchUpload if does not exist"

This reverts commit 6cf2e7d493.
file set attribute
2026-10-01 20:26:27 +00:00 · 2022-02-14 01:44:23 -08:00 · 2022-02-14 01:36:10 -08:00 · 2022-02-14 01:09:31 -08:00 · 2022-02-13 23:27:11 -08:00 · 2022-02-13 22:50:44 -08:00
186 changed files with 7908 additions and 2922 deletions
@@ -20,12 +20,12 @@ jobs:
          assets: |
            weed-*

-  build_dev_linux:
+  build_dev_linux_windows:
    needs: cleanup
    runs-on: ubuntu-latest
    strategy:
      matrix:
-        goos: [linux]
+        goos: [linux, windows]
        goarch: [amd64]

    steps:
@@ -68,7 +68,7 @@ jobs:
          asset_name: "weed-${{ env.BUILD_TIME }}-${{ matrix.goos }}-${{ matrix.goarch }}"

  build_dev_darwin:
-    needs: build_dev_linux
+    needs: build_dev_linux_windows
    runs-on: ubuntu-latest
    strategy:
      matrix:
@@ -41,6 +41,7 @@ Your support will be really appreciated by me and other supporters!
 - [Download Binaries for different platforms](https://github.com/chrislusf/seaweedfs/releases/latest)
 - [SeaweedFS on Slack](https://join.slack.com/t/seaweedfs/shared_invite/enQtMzI4MTMwMjU2MzA3LTEyYzZmZWYzOGQ3MDJlZWMzYmI0OTE4OTJiZjJjODBmMzUxNmYwODg0YjY3MTNlMjBmZDQ1NzQ5NDJhZWI2ZmY)
 - [SeaweedFS on Twitter](https://twitter.com/SeaweedFS)
+- [SeaweedFS on Telegram](https://t.me/Seaweedfs) 
 - [SeaweedFS Mailing List](https://groups.google.com/d/forum/seaweedfs)
 - [Wiki Documentation](https://github.com/chrislusf/seaweedfs/wiki)
 - [SeaweedFS White Paper](https://github.com/chrislusf/seaweedfs/wiki/SeaweedFS_Architecture.pdf)
@@ -9,7 +9,7 @@ ENV ROCKSDB_VERSION v6.22.1
 RUN cd /tmp && \
    git clone https://github.com/facebook/rocksdb.git /tmp/rocksdb --depth 1 --single-branch --branch $ROCKSDB_VERSION && \
    cd rocksdb && \
-    make static_lib && \
+    PORTABLE=1 make static_lib && \
    make install-static

 ENV CGO_CFLAGS "-I/tmp/rocksdb/include"
@@ -29,7 +29,7 @@ FROM alpine AS final
 LABEL author="Chris Lu"
 COPY --from=builder /go/bin/weed /usr/bin/
 RUN mkdir -p /etc/seaweedfs
-COPY --from=builder /go/src/github.com/chrislusf/seaweedfs/docker/filer.toml /etc/seaweedfs/filer.toml
+COPY --from=builder /go/src/github.com/chrislusf/seaweedfs/docker/filer_rocksdb.toml /etc/seaweedfs/filer.toml
 COPY --from=builder /go/src/github.com/chrislusf/seaweedfs/docker/entrypoint.sh /entrypoint.sh
 RUN apk add fuse snappy gflags

@@ -50,7 +50,7 @@ EXPOSE 8333
 # webdav server http port
 EXPOSE 7333

-RUN mkdir -p /data/filerldb2
+RUN mkdir -p /data/filer_rocksdb

 VOLUME /data

@@ -0,0 +1,3 @@
+[rocksdb]
+enabled = true
+dir = "/data/filer_rocksdb"
@@ -8,7 +8,6 @@ require (
 	cloud.google.com/go/storage v1.16.1
 	github.com/Azure/azure-pipeline-go v0.2.3
 	github.com/Azure/azure-storage-blob-go v0.14.0
-	github.com/DataDog/zstd v1.3.6-0.20190409195224-796139022798 // indirect
 	github.com/OneOfOne/xxhash v1.2.2
 	github.com/Shopify/sarama v1.23.1
 	github.com/aws/aws-sdk-go v1.35.3
@@ -100,7 +99,7 @@ require (
 	github.com/prometheus/procfs v0.6.0 // indirect
 	github.com/rcrowley/go-metrics v0.0.0-20190826022208-cac0b30c2563 // indirect
 	github.com/remyoudompheng/bigfft v0.0.0-20200410134404-eec4a21b6bb0 // indirect
-	github.com/seaweedfs/fuse v1.2.1
+	github.com/seaweedfs/fuse v1.2.2
 	github.com/seaweedfs/goexif v1.0.2
 	github.com/sirupsen/logrus v1.6.0 // indirect
 	github.com/spaolacci/murmur3 v1.1.0 // indirect
@@ -136,7 +135,7 @@ require (
 	golang.org/x/image v0.0.0-20200119044424-58c23975cae1
 	golang.org/x/net v0.0.0-20210813160813-60bc85c4be6d
 	golang.org/x/oauth2 v0.0.0-20210819190943-2bc19b11175f // indirect
-	golang.org/x/sys v0.0.0-20211216021012-1d35b9e2eb4e
+	golang.org/x/sys v0.0.0-20220111092808-5a964db01320
 	golang.org/x/text v0.3.7 // indirect
 	golang.org/x/tools v0.1.6-0.20210726203631-07bc1bf47fb2
 	golang.org/x/xerrors v0.0.0-20200804184101-5ec99f83aff1 // indirect
@@ -163,20 +162,22 @@ require (
 	modernc.org/token v1.0.0 // indirect
 )

+require (
+	github.com/fluent/fluent-logger-golang v1.8.0
+	github.com/hanwen/go-fuse/v2 v2.1.0
+)
+
 require (
 	cloud.google.com/go/kms v1.0.0 // indirect
-	github.com/apache/thrift v0.15.0 // indirect
+	github.com/DataDog/zstd v1.3.6-0.20190409195224-796139022798 // indirect
 	github.com/d4l3k/messagediff v1.2.1 // indirect
-	github.com/fluent/fluent-logger-golang v1.8.0 // indirect
 	github.com/gogo/protobuf v1.3.2 // indirect
-	github.com/gorilla/websocket v1.4.2 // indirect
 	github.com/jcmturner/aescts/v2 v2.0.0 // indirect
 	github.com/jcmturner/dnsutils/v2 v2.0.0 // indirect
 	github.com/jcmturner/goidentity/v6 v6.0.1 // indirect
 	github.com/jcmturner/rpc/v2 v2.0.2 // indirect
 	github.com/mattn/go-runewidth v0.0.7 // indirect
 	github.com/mattn/go-sqlite3 v2.0.1+incompatible // indirect
-	github.com/miekg/dns v1.1.44 // indirect
 	github.com/philhofer/fwd v1.1.1 // indirect
 	github.com/spf13/pflag v1.0.5 // indirect
 	github.com/tinylib/msgp v1.1.6 // indirect
@@ -124,8 +124,6 @@ github.com/alecthomas/units v0.0.0-20190924025748-f65c72e2690d/go.mod h1:rBZYJk5
 github.com/antihax/optional v1.0.0/go.mod h1:uupD/76wgC+ih3iEmQUL+0Ugr19nfwCT1kdvxnR2qWY=
 github.com/apache/thrift v0.12.0/go.mod h1:cp2SuWMxlEZw2r+iP2GNCdIi4C1qmUzdZFSVb+bacwQ=
 github.com/apache/thrift v0.13.0/go.mod h1:cp2SuWMxlEZw2r+iP2GNCdIi4C1qmUzdZFSVb+bacwQ=
-github.com/apache/thrift v0.15.0 h1:aGvdaR0v1t9XLgjtBYwxcBvBOTMqClzwE26CHOgjW1Y=
-github.com/apache/thrift v0.15.0/go.mod h1:PHK3hniurgQaNMZYaCLEqXKsYK8upmhPbmdP2FXSqgU=
 github.com/armon/circbuf v0.0.0-20150827004946-bbbad097214e/go.mod h1:3U/XgcO3hCbHZ8TKRvWD2dDTCfh9M9ya+I9JpbB7O8o=
 github.com/armon/consul-api v0.0.0-20180202201655-eb2c6b5be1b6/go.mod h1:grANhF5doyWs3UAsr3K4I6qtAmlQcZDesFNEHPZAzj8=
 github.com/armon/go-metrics v0.0.0-20180917152333-f0300d1749da/go.mod h1:Q73ZrmVTwzkszR9V5SSuryQ31EELlFMUz1kKyl939pY=
@@ -281,30 +279,6 @@ github.com/go-stack/stack v1.8.0 h1:5SgMzNM5HxrEjV0ww2lTmX6E2Izsfxas4+YHWRs3Lsk=
 github.com/go-stack/stack v1.8.0/go.mod h1:v0f6uXyyMGvRgIKkXu+yp6POWl0qKG85gN/melR3HDY=
 github.com/go-zookeeper/zk v1.0.2 h1:4mx0EYENAdX/B/rbunjlt5+4RTA/a9SMHBRuSKdGxPM=
 github.com/go-zookeeper/zk v1.0.2/go.mod h1:nOB03cncLtlp4t+UAkGSV+9beXP/akpekBwL+UX1Qcw=
-github.com/gobuffalo/attrs v0.0.0-20190224210810-a9411de4debd/go.mod h1:4duuawTqi2wkkpB4ePgWMaai6/Kc6WEz83bhFwpHzj0=
-github.com/gobuffalo/depgen v0.0.0-20190329151759-d478694a28d3/go.mod h1:3STtPUQYuzV0gBVOY3vy6CfMm/ljR4pABfrTeHNLHUY=
-github.com/gobuffalo/depgen v0.1.0/go.mod h1:+ifsuy7fhi15RWncXQQKjWS9JPkdah5sZvtHc2RXGlg=
-github.com/gobuffalo/envy v1.6.15/go.mod h1:n7DRkBerg/aorDM8kbduw5dN3oXGswK5liaSCx4T5NI=
-github.com/gobuffalo/envy v1.7.0/go.mod h1:n7DRkBerg/aorDM8kbduw5dN3oXGswK5liaSCx4T5NI=
-github.com/gobuffalo/flect v0.1.0/go.mod h1:d2ehjJqGOH/Kjqcoz+F7jHTBbmDb38yXA598Hb50EGs=
-github.com/gobuffalo/flect v0.1.1/go.mod h1:8JCgGVbRjJhVgD6399mQr4fx5rRfGKVzFjbj6RE/9UI=
-github.com/gobuffalo/flect v0.1.3/go.mod h1:8JCgGVbRjJhVgD6399mQr4fx5rRfGKVzFjbj6RE/9UI=
-github.com/gobuffalo/genny v0.0.0-20190329151137-27723ad26ef9/go.mod h1:rWs4Z12d1Zbf19rlsn0nurr75KqhYp52EAGGxTbBhNk=
-github.com/gobuffalo/genny v0.0.0-20190403191548-3ca520ef0d9e/go.mod h1:80lIj3kVJWwOrXWWMRzzdhW3DsrdjILVil/SFKBzF28=
-github.com/gobuffalo/genny v0.1.0/go.mod h1:XidbUqzak3lHdS//TPu2OgiFB+51Ur5f7CSnXZ/JDvo=
-github.com/gobuffalo/genny v0.1.1/go.mod h1:5TExbEyY48pfunL4QSXxlDOmdsD44RRq4mVZ0Ex28Xk=
-github.com/gobuffalo/gitgen v0.0.0-20190315122116-cc086187d211/go.mod h1:vEHJk/E9DmhejeLeNt7UVvlSGv3ziL+djtTr3yyzcOw=
-github.com/gobuffalo/gogen v0.0.0-20190315121717-8f38393713f5/go.mod h1:V9QVDIxsgKNZs6L2IYiGR8datgMhB577vzTDqypH360=
-github.com/gobuffalo/gogen v0.1.0/go.mod h1:8NTelM5qd8RZ15VjQTFkAW6qOMx5wBbW4dSCS3BY8gg=
-github.com/gobuffalo/gogen v0.1.1/go.mod h1:y8iBtmHmGc4qa3urIyo1shvOD8JftTtfcKi+71xfDNE=
-github.com/gobuffalo/logger v0.0.0-20190315122211-86e12af44bc2/go.mod h1:QdxcLw541hSGtBnhUc4gaNIXRjiDppFGaDqzbrBd3v8=
-github.com/gobuffalo/mapi v1.0.1/go.mod h1:4VAGh89y6rVOvm5A8fKFxYG+wIW6LO1FMTG9hnKStFc=
-github.com/gobuffalo/mapi v1.0.2/go.mod h1:4VAGh89y6rVOvm5A8fKFxYG+wIW6LO1FMTG9hnKStFc=
-github.com/gobuffalo/packd v0.0.0-20190315124812-a385830c7fc0/go.mod h1:M2Juc+hhDXf/PnmBANFCqx4DM3wRbgDvnVWeG2RIxq4=
-github.com/gobuffalo/packd v0.1.0/go.mod h1:M2Juc+hhDXf/PnmBANFCqx4DM3wRbgDvnVWeG2RIxq4=
-github.com/gobuffalo/packr/v2 v2.0.9/go.mod h1:emmyGweYTm6Kdper+iywB6YK5YzuKchGtJQZ0Odn4pQ=
-github.com/gobuffalo/packr/v2 v2.2.0/go.mod h1:CaAwI0GPIAv+5wKLtv8Afwl+Cm78K/I/VCm/3ptBN+0=
-github.com/gobuffalo/syncx v0.0.0-20190224160051-33c29581e754/go.mod h1:HhnNqWY95UYwwW3uSASeV7vtgYkT2t16hJgV3AEPUpw=
 github.com/gocql/gocql v0.0.0-20210707082121-9a3953d1826d h1:k544nNVphXK4Yt0FTduvOvCfJabEY/DMkdNw0zpCwBE=
 github.com/gocql/gocql v0.0.0-20210707082121-9a3953d1826d/go.mod h1:3gM2c4D3AnkISwBxGnMMsS8Oy4y2lhbPRsH4xnJrHG8=
 github.com/godbus/dbus/v5 v5.0.4/go.mod h1:xhWf0FNVPg57R7Z0UbKHbJfkEywrmjJnf7w5xrFpKfA=
@@ -431,8 +405,6 @@ github.com/gorilla/sessions v1.2.0 h1:S7P+1Hm5V/AT9cjEcUD5uDaQSX0OE577aCXgoaKpYb
 github.com/gorilla/sessions v1.2.0/go.mod h1:dk2InVEVJ0sfLlnXv9EAgkf6ecYs/i80K/zI+bUmuGM=
 github.com/gorilla/websocket v0.0.0-20170926233335-4201258b820c/go.mod h1:E7qHFY5m1UJ88s3WnNqhKjPHQ0heANvMoAMk2YaljkQ=
 github.com/gorilla/websocket v1.4.0/go.mod h1:E7qHFY5m1UJ88s3WnNqhKjPHQ0heANvMoAMk2YaljkQ=
-github.com/gorilla/websocket v1.4.2 h1:+/TMaTYc4QFitKJxsQ7Yye35DkWvkdLcvGKqM+x0Ufc=
-github.com/gorilla/websocket v1.4.2/go.mod h1:YR8l580nyteQvAITg2hZ9XVh4b55+EU/adAjf1fMHhE=
 github.com/grpc-ecosystem/go-grpc-middleware v1.0.0/go.mod h1:FiyG127CGDf3tlThmgyCl78X/SZQqEOJBCDaAfeWzPs=
 github.com/grpc-ecosystem/go-grpc-middleware v1.0.1-0.20190118093823-f849b5445de4/go.mod h1:FiyG127CGDf3tlThmgyCl78X/SZQqEOJBCDaAfeWzPs=
 github.com/grpc-ecosystem/go-grpc-middleware v1.1.0 h1:THDBEeQ9xZ8JEaCLyLQqXMMdRqNr0QAUJTIkQAUtFjg=
@@ -443,6 +415,10 @@ github.com/grpc-ecosystem/grpc-gateway v1.9.5/go.mod h1:vNeuVxBJEsws4ogUvrchl83t
 github.com/grpc-ecosystem/grpc-gateway v1.16.0/go.mod h1:BDjrQk3hbvj6Nolgz8mAMFbcEtjT1g+wF4CSlocrBnw=
 github.com/hailocab/go-hostpool v0.0.0-20160125115350-e80d13ce29ed h1:5upAirOpQc1Q53c0bnx2ufif5kANL7bfZWcc6VJWJd8=
 github.com/hailocab/go-hostpool v0.0.0-20160125115350-e80d13ce29ed/go.mod h1:tMWxXQ9wFIaZeTI9F+hmhFiGpFmhOHzyShyFUhRm0H4=
+github.com/hanwen/go-fuse v1.0.0 h1:GxS9Zrn6c35/BnfiVsZVWmsG803xwE7eVRDvcf/BEVc=
+github.com/hanwen/go-fuse v1.0.0/go.mod h1:unqXarDXqzAk0rt98O2tVndEPIpUgLD9+rwFisZH3Ok=
+github.com/hanwen/go-fuse/v2 v2.1.0 h1:+32ffteETaLYClUj0a3aHjZ1hOPxxaNEHiZiujuDaek=
+github.com/hanwen/go-fuse/v2 v2.1.0/go.mod h1:oRyA5eK+pvJyv5otpO/DgccS8y/RvYMaO00GgRLGryc=
 github.com/hashicorp/consul/api v1.3.0/go.mod h1:MmDNSzIMUjNpY/mQ398R4bk2FnqQLoPndWW5VkKPlCE=
 github.com/hashicorp/consul/sdk v0.3.0/go.mod h1:VKf9jXwCTEY1QZP2MOLRhb5i/I/ssyNV1vwHyQBF0x8=
 github.com/hashicorp/errwrap v1.0.0 h1:hLrqtEDnRye3+sgx6z4qVLNuviH3MR5aQ0ykNJa/UYA=
@@ -516,16 +492,11 @@ github.com/karlseguin/ccache/v2 v2.0.7 h1:y5Pfi4eiyYCOD6LS/Kj+o6Nb4M5Ngpw9qFQs+v
 github.com/karlseguin/ccache/v2 v2.0.7/go.mod h1:2BDThcfQMf/c0jnZowt16eW405XIqZPavt+HoYEtcxQ=
 github.com/karlseguin/expect v1.0.2-0.20190806010014-778a5f0c6003 h1:vJ0Snvo+SLMY72r5J4sEfkuE7AFbixEP2qRbEcum/wA=
 github.com/karlseguin/expect v1.0.2-0.20190806010014-778a5f0c6003/go.mod h1:zNBxMY8P21owkeogJELCLeHIt+voOSduHYTFUbwRAV8=
-github.com/karrick/godirwalk v1.8.0/go.mod h1:H5KPZjojv4lE+QYImBI8xVtrBRgYrIVsaRPx4tDPEn4=
-github.com/karrick/godirwalk v1.10.3/go.mod h1:RoGL9dQei4vP9ilrpETWE8CLOZ1kiN0LhBygSwrAsHA=
 github.com/kballard/go-shellquote v0.0.0-20180428030007-95032a82bc51 h1:Z9n2FFNUXsshfwJMBgNA0RU6/i7WVaAegv3PtuIHPMs=
 github.com/kballard/go-shellquote v0.0.0-20180428030007-95032a82bc51/go.mod h1:CzGEWj7cYgsdH8dAjBGEr58BoE7ScuLd+fwFZ44+/x8=
 github.com/kisielk/errcheck v1.1.0/go.mod h1:EZBBE59ingxPouuu3KfxchcWSUPOHkagtvWXihfKN4Q=
 github.com/kisielk/errcheck v1.5.0/go.mod h1:pFxgyoBC7bSaBwPgfKdkLd5X25qrDl4LWUI2bnpBCr8=
 github.com/kisielk/gotool v1.0.0/go.mod h1:XhKaO+MFFWcvkIS/tQcRk01m1F5IRFswLeQ+oQHNcck=
-github.com/klauspost/compress v1.9.5/go.mod h1:RyIbtBH6LamlWaDj8nUwkbUhJ87Yi3uG0guNDohfE1A=
-github.com/klauspost/compress v1.10.9 h1:pPRt1Z78crspaHISkpSSHjDlx+Tt9suHe519dsI0vF4=
-github.com/klauspost/compress v1.10.9/go.mod h1:aoV0uJVorq1K+umq18yTdKaF57EivdYsUV+/s2qKfXs=
 github.com/klauspost/compress v1.13.6 h1:P76CopJELS0TiO2mebmnzgWaajssP/EszplttgQxcgc=
 github.com/klauspost/compress v1.13.6/go.mod h1:/3/Vjq9QcHkK5uEr5lBEmyoZ1iFhe47etQ6QUkpK6sk=
 github.com/klauspost/cpuid v1.2.1 h1:vJi+O/nMdFt0vqm8NZBI6wzALWdA2X+egi0ogNyrC/w=
@@ -533,7 +504,6 @@ github.com/klauspost/cpuid v1.2.1/go.mod h1:Pj4uuM528wm8OyEC2QMXAi2YiTZ96dNQPGgo
 github.com/klauspost/reedsolomon v1.9.2 h1:E9CMS2Pqbv+C7tsrYad4YC9MfhnMVWhMRsTi7U0UB18=
 github.com/klauspost/reedsolomon v1.9.2/go.mod h1:CwCi+NUr9pqSVktrkN+Ondf06rkhYZ/pcNv7fu+8Un4=
 github.com/konsorten/go-windows-terminal-sequences v1.0.1/go.mod h1:T0+1ngSBFLxvqU3pZ+m/2kptfBszLMUkC4ZK/EgS/cQ=
-github.com/konsorten/go-windows-terminal-sequences v1.0.2/go.mod h1:T0+1ngSBFLxvqU3pZ+m/2kptfBszLMUkC4ZK/EgS/cQ=
 github.com/konsorten/go-windows-terminal-sequences v1.0.3 h1:CE8S1cTafDpPvMhIxNJKvHsGVBgn1xWYf1NbHQhywc8=
 github.com/konsorten/go-windows-terminal-sequences v1.0.3/go.mod h1:T0+1ngSBFLxvqU3pZ+m/2kptfBszLMUkC4ZK/EgS/cQ=
 github.com/kr/fs v0.1.0/go.mod h1:FFnZGqtBN9Gxj7eW1uZ42v5BccTP0vu6NEaFoC2HwRg=
@@ -546,6 +516,8 @@ github.com/kr/text v0.1.0 h1:45sCR5RtlFHMR4UwH9sdQ5TC8v0qDQCHnXt+kaKSTVE=
 github.com/kr/text v0.1.0/go.mod h1:4Jbv+DJW3UT/LiOwJeYQe1efqtUx/iVham/4vfdArNI=
 github.com/kurin/blazer v0.5.3 h1:SAgYv0TKU0kN/ETfO5ExjNAPyMt2FocO2s/UlCHfjAk=
 github.com/kurin/blazer v0.5.3/go.mod h1:4FCXMUWo9DllR2Do4TtBd377ezyAJ51vB5uTBjt0pGU=
+github.com/kylelemons/godebug v0.0.0-20170820004349-d65d576e9348 h1:MtvEpTB6LX3vkb4ax0b5D2DHbNAUsen0Gx5wZoq3lV4=
+github.com/kylelemons/godebug v0.0.0-20170820004349-d65d576e9348/go.mod h1:B69LEHPfb2qLo0BaaOLcbitczOKLWTsrBG9LczfCD4k=
 github.com/lib/pq v1.1.1/go.mod h1:5WUZQaWbwv1U+lTReE5YruASi9Al49XbQIvNi/34Woo=
 github.com/lib/pq v1.10.0 h1:Zx5DJFEYQXio93kgXnQ09fXNiUKsqv4OUEu2UtGcB1E=
 github.com/lib/pq v1.10.0/go.mod h1:AlVN5x4E4T544tWzH6hKfbfQvm3HdbOxrmggDNAPY9o=
@@ -559,8 +531,6 @@ github.com/magiconair/properties v1.8.1 h1:ZC2Vc7/ZFkGmsVC9KvOjumD+G5lXy2RtTKyzR
 github.com/magiconair/properties v1.8.1/go.mod h1:PppfXfuXeibc/6YijjN8zIbojt8czPbwD3XqdrwzmxQ=
 github.com/mailru/easyjson v0.7.1 h1:mdxE1MF9o53iCb2Ghj1VfWvh7ZOwHpnVG/xwXrV90U8=
 github.com/mailru/easyjson v0.7.1/go.mod h1:KAzv3t3aY1NaHWoQz1+4F1ccyAH66Jk7yos7ldAVICs=
-github.com/markbates/oncer v0.0.0-20181203154359-bf2de49a0be2/go.mod h1:Ld9puTsIW75CHf65OeIOkyKbteujpZVXDpWK6YGZbxE=
-github.com/markbates/safe v1.0.1/go.mod h1:nAqgmRi7cY2nqMc92/bSEeQA+R4OheNU2T1kNSCBdG0=
 github.com/mattn/go-colorable v0.0.9/go.mod h1:9vuHe8Xs5qXnSaW/c/ABM9alt+Vo+STaOChaDxuIBZU=
 github.com/mattn/go-ieproxy v0.0.0-20190610004146-91bb50d98149/go.mod h1:31jz6HNzdxOmlERGGEc4v/dMssOfmp2p5bT/okiKFFc=
 github.com/mattn/go-ieproxy v0.0.0-20190702010315-6dee0af9227d/go.mod h1:31jz6HNzdxOmlERGGEc4v/dMssOfmp2p5bT/okiKFFc=
@@ -580,8 +550,6 @@ github.com/mattn/go-sqlite3 v2.0.1+incompatible/go.mod h1:FPy6KqzDD04eiIsT53CuJW
 github.com/matttproud/golang_protobuf_extensions v1.0.1 h1:4hp9jkHxhMHkqkrB3Ix0jegS5sx/RkqARlsWZ6pIwiU=
 github.com/matttproud/golang_protobuf_extensions v1.0.1/go.mod h1:D8He9yQNgCq6Z5Ld7szi9bcBfOoFv/3dc6xSMkL2PC0=
 github.com/miekg/dns v1.0.14/go.mod h1:W1PPwlIAgtquWBMBEV9nkV9Cazfe8ScdGz/Lj7v3Nrg=
-github.com/miekg/dns v1.1.44 h1:4rpqcegYPVkvIeVhITrKP1sRR3KjfRc1nrOPMUZmLyc=
-github.com/miekg/dns v1.1.44/go.mod h1:e3IlAVfNqAllflbibAZEWOXOQ+Ynzk/dDozDxY7XnME=
 github.com/mitchellh/cli v1.0.0/go.mod h1:hNIlj7HEI86fIcpObd7a0FcrxTWetlwJDGcceTlRvqc=
 github.com/mitchellh/go-homedir v1.0.0/go.mod h1:SfyaCUpYCn1Vlf4IUYiD9fPX4A5wJrkLzIz1N1q0pr0=
 github.com/mitchellh/go-homedir v1.1.0/go.mod h1:SfyaCUpYCn1Vlf4IUYiD9fPX4A5wJrkLzIz1N1q0pr0=
@@ -721,25 +689,19 @@ github.com/remyoudompheng/bigfft v0.0.0-20200410134404-eec4a21b6bb0 h1:OdAsTTz6O
 github.com/remyoudompheng/bigfft v0.0.0-20200410134404-eec4a21b6bb0/go.mod h1:qqbHyh8v60DhA7CoWK5oRCqLrMHRGoxYCSS9EjAz6Eo=
 github.com/rogpeppe/fastuuid v0.0.0-20150106093220-6724a57986af/go.mod h1:XWv6SoW27p1b0cqNHllgS5HIMJraePCO15w5zCzIWYg=
 github.com/rogpeppe/fastuuid v1.2.0/go.mod h1:jVj6XXZzXRy/MSR5jhDC/2q6DgLz+nrA6LYCDYWNEvQ=
-github.com/rogpeppe/go-internal v1.1.0/go.mod h1:M8bDsm7K2OlrFYOpmOWEs/qY81heoFRclV5y23lUDJ4=
-github.com/rogpeppe/go-internal v1.2.2/go.mod h1:M8bDsm7K2OlrFYOpmOWEs/qY81heoFRclV5y23lUDJ4=
 github.com/rogpeppe/go-internal v1.3.0/go.mod h1:M8bDsm7K2OlrFYOpmOWEs/qY81heoFRclV5y23lUDJ4=
 github.com/russross/blackfriday/v2 v2.0.1/go.mod h1:+Rmxgy9KzJVeS9/2gXHxylqXiyQDYRxCVz55jmeOWTM=
 github.com/ryanuber/columnize v0.0.0-20160712163229-9b3edd62028f/go.mod h1:sm1tb6uqfes/u+d4ooFouqFdy9/2g9QGwK3SQygK0Ts=
 github.com/samuel/go-zookeeper v0.0.0-20190923202752-2cc03de413da/go.mod h1:gi+0XIa01GRL2eRQVjQkKGqKF3SF9vZR/HnPullcV2E=
 github.com/sean-/seed v0.0.0-20170313163322-e2103e2c3529/go.mod h1:DxrIzT+xaE7yg65j358z/aeFdxmN0P9QXhEzd20vsDc=
-github.com/seaweedfs/fuse v1.2.0 h1:m4Ar3I0TY/Tzyb80lfJiXUKFcMBHqA5e/cZwxQtilZo=
-github.com/seaweedfs/fuse v1.2.0/go.mod h1:iwbDQv5BZACY54r6AO/6xsLNuMaYcBKSkLTZVfmK594=
-github.com/seaweedfs/fuse v1.2.1 h1:920YH3+PnIqpNLo2lsZDTUJHUPLXnxgQGEXoVb/yMno=
-github.com/seaweedfs/fuse v1.2.1/go.mod h1:iwbDQv5BZACY54r6AO/6xsLNuMaYcBKSkLTZVfmK594=
+github.com/seaweedfs/fuse v1.2.2 h1:01l8OjIdyATRNqVc/gDPgFobuC8ubQF3hRKOPColROw=
+github.com/seaweedfs/fuse v1.2.2/go.mod h1:iwbDQv5BZACY54r6AO/6xsLNuMaYcBKSkLTZVfmK594=
 github.com/seaweedfs/goexif v1.0.2 h1:p+rTXYdQ2mgxd+1JaTrQ9N8DvYuw9UH9xgYmJ+Bb29E=
 github.com/seaweedfs/goexif v1.0.2/go.mod h1:MrKs5LK0HXdffrdCZrW3OIMegL2xXpC6ThLyXMyjdrk=
 github.com/secsy/goftp v0.0.0-20190720192957-f31499d7c79a h1:C6IhVTxNkhlb0tlCB6JfHOUv1f0xHPK7V8X4HlJZEJw=
 github.com/secsy/goftp v0.0.0-20190720192957-f31499d7c79a/go.mod h1:MnkX001NG75g3p8bhFycnyIjeQoOjGL6CEIsdE/nKSY=
 github.com/shurcooL/sanitized_anchor_name v1.0.0/go.mod h1:1NzhyTcUVG4SuEtjjoZeVRXNmyL/1OwPU0+IJeTBvfc=
 github.com/sirupsen/logrus v1.2.0/go.mod h1:LxeOpSwHxABJmUn/MG1IvRgCAasNZTLOkJPxbbu5VWo=
-github.com/sirupsen/logrus v1.4.0/go.mod h1:LxeOpSwHxABJmUn/MG1IvRgCAasNZTLOkJPxbbu5VWo=
-github.com/sirupsen/logrus v1.4.1/go.mod h1:ni0Sbl8bgC9z8RoU9G6nDWqqs/fq4eDPysMBDgk/93Q=
 github.com/sirupsen/logrus v1.4.2/go.mod h1:tLMulIdttU9McNUspp0xgXVQah82FyeX6MwdIuYE2rE=
 github.com/sirupsen/logrus v1.5.0/go.mod h1:+F7Ogzej0PZc/94MaYx/nvG9jOFMD2osvC3s+Squfpo=
 github.com/sirupsen/logrus v1.6.0 h1:UBcNElsrwanuuMsnGSlYmtmgbb23qDR5dG+6X6Oo89I=
@@ -790,17 +752,11 @@ github.com/stvp/tempredis v0.0.0-20181119212430-b82af8480203 h1:QVqDTf3h2WHt08Yu
 github.com/stvp/tempredis v0.0.0-20181119212430-b82af8480203/go.mod h1:oqN97ltKNihBbwlX8dLpwxCl3+HnXKV/R0e+sRLd9C8=
 github.com/syndtr/goleveldb v1.0.1-0.20190318030020-c3a204f8e965 h1:1oFLiOyVl+W7bnBzGhf7BbIv9loSFQcieWWYIjLqcAw=
 github.com/syndtr/goleveldb v1.0.1-0.20190318030020-c3a204f8e965/go.mod h1:9OrXJhf154huy1nPWmuSrkgjPUtUNhA+Zmy+6AESzuA=
-github.com/tidwall/gjson v1.8.1 h1:8j5EE9Hrh3l9Od1OIEDAb7IpezNA20UdRngNAj5N0WU=
-github.com/tidwall/gjson v1.8.1/go.mod h1:5/xDoumyyDNerp2U36lyolv46b3uF/9Bu6OfyQ9GImk=
 github.com/tidwall/gjson v1.10.2 h1:APbLGOM0rrEkd8WBw9C24nllro4ajFuJu0Sc9hRz8Bo=
 github.com/tidwall/gjson v1.10.2/go.mod h1:/wbyibRr2FHMks5tjHJ5F8dMZh3AcwJEMf5vlfC0lxk=
-github.com/tidwall/match v1.0.3 h1:FQUVvBImDutD8wJLN6c5eMzWtjgONK9MwIBCOrUJKeE=
-github.com/tidwall/match v1.0.3/go.mod h1:eRSPERbgtNPcGhD8UCthc6PmLEQXEWd3PRB5JTxsfmM=
 github.com/tidwall/match v1.1.1 h1:+Ho715JplO36QYgwN9PGYNhgZvoUSc9X2c80KVTi+GA=
 github.com/tidwall/match v1.1.1/go.mod h1:eRSPERbgtNPcGhD8UCthc6PmLEQXEWd3PRB5JTxsfmM=
 github.com/tidwall/pretty v1.0.0/go.mod h1:XNkn88O1ChpSDQmQeStsy+sBenx6DDtFZJxhVysOjyk=
-github.com/tidwall/pretty v1.1.0 h1:K3hMW5epkdAVwibsQEfR/7Zj0Qgt4DxtNumTq/VloO8=
-github.com/tidwall/pretty v1.1.0/go.mod h1:XNkn88O1ChpSDQmQeStsy+sBenx6DDtFZJxhVysOjyk=
 github.com/tidwall/pretty v1.2.0 h1:RWIZEg2iJ8/g6fDDYzMpobmaoGh5OLl4AXtGUGPcqCs=
 github.com/tidwall/pretty v1.2.0/go.mod h1:ITEVvHYasfjBbM0u2Pg8T2nJnzm8xPwvNhhsoaGGjNU=
 github.com/tinylib/msgp v1.1.6 h1:i+SbKraHhnrf9M5MYmvQhFnbLhAXSDWF8WWsuyRdocw=
@@ -851,8 +807,6 @@ go.etcd.io/etcd/client/pkg/v3 v3.5.0 h1:2aQv6F436YnN7I4VbI8PPYrBhu+SmrTaADcf8Mi/
 go.etcd.io/etcd/client/pkg/v3 v3.5.0/go.mod h1:IJHfcCEKxYu1Os13ZdwCwIUTUVGYTSAM3YSwc9/Ac1g=
 go.etcd.io/etcd/client/v3 v3.5.0 h1:62Eh0XOro+rDwkrypAGDfgmNh5Joq+z+W9HZdlXMzek=
 go.etcd.io/etcd/client/v3 v3.5.0/go.mod h1:AIKXXVX/DQXtfTEqBryiLTUXwON+GuvO6Z7lLS/oTh0=
-go.mongodb.org/mongo-driver v1.7.0 h1:hHrvOBWlWB2c7+8Gh/Xi5jj82AgidK/t7KVXBZ+IyUA=
-go.mongodb.org/mongo-driver v1.7.0/go.mod h1:Q4oFMbo1+MSNqICAdYMlC/zSTrwCogR4R8NzkI+yfU8=
 go.mongodb.org/mongo-driver v1.8.0 h1:R/P/JJzu8LJvJ1lDfph9GLNIKQxEtIHFfnUUUve35zY=
 go.mongodb.org/mongo-driver v1.8.0/go.mod h1:0sQWfOeY63QTntERDJJ/0SuKK0T1uVSgKCuAROlKEPY=
 go.opencensus.io v0.15.0/go.mod h1:UffZAU+4sDEINUGP/B7UfBBkq4fqLu9zXAX7ke6CHW0=
@@ -896,7 +850,6 @@ golang.org/x/crypto v0.0.0-20181029021203-45a5f77698d3/go.mod h1:6SG95UA2DQfeDnf
 golang.org/x/crypto v0.0.0-20181203042331-505ab145d0a9/go.mod h1:6SG95UA2DQfeDnfUPMdvaQW0Q7yPrPDi9nlGo2tz2b4=
 golang.org/x/crypto v0.0.0-20190308221718-c2843e01d9a2/go.mod h1:djNgcEr1/C05ACkg1iLfiJU5Ep61QUkGW8qpdssI0+w=
 golang.org/x/crypto v0.0.0-20190404164418-38d8ce5564a5/go.mod h1:WFFai1msRO1wXaEeE5yQxYXgSfI8pQAWXbQop6sCtWE=
-golang.org/x/crypto v0.0.0-20190422162423-af44ce270edf/go.mod h1:WFFai1msRO1wXaEeE5yQxYXgSfI8pQAWXbQop6sCtWE=
 golang.org/x/crypto v0.0.0-20190510104115-cbcb75029529/go.mod h1:yigFU9vqHzYiE8UmvKecakEJjdnWj3jj499lnFckfCI=
 golang.org/x/crypto v0.0.0-20190530122614-20be4c3c3ed5/go.mod h1:yigFU9vqHzYiE8UmvKecakEJjdnWj3jj499lnFckfCI=
 golang.org/x/crypto v0.0.0-20190605123033-f99c8df09eb5/go.mod h1:yigFU9vqHzYiE8UmvKecakEJjdnWj3jj499lnFckfCI=
@@ -905,7 +858,6 @@ golang.org/x/crypto v0.0.0-20190820162420-60c769a6c586/go.mod h1:yigFU9vqHzYiE8U
 golang.org/x/crypto v0.0.0-20191011191535-87dc89f01550/go.mod h1:yigFU9vqHzYiE8UmvKecakEJjdnWj3jj499lnFckfCI=
 golang.org/x/crypto v0.0.0-20191206172530-e9b2fee46413/go.mod h1:LzIPMQfyMNhhGPhUkYOs5KpL4U8rLKemX1yGLhDgUto=
 golang.org/x/crypto v0.0.0-20200117160349-530e935923ad/go.mod h1:LzIPMQfyMNhhGPhUkYOs5KpL4U8rLKemX1yGLhDgUto=
-golang.org/x/crypto v0.0.0-20200302210943-78000ba7a073/go.mod h1:LzIPMQfyMNhhGPhUkYOs5KpL4U8rLKemX1yGLhDgUto=
 golang.org/x/crypto v0.0.0-20200323165209-0ec3e9974c59/go.mod h1:LzIPMQfyMNhhGPhUkYOs5KpL4U8rLKemX1yGLhDgUto=
 golang.org/x/crypto v0.0.0-20200604202706-70a84ac30bf9/go.mod h1:LzIPMQfyMNhhGPhUkYOs5KpL4U8rLKemX1yGLhDgUto=
 golang.org/x/crypto v0.0.0-20200622213623-75b288015ac9/go.mod h1:LzIPMQfyMNhhGPhUkYOs5KpL4U8rLKemX1yGLhDgUto=
@@ -1001,7 +953,6 @@ golang.org/x/net v0.0.0-20210226172049-e18ecbb05110/go.mod h1:m0MpNAwzfU5UDzcl9v
 golang.org/x/net v0.0.0-20210316092652-d523dce5a7f4/go.mod h1:RBQZq4jEuRlivfhVLdyRGr576XBO4/greRjx4P4O3yc=
 golang.org/x/net v0.0.0-20210405180319-a5a99cb37ef4/go.mod h1:p54w0d4576C0XHj96bSt6lcn1PtDYWL6XObtHCRCNQM=
 golang.org/x/net v0.0.0-20210503060351-7fd8e65b6420/go.mod h1:9nx3DQGgdP8bBQD5qxJ1jj9UTztislL4KSBs9R2vV5Y=
-golang.org/x/net v0.0.0-20210726213435-c6fcb2dbf985/go.mod h1:9nx3DQGgdP8bBQD5qxJ1jj9UTztislL4KSBs9R2vV5Y=
 golang.org/x/net v0.0.0-20210813160813-60bc85c4be6d h1:LO7XpTYMwTqxjLcGWPijK3vRXg1aWdlNOVOHRq45d7c=
 golang.org/x/net v0.0.0-20210813160813-60bc85c4be6d/go.mod h1:9nx3DQGgdP8bBQD5qxJ1jj9UTztislL4KSBs9R2vV5Y=
 golang.org/x/oauth2 v0.0.0-20180821212333-d2e6202438be/go.mod h1:N/0e6XlmueqKjAGxoOufVs8QHGRruUQn6yWY3a++T0U=
@@ -1025,7 +976,6 @@ golang.org/x/sync v0.0.0-20180314180146-1d60e4601c6f/go.mod h1:RxMgew5VJxzue5/jJ
 golang.org/x/sync v0.0.0-20181108010431-42b317875d0f/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
 golang.org/x/sync v0.0.0-20181221193216-37e7f081c4d4/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
 golang.org/x/sync v0.0.0-20190227155943-e225da77a7e6/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
-golang.org/x/sync v0.0.0-20190412183630-56d357773e84/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
 golang.org/x/sync v0.0.0-20190423024810-112230192c58/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
 golang.org/x/sync v0.0.0-20190911185100-cd5d95a43a6e/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
 golang.org/x/sync v0.0.0-20200317015054-43a5402ce75a/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
@@ -1046,11 +996,9 @@ golang.org/x/sys v0.0.0-20190215142949-d0b11bdaac8a/go.mod h1:STP8DvDyc/dI5b8T5h
 golang.org/x/sys v0.0.0-20190312061237-fead79001313/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
 golang.org/x/sys v0.0.0-20190403152447-81d4e9dc473e/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
 golang.org/x/sys v0.0.0-20190412213103-97732733099d/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
-golang.org/x/sys v0.0.0-20190419153524-e8e3143a4f4a/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
 golang.org/x/sys v0.0.0-20190422165155-953cdadca894/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
 golang.org/x/sys v0.0.0-20190502145724-3ef323f4f1fd/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
 golang.org/x/sys v0.0.0-20190507160741-ecd444e8653b/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
-golang.org/x/sys v0.0.0-20190531175056-4c3a928424d2/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
 golang.org/x/sys v0.0.0-20190606165138-5da285871e9c/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
 golang.org/x/sys v0.0.0-20190624142023-c5567b49c5d0/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
 golang.org/x/sys v0.0.0-20190726091711-fc99dfbffb4e/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
@@ -1110,10 +1058,9 @@ golang.org/x/sys v0.0.0-20210630005230-0f9fa26af87c/go.mod h1:oPkhp1MJrh7nUepCBc
 golang.org/x/sys v0.0.0-20210806184541-e5e7981a1069/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.0.0-20210817142637-7d9622a276b7/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.0.0-20210823070655-63515b42dcdf/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
-golang.org/x/sys v0.0.0-20210908233432-aa78b53d3365 h1:6wSTsvPddg9gc/mVEEyk9oOAoxn+bT4Z9q1zx+4RwA4=
 golang.org/x/sys v0.0.0-20210908233432-aa78b53d3365/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
-golang.org/x/sys v0.0.0-20211216021012-1d35b9e2eb4e h1:fLOSk5Q00efkSvAm+4xcoXD+RRmLmmulPn5I3Y9F2EM=
-golang.org/x/sys v0.0.0-20211216021012-1d35b9e2eb4e/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
+golang.org/x/sys v0.0.0-20220111092808-5a964db01320 h1:0jf+tOCoZ3LyutmCOWpVni1chK4VfFLhRsDK7MhqGRY=
+golang.org/x/sys v0.0.0-20220111092808-5a964db01320/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/term v0.0.0-20201117132131-f5c789dd3221/go.mod h1:Nr5EML6q2oocZ2LXRh80K7BxOlk5/8JxuGnuhpl+muw=
 golang.org/x/term v0.0.0-20201126162022-7de9c90e9dd1/go.mod h1:bj7SfCRtBDWHUb9snDiAeCFNEtKQo2Wmx5Cou7ajbmo=
 golang.org/x/text v0.0.0-20170915032832-14c0d48ead0c/go.mod h1:NqM8EUOU14njkJ3fqMW+pc6Ldnwhi/IjpwHt7yyuwOQ=
@@ -1123,7 +1070,6 @@ golang.org/x/text v0.3.2/go.mod h1:bEr9sfX3Q8Zfm5fL9x+3itogRgK3+ptLWKqgva+5dAk=
 golang.org/x/text v0.3.3/go.mod h1:5Zoc/QRtKVWzQhOtBMvqHzDpF6irO9z98xDceosuGiQ=
 golang.org/x/text v0.3.4/go.mod h1:5Zoc/QRtKVWzQhOtBMvqHzDpF6irO9z98xDceosuGiQ=
 golang.org/x/text v0.3.5/go.mod h1:5Zoc/QRtKVWzQhOtBMvqHzDpF6irO9z98xDceosuGiQ=
-golang.org/x/text v0.3.6 h1:aRYxNxv6iGQlyVaZmk6ZgYEDa+Jg18DxebPSrd6bg1M=
 golang.org/x/text v0.3.6/go.mod h1:5Zoc/QRtKVWzQhOtBMvqHzDpF6irO9z98xDceosuGiQ=
 golang.org/x/text v0.3.7 h1:olpwvP2KacW1ZWvsR7uQhoyTYvKAupfQrRGBFM352Gk=
 golang.org/x/text v0.3.7/go.mod h1:u+2+/6zg+i71rQMx5EYifcz6MCKuco9NR6JIITiCfzQ=
@@ -1140,9 +1086,6 @@ golang.org/x/tools v0.0.0-20190311212946-11955173bddd/go.mod h1:LCzVGOaR6xXOjkQ3
 golang.org/x/tools v0.0.0-20190312151545-0bb0c0a6e846/go.mod h1:LCzVGOaR6xXOjkQ3onu1FJEFr0SW1gC7cKk1uF8kGRs=
 golang.org/x/tools v0.0.0-20190312170243-e65039ee4138/go.mod h1:LCzVGOaR6xXOjkQ3onu1FJEFr0SW1gC7cKk1uF8kGRs=
 golang.org/x/tools v0.0.0-20190328211700-ab21143f2384/go.mod h1:LCzVGOaR6xXOjkQ3onu1FJEFr0SW1gC7cKk1uF8kGRs=
-golang.org/x/tools v0.0.0-20190329151228-23e29df326fe/go.mod h1:LCzVGOaR6xXOjkQ3onu1FJEFr0SW1gC7cKk1uF8kGRs=
-golang.org/x/tools v0.0.0-20190416151739-9c9e1878f421/go.mod h1:LCzVGOaR6xXOjkQ3onu1FJEFr0SW1gC7cKk1uF8kGRs=
-golang.org/x/tools v0.0.0-20190420181800-aa740d480789/go.mod h1:LCzVGOaR6xXOjkQ3onu1FJEFr0SW1gC7cKk1uF8kGRs=
 golang.org/x/tools v0.0.0-20190422233926-fe54fb35175b/go.mod h1:LCzVGOaR6xXOjkQ3onu1FJEFr0SW1gC7cKk1uF8kGRs=
 golang.org/x/tools v0.0.0-20190425150028-36563e24a262/go.mod h1:RgjU9mgBXZiqYHBnxXauZ1Gv1EHHAz9KjViQ78xBX0Q=
 golang.org/x/tools v0.0.0-20190506145303-2d16b83fe98c/go.mod h1:RgjU9mgBXZiqYHBnxXauZ1Gv1EHHAz9KjViQ78xBX0Q=
@@ -1201,7 +1144,6 @@ golang.org/x/tools v0.1.1/go.mod h1:o0xws9oXOQQZyjljx8fwUC0k7L1pTE6eaCbjGeHmOkk=
 golang.org/x/tools v0.1.2/go.mod h1:o0xws9oXOQQZyjljx8fwUC0k7L1pTE6eaCbjGeHmOkk=
 golang.org/x/tools v0.1.3/go.mod h1:o0xws9oXOQQZyjljx8fwUC0k7L1pTE6eaCbjGeHmOkk=
 golang.org/x/tools v0.1.4/go.mod h1:o0xws9oXOQQZyjljx8fwUC0k7L1pTE6eaCbjGeHmOkk=
-golang.org/x/tools v0.1.5 h1:ouewzE6p+/VEB31YYnTbEJdi8pFqKp4P4n85vwo3DHA=
 golang.org/x/tools v0.1.5/go.mod h1:o0xws9oXOQQZyjljx8fwUC0k7L1pTE6eaCbjGeHmOkk=
 golang.org/x/tools v0.1.6-0.20210726203631-07bc1bf47fb2 h1:BonxutuHCTL0rBDnZlKjpGIQFTjyUVTexFOdWkB6Fg0=
 golang.org/x/tools v0.1.6-0.20210726203631-07bc1bf47fb2/go.mod h1:o0xws9oXOQQZyjljx8fwUC0k7L1pTE6eaCbjGeHmOkk=
@@ -1,5 +1,5 @@
 apiVersion: v1
 description: SeaweedFS
 name: seaweedfs
-appVersion: "2.85"
-version: "2.85"
+appVersion: "2.88"
+version: "2.88"
@@ -40,7 +40,7 @@ spec:
      imagePullSecrets:
        - name: {{ .Values.global.imagePullSecrets }}
      {{- end }}
-      terminationGracePeriodSeconds: 10
+      terminationGracePeriodSeconds: 150
      {{- if .Values.volume.priorityClassName }}
      priorityClassName: {{ .Values.volume.priorityClassName | quote }}
      {{- end }}
@@ -114,6 +114,7 @@ message Entry {
    bytes content = 9; // if not empty, the file content

    RemoteEntry remote_entry = 10;
+    int64 quota = 11; // for bucket only. Positive/Negative means enabled/disabled.
 }

 message FullEntry {
@@ -23,7 +23,7 @@
        <dependency>
            <groupId>org.apache.hadoop</groupId>
            <artifactId>hadoop-common</artifactId>
-            <version>2.9.2</version>
+            <version>2.10.1</version>
            <scope>compile</scope>
        </dependency>
    </dependencies>
@@ -36,7 +36,7 @@
        <dependency>
            <groupId>org.apache.hadoop</groupId>
            <artifactId>hadoop-common</artifactId>
-            <version>3.2.1</version>
+            <version>3.2.2</version>
        </dependency>
        <dependency>
            <groupId>org.apache.hadoop</groupId>
@@ -6,7 +6,7 @@

    <properties>
        <seaweedfs.client.version>2.85</seaweedfs.client.version>
-        <hadoop.version>2.9.2</hadoop.version>
+        <hadoop.version>2.10.1</hadoop.version>
    </properties>

    <groupId>com.github.chrislusf</groupId>
@@ -6,7 +6,7 @@

    <properties>
        <seaweedfs.client.version>2.85</seaweedfs.client.version>
-        <hadoop.version>3.1.1</hadoop.version>
+        <hadoop.version>3.1.4</hadoop.version>
    </properties>

    <groupId>com.github.chrislusf</groupId>
@@ -19,15 +19,19 @@ debug_shell:

 debug_mount:
 	go build -gcflags="all=-N -l"
-	dlv --listen=:2345 --headless=true --api-version=2 --accept-multiclient exec weed -- -v=4 mount -dir=~/tmp/mm -cacheCapacityMB=0 -filer.path=/
+	dlv --listen=:2345 --headless=true --api-version=2 --accept-multiclient exec weed -- -v=4 mount -dir=~/tmp/mm -cacheCapacityMB=0 -filer.path=/ -umask=000
+
+debug_mount2:
+	go build -gcflags="all=-N -l"
+	dlv --listen=:2345 --headless=true --api-version=2 --accept-multiclient exec weed -- -v=4 mount2 -dir=~/tmp/mm -cacheCapacityMB=0 -filer.path=/ -umask=000

 debug_server:
 	go build -gcflags="all=-N -l"
-	dlv --listen=:2345 --headless=true --api-version=2 --accept-multiclient exec weed -- server -dir=/Volumes/mobile_disk/99 -filer -volume.port=8343 -s3 -volume.max=0 -master.volumeSizeLimitMB=1024 -volume.preStopSeconds=1
+	dlv --listen=:2345 --headless=true --api-version=2 --accept-multiclient exec weed -- server -dir=~/tmp/99 -filer -volume.port=8343 -s3 -volume.max=0 -master.volumeSizeLimitMB=1024 -volume.preStopSeconds=1

 debug_volume:
 	go build -gcflags="all=-N -l"
-	dlv --listen=:2345 --headless=true --api-version=2 --accept-multiclient exec weed -- volume -dir=/Volumes/mobile_disk/100 -port 8564 -max=30 -preStopSeconds=2
+	dlv --listen=:2345 --headless=true --api-version=2 --accept-multiclient exec weed -- volume -dir=~/tmp/100 -port 8564 -max=30 -preStopSeconds=2

 debug_webdav:
 	go build -gcflags="all=-N -l"
@@ -52,4 +56,3 @@ debug_filer_remote_sync_buckets:
 debug_master_follower:
 	go build -gcflags="all=-N -l"
 	dlv --listen=:2345 --headless=true --api-version=2 --accept-multiclient exec weed -- -v=4 master.follower
-
@@ -30,6 +30,7 @@ var Commands = []*Command{
 	cmdMaster,
 	cmdMasterFollower,
 	cmdMount,
+	cmdMount2,
 	cmdS3,
 	cmdIam,
 	cmdMsgBroker,
@@ -8,7 +8,6 @@ import (
 	"github.com/chrislusf/seaweedfs/weed/pb/filer_pb"
 	"github.com/chrislusf/seaweedfs/weed/wdclient"
 	"google.golang.org/grpc"
-	"math"
 	"net/url"
 	"os"
 	"strings"
@@ -115,7 +114,7 @@ func runFilerCat(cmd *Command, args []string) bool {

 		filerCat.filerClient = client

-		return filer.StreamContent(&filerCat, writer, respLookupEntry.Entry.Chunks, 0, math.MaxInt64)
+		return filer.StreamContent(&filerCat, writer, respLookupEntry.Entry.Chunks, 0, int64(filer.FileSize(respLookupEntry.Entry)))

 	})

@@ -212,7 +212,7 @@ func genFileCopyTask(fileOrDir string, destPath string, fileCopyTaskChan chan Fi
 	if mode.IsDir() {
 		files, _ := os.ReadDir(fileOrDir)
 		for _, subFileOrDir := range files {
-			cleanedDestDirectory := filepath.Clean(destPath + fi.Name())
+			cleanedDestDirectory := destPath + fi.Name()
 			if err = genFileCopyTask(fileOrDir+"/"+subFileOrDir.Name(), cleanedDestDirectory+"/", fileCopyTaskChan); err != nil {
 				return err
 			}
@@ -482,7 +482,8 @@ func (worker *FileCopyWorker) uploadFileInChunks(task FileCopyTask, f *os.File,
 				})
 			})
 			if err != nil {
-				fmt.Printf("Failed to assign from %v: %v\n", worker.options.masters, err)
+				uploadError = fmt.Errorf("Failed to assign from %v: %v\n", worker.options.masters, err)
+				return
 			}

 			targetUrl := "http://" + assignResult.Location.Url + "/" + assignResult.FileId
@@ -264,7 +264,7 @@ func (option *RemoteGatewayOptions) makeBucketedEventProcessor(filerSource *sour
 						// update directory property
 						return nil
 					}
-					if filer.IsSameData(message.OldEntry, message.NewEntry) {
+					if message.OldEntry.RemoteEntry != nil && filer.IsSameData(message.OldEntry, message.NewEntry) {
 						glog.V(2).Infof("update meta: %+v", resp)
 						oldDest := toRemoteStorageLocation(oldBucket, util.NewFullPath(resp.Directory, message.OldEntry.Name), oldRemoteStorageMountLocation)
 						return client.UpdateFileMetadata(oldDest, message.OldEntry, message.NewEntry)
@@ -169,8 +169,11 @@ func doSubscribeFilerMetaChanges(clientId int32, grpcDialOption grpc.DialOption,
 		return persistEventFn(resp)
 	}

+	var lastLogTsNs = time.Now().Nanosecond()
 	processEventFnWithOffset := pb.AddOffsetFunc(processEventFn, 3*time.Second, func(counter int64, lastTsNs int64) error {
-		glog.V(0).Infof("sync %s to %s progressed to %v %0.2f/sec", sourceFiler, targetFiler, time.Unix(0, lastTsNs), float64(counter)/float64(3))
+		now := time.Now().Nanosecond()
+		glog.V(0).Infof("sync %s to %s progressed to %v %0.2f/sec", sourceFiler, targetFiler, time.Unix(0, lastTsNs), float64(counter)/(float64(now-lastLogTsNs)/1e9))
+		lastLogTsNs = now
 		return setOffset(grpcDialOption, targetFiler, SyncKeyPrefix, sourceFilerSignature, lastTsNs)
 	})

@@ -2,7 +2,7 @@ package command

 import (
 	"fmt"
-	"io/ioutil"
+	"io/fs"
 	"os"
 	"path"
 	"strconv"
@@ -22,17 +22,15 @@ func init() {
 }

 var cmdFix = &Command{
-	UsageLine: "fix -dir=/tmp -volumeId=234",
-	Short:     "run weed tool fix on index file if corrupted",
-	Long: `Fix runs the SeaweedFS fix command to re-create the index .idx file.
-
+	UsageLine: "fix [-volumeId=234] [-collection=bigData] /tmp",
+	Short:     "run weed tool fix on files or whole folders to recreate index file(s) if corrupted",
+	Long: `Fix runs the SeaweedFS fix command on dat files or whole folders to re-create the index .idx file.
  `,
 }

 var (
-	fixVolumePath       = cmdFix.Flag.String("dir", ".", "data directory to store files")
-	fixVolumeCollection = cmdFix.Flag.String("collection", "", "the volume collection name")
-	fixVolumeId         = cmdFix.Flag.Int("volumeId", 0, "an optional volume id.")
+	fixVolumeCollection = cmdFix.Flag.String("collection", "", "an optional volume collection name, if specified only it will be processed")
+	fixVolumeId         = cmdFix.Flag.Int64("volumeId", 0, "an optional volume id, if not 0 (default) only it will be processed")
 )

 type VolumeFileScanner4Fix struct {
@@ -62,63 +60,68 @@ func (scanner *VolumeFileScanner4Fix) VisitNeedle(n *needle.Needle, offset int64
 }

 func runFix(cmd *Command, args []string) bool {
+	for _, arg := range args {
+		basePath, f := path.Split(util.ResolvePath(arg))

-	dir := util.ResolvePath(*fixVolumePath)
-	if *fixVolumeId != 0 {
-		doFixOneVolume(dir, *fixVolumeCollection, needle.VolumeId(*fixVolumeId))
-		return true
-	}
-
-	files, err := ioutil.ReadDir(dir)
-	if err != nil {
-		fmt.Println(err)
-		return false
-	}
-
-	for _, file := range files {
-		if !strings.HasSuffix(file.Name(), ".dat") {
-			continue
+		files := []fs.DirEntry{}
+		if f == "" {
+			fileInfo, err := os.ReadDir(basePath)
+			if err != nil {
+				fmt.Println(err)
+				return false
+			}
+			files = fileInfo
+		} else {
+			fileInfo, err := os.Stat(basePath + f)
+			if err != nil {
+				fmt.Println(err)
+				return false
+			}
+			files = []fs.DirEntry{fs.FileInfoToDirEntry(fileInfo)}
 		}
-		if *fixVolumeCollection != "" {
-			if !strings.HasPrefix(file.Name(), *fixVolumeCollection+"_") {
+
+		for _, file := range files {
+			if !strings.HasSuffix(file.Name(), ".dat") {
 				continue
 			}
+			if *fixVolumeCollection != "" {
+				if !strings.HasPrefix(file.Name(), *fixVolumeCollection+"_") {
+					continue
+				}
+			}
+			baseFileName := file.Name()[:len(file.Name())-4]
+			collection, volumeIdStr := "", baseFileName
+			if sepIndex := strings.LastIndex(baseFileName, "_"); sepIndex > 0 {
+				collection = baseFileName[:sepIndex]
+				volumeIdStr = baseFileName[sepIndex+1:]
+			}
+			volumeId, parseErr := strconv.ParseInt(volumeIdStr, 10, 64)
+			if parseErr != nil {
+				fmt.Printf("Failed to parse volume id from %s: %v\n", baseFileName, parseErr)
+				return false
+			}
+			if *fixVolumeId != 0 && *fixVolumeId != volumeId {
+				continue
+			}
+			doFixOneVolume(basePath, baseFileName, collection, volumeId)
 		}
-		baseFileName := file.Name()[:len(file.Name())-4]
-		collection, volumeIdStr := "", baseFileName
-		if sepIndex := strings.LastIndex(baseFileName, "_"); sepIndex > 0 {
-			collection = baseFileName[:sepIndex]
-			volumeIdStr = baseFileName[sepIndex+1:]
-		}
-		volumeId, parseErr := strconv.ParseInt(volumeIdStr, 10, 64)
-		if parseErr != nil {
-			fmt.Printf("Failed to parse volume id from %s: %v\n", baseFileName, parseErr)
-			return false
-		}
-		doFixOneVolume(dir, collection, needle.VolumeId(volumeId))
 	}
-
 	return true
 }

-func doFixOneVolume(dir, collection string, volumeId needle.VolumeId) {
+func doFixOneVolume(basepath string, baseFileName string, collection string, volumeId int64) {

-	baseFileName := strconv.Itoa(int(volumeId))
-	if collection != "" {
-		baseFileName = collection + "_" + baseFileName
-	}
-
-	indexFileName := path.Join(dir, baseFileName+".idx")
+	indexFileName := path.Join(basepath, baseFileName+".idx")

 	nm := needle_map.NewMemDb()
 	defer nm.Close()

-	vid := needle.VolumeId(*fixVolumeId)
+	vid := needle.VolumeId(volumeId)
 	scanner := &VolumeFileScanner4Fix{
 		nm: nm,
 	}

-	if err := storage.ScanVolumeFile(dir, collection, vid, storage.NeedleMapInMemory, scanner); err != nil {
+	if err := storage.ScanVolumeFile(basepath, collection, vid, storage.NeedleMapInMemory, scanner); err != nil {
 		glog.Fatalf("scan .dat File: %v", err)
 		os.Remove(indexFileName)
 	}
@@ -2,6 +2,7 @@ package command

 import (
 	"github.com/chrislusf/raft/protobuf"
+	stats_collect "github.com/chrislusf/seaweedfs/weed/stats"
 	"github.com/gorilla/mux"
 	"google.golang.org/grpc/reflection"
 	"net/http"
@@ -42,6 +43,7 @@ type MasterOptions struct {
 	metricsAddress     *string
 	metricsIntervalSec *int
 	raftResumeState    *bool
+	metricsHttpPort    *int
 }

 func init() {
@@ -61,6 +63,7 @@ func init() {
 	m.disableHttp = cmdMaster.Flag.Bool("disableHttp", false, "disable http requests, only gRPC operations are allowed.")
 	m.metricsAddress = cmdMaster.Flag.String("metrics.address", "", "Prometheus gateway address <host>:<port>")
 	m.metricsIntervalSec = cmdMaster.Flag.Int("metrics.intervalSeconds", 15, "Prometheus push interval in seconds")
+	m.metricsHttpPort = cmdMaster.Flag.Int("metricsPort", 0, "Prometheus metrics listen port")
 	m.raftResumeState = cmdMaster.Flag.Bool("resumeState", false, "resume previous state on start master server")
 }

@@ -104,6 +107,7 @@ func runMaster(cmd *Command, args []string) bool {
 		glog.Fatalf("volumeSizeLimitMB should be smaller than 30000")
 	}

+	go stats_collect.StartMetricsServer(*m.metricsHttpPort)
 	startMaster(m, masterWhiteList)

 	return true
@@ -26,6 +26,8 @@ type MountOptions struct {
 	uidMap             *string
 	gidMap             *string
 	readOnly           *bool
+	debug              *bool
+	debugPort          *int
 }

 var (
@@ -57,6 +59,8 @@ func init() {
 	mountOptions.uidMap = cmdMount.Flag.String("map.uid", "", "map local uid to uid on filer, comma-separated <local_uid>:<filer_uid>")
 	mountOptions.gidMap = cmdMount.Flag.String("map.gid", "", "map local gid to gid on filer, comma-separated <local_gid>:<filer_gid>")
 	mountOptions.readOnly = cmdMount.Flag.Bool("readOnly", false, "read only")
+	mountOptions.debug = cmdMount.Flag.Bool("debug", false, "serves runtime profiling data, e.g., http://localhost:<debug.port>/debug/pprof/goroutine?debug=2")
+	mountOptions.debugPort = cmdMount.Flag.Int("debug.port", 6061, "http port for debugging")

 	mountCpuProfile = cmdMount.Flag.String("cpuprofile", "", "cpu profile output file")
 	mountMemProfile = cmdMount.Flag.String("memprofile", "", "memory profile output file")
@@ -0,0 +1,83 @@
+package command
+
+import (
+	"os"
+	"time"
+)
+
+type Mount2Options struct {
+	filer              *string
+	filerMountRootPath *string
+	dir                *string
+	dirAutoCreate      *bool
+	collection         *string
+	replication        *string
+	diskType           *string
+	ttlSec             *int
+	chunkSizeLimitMB   *int
+	concurrentWriters  *int
+	cacheDir           *string
+	cacheSizeMB        *int64
+	dataCenter         *string
+	allowOthers        *bool
+	umaskString        *string
+	nonempty           *bool
+	volumeServerAccess *string
+	uidMap             *string
+	gidMap             *string
+	readOnly           *bool
+	debug              *bool
+	debugPort          *int
+}
+
+var (
+	mount2Options Mount2Options
+)
+
+func init() {
+	cmdMount2.Run = runMount2 // break init cycle
+	mount2Options.filer = cmdMount2.Flag.String("filer", "localhost:8888", "comma-separated weed filer location")
+	mount2Options.filerMountRootPath = cmdMount2.Flag.String("filer.path", "/", "mount this remote path from filer server")
+	mount2Options.dir = cmdMount2.Flag.String("dir", ".", "mount weed filer to this directory")
+	mount2Options.dirAutoCreate = cmdMount2.Flag.Bool("dirAutoCreate", false, "auto create the directory to mount to")
+	mount2Options.collection = cmdMount2.Flag.String("collection", "", "collection to create the files")
+	mount2Options.replication = cmdMount2.Flag.String("replication", "", "replication(e.g. 000, 001) to create to files. If empty, let filer decide.")
+	mount2Options.diskType = cmdMount2.Flag.String("disk", "", "[hdd|ssd|<tag>] hard drive or solid state drive or any tag")
+	mount2Options.ttlSec = cmdMount2.Flag.Int("ttl", 0, "file ttl in seconds")
+	mount2Options.chunkSizeLimitMB = cmdMount2.Flag.Int("chunkSizeLimitMB", 2, "local write buffer size, also chunk large files")
+	mount2Options.concurrentWriters = cmdMount2.Flag.Int("concurrentWriters", 32, "limit concurrent goroutine writers if not 0")
+	mount2Options.cacheDir = cmdMount2.Flag.String("cacheDir", os.TempDir(), "local cache directory for file chunks and meta data")
+	mount2Options.cacheSizeMB = cmdMount2.Flag.Int64("cacheCapacityMB", 1000, "local file chunk cache capacity in MB (0 will disable cache)")
+	mount2Options.dataCenter = cmdMount2.Flag.String("dataCenter", "", "prefer to write to the data center")
+	mount2Options.allowOthers = cmdMount2.Flag.Bool("allowOthers", true, "allows other users to access the file system")
+	mount2Options.umaskString = cmdMount2.Flag.String("umask", "022", "octal umask, e.g., 022, 0111")
+	mount2Options.nonempty = cmdMount2.Flag.Bool("nonempty", false, "allows the mounting over a non-empty directory")
+	mount2Options.volumeServerAccess = cmdMount2.Flag.String("volumeServerAccess", "direct", "access volume servers by [direct|publicUrl|filerProxy]")
+	mount2Options.uidMap = cmdMount2.Flag.String("map.uid", "", "map local uid to uid on filer, comma-separated <local_uid>:<filer_uid>")
+	mount2Options.gidMap = cmdMount2.Flag.String("map.gid", "", "map local gid to gid on filer, comma-separated <local_gid>:<filer_gid>")
+	mount2Options.readOnly = cmdMount2.Flag.Bool("readOnly", false, "read only")
+	mount2Options.debug = cmdMount2.Flag.Bool("debug", false, "serves runtime profiling data, e.g., http://localhost:<debug.port>/debug/pprof/goroutine?debug=2")
+	mount2Options.debugPort = cmdMount2.Flag.Int("debug.port", 6061, "http port for debugging")
+
+	mountCpuProfile = cmdMount2.Flag.String("cpuprofile", "", "cpu profile output file")
+	mountMemProfile = cmdMount2.Flag.String("memprofile", "", "memory profile output file")
+	mountReadRetryTime = cmdMount2.Flag.Duration("readRetryTime", 6*time.Second, "maximum read retry wait time")
+}
+
+var cmdMount2 = &Command{
+	UsageLine: "mount2 -filer=localhost:8888 -dir=/some/dir",
+	Short:     "<WIP> mount weed filer to a directory as file system in userspace(FUSE)",
+	Long: `mount weed filer to userspace.
+
+  Pre-requisites:
+  1) have SeaweedFS master and volume servers running
+  2) have a "weed filer" running
+  These 2 requirements can be achieved with one command "weed server -filer=true"
+
+  This uses github.com/seaweedfs/fuse, which enables writing FUSE file systems on
+  Linux, and OS X.
+
+  On OS X, it requires OSXFUSE (http://osxfuse.github.com/).
+
+  `,
+}
@@ -0,0 +1,210 @@
+package command
+
+import (
+	"context"
+	"fmt"
+	"github.com/chrislusf/seaweedfs/weed/glog"
+	"github.com/chrislusf/seaweedfs/weed/mount"
+	"github.com/chrislusf/seaweedfs/weed/mount/meta_cache"
+	"github.com/chrislusf/seaweedfs/weed/mount/unmount"
+	"github.com/chrislusf/seaweedfs/weed/pb"
+	"github.com/chrislusf/seaweedfs/weed/pb/filer_pb"
+	"github.com/chrislusf/seaweedfs/weed/security"
+	"github.com/chrislusf/seaweedfs/weed/storage/types"
+	"github.com/hanwen/go-fuse/v2/fuse"
+	"net/http"
+	"os"
+	"os/user"
+	"runtime"
+	"strconv"
+	"strings"
+	"time"
+
+	"github.com/chrislusf/seaweedfs/weed/util"
+	"github.com/chrislusf/seaweedfs/weed/util/grace"
+)
+
+func runMount2(cmd *Command, args []string) bool {
+
+	if *mountOptions.debug {
+		go http.ListenAndServe(fmt.Sprintf(":%d", *mountOptions.debugPort), nil)
+	}
+
+	grace.SetupProfiling(*mountCpuProfile, *mountMemProfile)
+	if *mountReadRetryTime < time.Second {
+		*mountReadRetryTime = time.Second
+	}
+	util.RetryWaitTime = *mountReadRetryTime
+
+	umask, umaskErr := strconv.ParseUint(*mountOptions.umaskString, 8, 64)
+	if umaskErr != nil {
+		fmt.Printf("can not parse umask %s", *mountOptions.umaskString)
+		return false
+	}
+
+	if len(args) > 0 {
+		return false
+	}
+
+	return RunMount2(&mount2Options, os.FileMode(umask))
+}
+
+func RunMount2(option *Mount2Options, umask os.FileMode) bool {
+
+	// basic checks
+	chunkSizeLimitMB := *mountOptions.chunkSizeLimitMB
+	if chunkSizeLimitMB <= 0 {
+		fmt.Printf("Please specify a reasonable buffer size.")
+		return false
+	}
+
+	// try to connect to filer
+	filerAddresses := pb.ServerAddresses(*option.filer).ToAddresses()
+	util.LoadConfiguration("security", false)
+	grpcDialOption := security.LoadClientTLS(util.GetViper(), "grpc.client")
+	var cipher bool
+	var err error
+	for i := 0; i < 10; i++ {
+		err = pb.WithOneOfGrpcFilerClients(false, filerAddresses, grpcDialOption, func(client filer_pb.SeaweedFilerClient) error {
+			resp, err := client.GetFilerConfiguration(context.Background(), &filer_pb.GetFilerConfigurationRequest{})
+			if err != nil {
+				return fmt.Errorf("get filer grpc address %v configuration: %v", filerAddresses, err)
+			}
+			cipher = resp.Cipher
+			return nil
+		})
+		if err != nil {
+			glog.V(0).Infof("failed to talk to filer %v: %v", filerAddresses, err)
+			glog.V(0).Infof("wait for %d seconds ...", i+1)
+			time.Sleep(time.Duration(i+1) * time.Second)
+		}
+	}
+	if err != nil {
+		glog.Errorf("failed to talk to filer %v: %v", filerAddresses, err)
+		return true
+	}
+
+	filerMountRootPath := *option.filerMountRootPath
+
+	// clean up mount point
+	dir := util.ResolvePath(*option.dir)
+	if dir == "" {
+		fmt.Printf("Please specify the mount directory via \"-dir\"")
+		return false
+	}
+
+	unmount.Unmount(dir)
+
+	// detect mount folder mode
+	if *option.dirAutoCreate {
+		os.MkdirAll(dir, os.FileMode(0777)&^umask)
+	}
+	fileInfo, err := os.Stat(dir)
+
+	// collect uid, gid
+	uid, gid := uint32(0), uint32(0)
+	mountMode := os.ModeDir | 0777
+	if err == nil {
+		mountMode = os.ModeDir | os.FileMode(0777)&^umask
+		uid, gid = util.GetFileUidGid(fileInfo)
+		fmt.Printf("mount point owner uid=%d gid=%d mode=%s\n", uid, gid, mountMode)
+	} else {
+		fmt.Printf("can not stat %s\n", dir)
+		return false
+	}
+
+	// detect uid, gid
+	if uid == 0 {
+		if u, err := user.Current(); err == nil {
+			if parsedId, pe := strconv.ParseUint(u.Uid, 10, 32); pe == nil {
+				uid = uint32(parsedId)
+			}
+			if parsedId, pe := strconv.ParseUint(u.Gid, 10, 32); pe == nil {
+				gid = uint32(parsedId)
+			}
+			fmt.Printf("current uid=%d gid=%d\n", uid, gid)
+		}
+	}
+
+	// mapping uid, gid
+	uidGidMapper, err := meta_cache.NewUidGidMapper(*option.uidMap, *option.gidMap)
+	if err != nil {
+		fmt.Printf("failed to parse %s %s: %v\n", *option.uidMap, *option.gidMap, err)
+		return false
+	}
+
+	// Ensure target mount point availability
+	if isValid := checkMountPointAvailable(dir); !isValid {
+		glog.Fatalf("Expected mount to still be active, target mount point: %s, please check!", dir)
+		return true
+	}
+
+	// mount fuse
+	fuseMountOptions := &fuse.MountOptions{
+		AllowOther:               *option.allowOthers,
+		Options:                  nil,
+		MaxBackground:            128,
+		MaxWrite:                 1024 * 1024 * 2,
+		MaxReadAhead:             1024 * 1024 * 2,
+		IgnoreSecurityLabels:     false,
+		RememberInodes:           false,
+		FsName:                   *option.filer + ":" + filerMountRootPath,
+		Name:                     "seaweedfs",
+		SingleThreaded:           false,
+		DisableXAttrs:            false,
+		Debug:                    true,
+		EnableLocks:              false,
+		ExplicitDataCacheControl: false,
+		// SyncRead:                 false, // set to false to enable the FUSE_CAP_ASYNC_READ capability
+		DirectMount:      true,
+		DirectMountFlags: 0,
+		// EnableAcl:                false,
+	}
+
+	// find mount point
+	mountRoot := filerMountRootPath
+	if mountRoot != "/" && strings.HasSuffix(mountRoot, "/") {
+		mountRoot = mountRoot[0 : len(mountRoot)-1]
+	}
+
+	seaweedFileSystem := mount.NewSeaweedFileSystem(&mount.Option{
+		MountDirectory:     dir,
+		FilerAddresses:     filerAddresses,
+		GrpcDialOption:     grpcDialOption,
+		FilerMountRootPath: mountRoot,
+		Collection:         *option.collection,
+		Replication:        *option.replication,
+		TtlSec:             int32(*option.ttlSec),
+		DiskType:           types.ToDiskType(*option.diskType),
+		ChunkSizeLimit:     int64(chunkSizeLimitMB) * 1024 * 1024,
+		ConcurrentWriters:  *option.concurrentWriters,
+		CacheDir:           *option.cacheDir,
+		CacheSizeMB:        *option.cacheSizeMB,
+		DataCenter:         *option.dataCenter,
+		MountUid:           uid,
+		MountGid:           gid,
+		MountMode:          mountMode,
+		MountCtime:         fileInfo.ModTime(),
+		MountMtime:         time.Now(),
+		Umask:              umask,
+		VolumeServerAccess: *mountOptions.volumeServerAccess,
+		Cipher:             cipher,
+		UidGidMapper:       uidGidMapper,
+	})
+
+	server, err := fuse.NewServer(seaweedFileSystem, dir, fuseMountOptions)
+	if err != nil {
+		glog.Fatalf("Mount fail: %v", err)
+	}
+	grace.OnInterrupt(func() {
+		unmount.Unmount(dir)
+	})
+
+	seaweedFileSystem.StartBackgroundTasks()
+
+	fmt.Printf("This is SeaweedFS version %s %s %s\n", util.Version(), runtime.GOOS, runtime.GOARCH)
+
+	server.Serve()
+
+	return true
+}
@@ -6,6 +6,7 @@ package command
 import (
 	"context"
 	"fmt"
+	"net/http"
 	"os"
 	"os/user"
 	"path"
@@ -34,6 +35,10 @@ import (

 func runMount(cmd *Command, args []string) bool {

+	if *mountOptions.debug {
+		go http.ListenAndServe(fmt.Sprintf(":%d", *mountOptions.debugPort), nil)
+	}
+
 	grace.SetupProfiling(*mountCpuProfile, *mountMemProfile)
 	if *mountReadRetryTime < time.Second {
 		*mountReadRetryTime = time.Second
@@ -171,6 +176,7 @@ func RunMount(option *MountOptions, umask os.FileMode) bool {
 		fuse.NoAppleXattr(),
 		fuse.ExclCreate(),
 		fuse.DaemonTimeout("3600"),
+		fuse.AllowDev(),
 		fuse.AllowSUID(),
 		fuse.DefaultPermissions(),
 		fuse.MaxReadahead(1024 * 512),
@@ -2,7 +2,7 @@ package command

 import (
 	"fmt"
-	"os"
+	"github.com/chrislusf/seaweedfs/weed/util"
 	"path/filepath"

 	"github.com/chrislusf/seaweedfs/weed/command/scaffold"
@@ -56,7 +56,7 @@ func runScaffold(cmd *Command, args []string) bool {
 	}

 	if *outputPath != "" {
-		os.WriteFile(filepath.Join(*outputPath, *config+".toml"), []byte(content), 0644)
+		util.WriteFile(filepath.Join(*outputPath, *config+".toml"), []byte(content), 0644)
 	} else {
 		fmt.Println(content)
 	}
@@ -153,6 +153,8 @@ password = ""
 superLargeDirectories = []
 # Name of the datacenter local to this filer, used as host selection fallback.
 localDC = ""
+# Gocql connection timeout, default: 600ms
+connection_timeout_millisecond = 600

 [hbase]
 enabled = false
@@ -19,9 +19,6 @@ scripts = """
 """
 sleep_minutes = 17          # sleep minutes between each script execution

-[master.filer]
-default = "localhost:8888"    # used by maintenance scripts if the scripts needs to use fs related commands
-

 [master.sequencer]
 type = "raft"     # Choose [raft|snowflake] type for storing the file id sequence
@@ -28,6 +28,7 @@ var (
 	masterOptions    MasterOptions
 	filerOptions     FilerOptions
 	s3Options        S3Options
+	iamOptions       IamOptions
 	webdavOptions    WebDavOption
 	msgBrokerOptions MessageBrokerOptions
 )
@@ -61,7 +62,7 @@ var (
 	serverWhiteListOption     = cmdServer.Flag.String("whiteList", "", "comma separated Ip addresses having write permission. No limit if empty.")
 	serverDisableHttp         = cmdServer.Flag.Bool("disableHttp", false, "disable http requests, only gRPC operations are allowed.")
 	volumeDataFolders         = cmdServer.Flag.String("dir", os.TempDir(), "directories to store data files. dir[,dir]...")
-	volumeMaxDataVolumeCounts = cmdServer.Flag.String("volume.max", "8", "maximum numbers of volumes, count[,count]... If set to zero, the limit will be auto configured.")
+	volumeMaxDataVolumeCounts = cmdServer.Flag.String("volume.max", "8", "maximum numbers of volumes, count[,count]... If set to zero, the limit will be auto configured as free disk space divided by volume size.")
 	volumeMinFreeSpacePercent = cmdServer.Flag.String("volume.minFreeSpacePercent", "1", "minimum free disk space (default to 1%). Low disk space will mark all volumes as ReadOnly (deprecated, use minFreeSpace instead).")
 	volumeMinFreeSpace        = cmdServer.Flag.String("volume.minFreeSpace", "", "min free disk space (value<=100 as percentage like 1, other as human readable bytes, like 10GiB). Low disk space will mark all volumes as ReadOnly.")
 	serverMetricsHttpPort     = cmdServer.Flag.Int("metricsPort", 0, "Prometheus metrics listen port")
@@ -71,6 +72,7 @@ var (
 	isStartingVolumeServer = cmdServer.Flag.Bool("volume", true, "whether to start volume server")
 	isStartingFiler        = cmdServer.Flag.Bool("filer", false, "whether to start filer")
 	isStartingS3           = cmdServer.Flag.Bool("s3", false, "whether to start S3 gateway")
+	isStartingIam          = cmdServer.Flag.Bool("iam", false, "whether to start IAM service")
 	isStartingWebDav       = cmdServer.Flag.Bool("webdav", false, "whether to start WebDAV gateway")
 	isStartingMsgBroker    = cmdServer.Flag.Bool("msgBroker", false, "whether to start message broker")

@@ -134,6 +136,8 @@ func init() {
 	s3Options.auditLogConfig = cmdServer.Flag.String("s3.auditLogConfig", "", "path to the audit log config file")
 	s3Options.allowEmptyFolder = cmdServer.Flag.Bool("s3.allowEmptyFolder", true, "allow empty folders")

+	iamOptions.port = cmdServer.Flag.Int("iam.port", 8111, "iam server http listen port")
+
 	webdavOptions.port = cmdServer.Flag.Int("webdav.port", 7333, "webdav server http listen port")
 	webdavOptions.collection = cmdServer.Flag.String("webdav.collection", "", "collection to create the files")
 	webdavOptions.replication = cmdServer.Flag.String("webdav.replication", "", "replication to create the files")
@@ -161,6 +165,9 @@ func runServer(cmd *Command, args []string) bool {
 	if *isStartingS3 {
 		*isStartingFiler = true
 	}
+	if *isStartingIam {
+		*isStartingFiler = true
+	}
 	if *isStartingWebDav {
 		*isStartingFiler = true
 	}
@@ -201,6 +208,7 @@ func runServer(cmd *Command, args []string) bool {

 	filerAddress := string(pb.NewServerAddress(*serverIp, *filerOptions.port, *filerOptions.portGrpc))
 	s3Options.filer = &filerAddress
+	iamOptions.filer = &filerAddress
 	webdavOptions.filer = &filerAddress
 	msgBrokerOptions.filer = &filerAddress

@@ -227,25 +235,27 @@ func runServer(cmd *Command, args []string) bool {
 	if *isStartingFiler {
 		go func() {
 			time.Sleep(1 * time.Second)
-
 			filerOptions.startFiler()
-
 		}()
 	}

 	if *isStartingS3 {
 		go func() {
 			time.Sleep(2 * time.Second)
-
 			s3Options.startS3Server()
+		}()
+	}

+	if *isStartingIam {
+		go func() {
+			time.Sleep(2 * time.Second)
+			iamOptions.startIamServer()
 		}()
 	}

 	if *isStartingWebDav {
 		go func() {
 			time.Sleep(2 * time.Second)
-
 			webdavOptions.startWebDav()

 		}()
@@ -107,7 +107,7 @@ var cmdVolume = &Command{

 var (
 	volumeFolders         = cmdVolume.Flag.String("dir", os.TempDir(), "directories to store data files. dir[,dir]...")
-	maxVolumeCounts       = cmdVolume.Flag.String("max", "8", "maximum numbers of volumes, count[,count]... If set to zero, the limit will be auto configured.")
+	maxVolumeCounts       = cmdVolume.Flag.String("max", "8", "maximum numbers of volumes, count[,count]... If set to zero, the limit will be auto configured as free disk space divided by volume size.")
 	volumeWhiteListOption = cmdVolume.Flag.String("whiteList", "", "comma separated Ip addresses having write permission. No limit if empty.")
 	minFreeSpacePercent   = cmdVolume.Flag.String("minFreeSpacePercent", "1", "minimum free disk space (default to 1%). Low disk space will mark all volumes as ReadOnly (deprecated, use minFreeSpace instead).")
 	minFreeSpace          = cmdVolume.Flag.String("minFreeSpace", "", "min free disk space (value<=100 as percentage like 1, other as human readable bytes, like 10GiB). Low disk space will mark all volumes as ReadOnly.")
@@ -4,6 +4,7 @@ import (
 	"context"
 	"fmt"
 	"github.com/gocql/gocql"
+	"time"

 	"github.com/chrislusf/seaweedfs/weed/filer"
 	"github.com/chrislusf/seaweedfs/weed/glog"
@@ -33,6 +34,7 @@ func (store *CassandraStore) Initialize(configuration util.Configuration, prefix
 		configuration.GetString(prefix+"password"),
 		configuration.GetStringSlice(prefix+"superLargeDirectories"),
 		configuration.GetString(prefix+"localDC"),
+		configuration.GetInt(prefix+"connection_timeout_millisecond"),
 	)
 }

@@ -41,12 +43,14 @@ func (store *CassandraStore) isSuperLargeDirectory(dir string) (dirHash string,
 	return
 }

-func (store *CassandraStore) initialize(keyspace string, hosts []string, username string, password string, superLargeDirectories []string, localDC string) (err error) {
+func (store *CassandraStore) initialize(keyspace string, hosts []string, username string, password string, superLargeDirectories []string, localDC string, timeout int) (err error) {
 	store.cluster = gocql.NewCluster(hosts...)
 	if username != "" && password != "" {
 		store.cluster.Authenticator = gocql.PasswordAuthenticator{Username: username, Password: password}
 	}
 	store.cluster.Keyspace = keyspace
+	store.cluster.Timeout = time.Duration(timeout) * time.Millisecond
+	glog.V(0).Infof("timeout = %d", timeout)
 	fallback := gocql.RoundRobinHostPolicy()
 	if localDC != "" {
 		fallback = gocql.DCAwareRoundRobinPolicy(localDC)
@@ -43,6 +43,7 @@ type Entry struct {
 	HardLinkCounter int32
 	Content         []byte
 	Remote          *filer_pb.RemoteEntry
+	Quota           int64
 }

 func (entry *Entry) Size() uint64 {
@@ -70,6 +71,7 @@ func (entry *Entry) ShallowClone() *Entry {
 	newEntry.HardLinkCounter = entry.HardLinkCounter
 	newEntry.Content = entry.Content
 	newEntry.Remote = entry.Remote
+	newEntry.Quota = entry.Quota

 	return newEntry
 }
@@ -96,6 +98,7 @@ func (entry *Entry) ToExistingProtoEntry(message *filer_pb.Entry) {
 	message.HardLinkCounter = entry.HardLinkCounter
 	message.Content = entry.Content
 	message.RemoteEntry = entry.Remote
+	message.Quota = entry.Quota
 }

 func FromPbEntryToExistingEntry(message *filer_pb.Entry, fsEntry *Entry) {
@@ -106,6 +109,7 @@ func FromPbEntryToExistingEntry(message *filer_pb.Entry, fsEntry *Entry) {
 	fsEntry.HardLinkCounter = message.HardLinkCounter
 	fsEntry.Content = message.Content
 	fsEntry.Remote = message.RemoteEntry
+	fsEntry.Quota = message.Quota
 }

 func (entry *Entry) ToProtoFullEntry() *filer_pb.FullEntry {
@@ -118,6 +118,9 @@ func EqualEntry(a, b *Entry) bool {
 	if !proto.Equal(a.Remote, b.Remote) {
 		return false
 	}
+	if a.Quota != b.Quota {
+		return false
+	}
 	return true
 }

@@ -101,6 +101,21 @@ func DoMinusChunks(as, bs []*filer_pb.FileChunk) (delta []*filer_pb.FileChunk) {
 	return
 }

+func DoMinusChunksBySourceFileId(as, bs []*filer_pb.FileChunk) (delta []*filer_pb.FileChunk) {
+
+	fileIds := make(map[string]bool)
+	for _, interval := range bs {
+		fileIds[interval.GetFileIdString()] = true
+	}
+	for _, chunk := range as {
+		if _, found := fileIds[chunk.GetSourceFileId()]; !found {
+			delta = append(delta, chunk)
+		}
+	}
+
+	return
+}
+
 type ChunkView struct {
 	FileId      string
 	Offset      int64
@@ -75,12 +75,12 @@ func (fc *FilerConf) loadFromFiler(filer *Filer) (err error) {
 		return fc.LoadFromBytes(entry.Content)
 	}

-	return fc.loadFromChunks(filer, entry.Content, entry.Chunks)
+	return fc.loadFromChunks(filer, entry.Content, entry.Chunks, entry.Size())
 }

-func (fc *FilerConf) loadFromChunks(filer *Filer, content []byte, chunks []*filer_pb.FileChunk) (err error) {
+func (fc *FilerConf) loadFromChunks(filer *Filer, content []byte, chunks []*filer_pb.FileChunk, size uint64) (err error) {
 	if len(content) == 0 {
-		content, err = filer.readEntry(chunks)
+		content, err = filer.readEntry(chunks, size)
 		if err != nil {
 			glog.Errorf("read filer conf content: %v", err)
 			return
@@ -2,8 +2,6 @@ package filer

 import (
 	"bytes"
-	"math"
-
 	"github.com/chrislusf/seaweedfs/weed/glog"
 	"github.com/chrislusf/seaweedfs/weed/pb/filer_pb"
 	"github.com/chrislusf/seaweedfs/weed/util"
@@ -55,9 +53,9 @@ func (f *Filer) maybeReloadFilerConfiguration(event *filer_pb.SubscribeMetadataR
 	}
 }

-func (f *Filer) readEntry(chunks []*filer_pb.FileChunk) ([]byte, error) {
+func (f *Filer) readEntry(chunks []*filer_pb.FileChunk, size uint64) ([]byte, error) {
 	var buf bytes.Buffer
-	err := StreamContent(f.MasterClient, &buf, chunks, 0, math.MaxInt64)
+	err := StreamContent(f.MasterClient, &buf, chunks, 0, int64(size))
 	if err != nil {
 		return nil, err
 	}
@@ -66,7 +64,7 @@ func (f *Filer) readEntry(chunks []*filer_pb.FileChunk) ([]byte, error) {

 func (f *Filer) reloadFilerConfiguration(entry *filer_pb.Entry) {
 	fc := NewFilerConf()
-	err := fc.loadFromChunks(f, entry.Content, entry.Chunks)
+	err := fc.loadFromChunks(f, entry.Content, entry.Chunks, FileSize(entry))
 	if err != nil {
 		glog.Errorf("read filer conf chunks: %v", err)
 		return
@@ -4,6 +4,7 @@ import (
 	"context"
 	"errors"
 	"github.com/chrislusf/seaweedfs/weed/util"
+	"io"
 )

 var (
@@ -45,3 +46,7 @@ type BucketAware interface {
 	OnBucketDeletion(bucket string)
 	CanDropWholeBucket() bool
 }
+
+type Debuggable interface {
+	Debug(writer io.Writer)
+}
@@ -4,6 +4,7 @@ import (
 	"context"
 	"github.com/chrislusf/seaweedfs/weed/glog"
 	"github.com/viant/ptrie"
+	"io"
 	"math"
 	"strings"
 	"time"
@@ -15,6 +16,7 @@ import (

 var (
 	_ = VirtualFilerStore(&FilerStoreWrapper{})
+	_ = Debuggable(&FilerStoreWrapper{})
 )

 type VirtualFilerStore interface {
@@ -333,3 +335,9 @@ func (fsw *FilerStoreWrapper) KvGet(ctx context.Context, key []byte) (value []by
 func (fsw *FilerStoreWrapper) KvDelete(ctx context.Context, key []byte) (err error) {
 	return fsw.getDefaultStore().KvDelete(ctx, key)
 }
+
+func (fsw *FilerStoreWrapper) Debug(writer io.Writer) {
+	if debuggable, ok := fsw.getDefaultStore().(Debuggable); ok {
+		debuggable.Debug(writer)
+	}
+}
@@ -9,6 +9,7 @@ import (
 	"github.com/syndtr/goleveldb/leveldb/filter"
 	"github.com/syndtr/goleveldb/leveldb/opt"
 	leveldb_util "github.com/syndtr/goleveldb/leveldb/util"
+	"io"
 	"os"

 	"github.com/chrislusf/seaweedfs/weed/filer"
@@ -21,6 +22,10 @@ const (
 	DIR_FILE_SEPARATOR = byte(0x00)
 )

+var (
+	_ = filer.Debuggable(&LevelDBStore{})
+)
+
 func init() {
 	filer.Stores = append(filer.Stores, &LevelDBStore{})
 }
@@ -242,3 +247,13 @@ func getNameFromKey(key []byte) string {
 func (store *LevelDBStore) Shutdown() {
 	store.db.Close()
 }
+
+func (store *LevelDBStore) Debug(writer io.Writer) {
+	iter := store.db.NewIterator(&leveldb_util.Range{}, nil)
+	for iter.Next() {
+		key := iter.Key()
+		fullName := bytes.Replace(key, []byte{DIR_FILE_SEPARATOR}, []byte{' '}, 1)
+		fmt.Fprintf(writer, "%v\n", string(fullName))
+	}
+	iter.Release()
+}
@@ -165,13 +165,14 @@ func (ma *MetaAggregator) subscribeToOneFiler(f *Filer, self pb.ServerAddress, p

 	for {
 		glog.V(4).Infof("subscribing remote %s meta change: %v", peer, time.Unix(0, lastTsNs))
-		err := pb.WithFilerClient(false, peer, ma.grpcDialOption, func(client filer_pb.SeaweedFilerClient) error {
+		err := pb.WithFilerClient(true, peer, ma.grpcDialOption, func(client filer_pb.SeaweedFilerClient) error {
 			ctx, cancel := context.WithCancel(context.Background())
 			defer cancel()
 			stream, err := client.SubscribeLocalMetadata(ctx, &filer_pb.SubscribeMetadataRequest{
 				ClientName: "filer:" + string(self),
 				PathPrefix: "/",
 				SinceNs:    lastTsNs,
+				ClientId:   int32(ma.filer.UniqueFileId),
 			})
 			if err != nil {
 				return fmt.Errorf("subscribe: %v", err)
@@ -1,22 +0,0 @@
-package filer
-
-func hasWritePermission(dir *Entry, entry *Entry) bool {
-
-	if dir == nil {
-		return false
-	}
-
-	if dir.Uid == entry.Uid && dir.Mode&0200 > 0 {
-		return true
-	}
-
-	if dir.Gid == entry.Gid && dir.Mode&0020 > 0 {
-		return true
-	}
-
-	if dir.Mode&0002 > 0 {
-		return true
-	}
-
-	return false
-}
@@ -4,7 +4,6 @@ import (
 	"bytes"
 	"github.com/chrislusf/seaweedfs/weed/pb/filer_pb"
 	"github.com/chrislusf/seaweedfs/weed/wdclient"
-	"math"
 	"time"
 )

@@ -23,7 +22,7 @@ func ReadEntry(masterClient *wdclient.MasterClient, filerClient filer_pb.Seaweed
 		return err
 	}

-	return StreamContent(masterClient, byteBuffer, respLookupEntry.Entry.Chunks, 0, math.MaxInt64)
+	return StreamContent(masterClient, byteBuffer, respLookupEntry.Entry.Chunks, 0, int64(FileSize(respLookupEntry.Entry)))

 }

@@ -129,7 +129,7 @@ func (c *ChunkReadAt) doReadAt(p []byte, offset int64) (n int, err error) {
 		}
 		if startOffset < chunk.LogicOffset {
 			gap := int(chunk.LogicOffset - startOffset)
-			glog.V(4).Infof("zero [%d,%d)", startOffset, startOffset+int64(gap))
+			glog.V(4).Infof("zero [%d,%d)", startOffset, chunk.LogicOffset)
 			n += int(min(int64(gap), remaining))
 			startOffset, remaining = chunk.LogicOffset, remaining-int64(gap)
 			if remaining <= 0 {
@@ -80,11 +80,23 @@ func StreamContent(masterClient wdclient.HasLookupFileIdFunction, writer io.Writ
 		fileId2Url[chunkView.FileId] = urlStrings
 	}

+	remaining := size
 	for _, chunkView := range chunkViews {
-
+		if offset < chunkView.LogicOffset {
+			gap := chunkView.LogicOffset - offset
+			remaining -= gap
+			glog.V(4).Infof("zero [%d,%d)", offset, chunkView.LogicOffset)
+			err := writeZero(writer, gap)
+			if err != nil {
+				return fmt.Errorf("write zero [%d,%d)", offset, chunkView.LogicOffset)
+			}
+			offset = chunkView.LogicOffset
+		}
 		urlStrings := fileId2Url[chunkView.FileId]
 		start := time.Now()
 		err := retriedStreamFetchChunkData(writer, urlStrings, chunkView.CipherKey, chunkView.IsGzipped, chunkView.IsFullChunk(), chunkView.Offset, int(chunkView.Size))
+		offset += int64(chunkView.Size)
+		remaining -= int64(chunkView.Size)
 		stats.FilerRequestHistogram.WithLabelValues("chunkDownload").Observe(time.Since(start).Seconds())
 		if err != nil {
 			stats.FilerRequestCounter.WithLabelValues("chunkDownloadError").Inc()
@@ -92,6 +104,11 @@ func StreamContent(masterClient wdclient.HasLookupFileIdFunction, writer io.Writ
 		}
 		stats.FilerRequestCounter.WithLabelValues("chunkDownload").Inc()
 	}
+	glog.V(4).Infof("zero [%d,%d)", offset, offset+remaining)
+	err := writeZero(writer, remaining)
+	if err != nil {
+		return fmt.Errorf("write zero [%d,%d)", offset, offset+remaining)
+	}

 	return nil

@@ -99,6 +116,23 @@ func StreamContent(masterClient wdclient.HasLookupFileIdFunction, writer io.Writ

 // ----------------  ReadAllReader ----------------------------------

+func writeZero(w io.Writer, size int64) (err error) {
+	zeroPadding := make([]byte, 1024)
+	var written int
+	for size > 0 {
+		if size > 1024 {
+			written, err = w.Write(zeroPadding)
+		} else {
+			written, err = w.Write(zeroPadding[:size])
+		}
+		size -= int64(written)
+		if err != nil {
+			return
+		}
+	}
+	return
+}
+
 func ReadAll(masterClient *wdclient.MasterClient, chunks []*filer_pb.FileChunk) ([]byte, error) {

 	buffer := bytes.Buffer{}
@@ -66,7 +66,7 @@ func (dir *Dir) Attr(ctx context.Context, attr *fuse.Attr) error {
 	attr.Mode = os.FileMode(entry.Attributes.FileMode) | os.ModeDir
 	attr.Mtime = time.Unix(entry.Attributes.Mtime, 0)
 	attr.Crtime = time.Unix(entry.Attributes.Crtime, 0)
-	attr.Ctime = time.Unix(entry.Attributes.Crtime, 0)
+	attr.Ctime = time.Unix(entry.Attributes.Mtime, 0)
 	attr.Atime = time.Unix(entry.Attributes.Mtime, 0)
 	attr.Gid = entry.Attributes.Gid
 	attr.Uid = entry.Attributes.Uid
@@ -100,10 +100,10 @@ func (dir *Dir) Fsync(ctx context.Context, req *fuse.FsyncRequest) error {
 	return nil
 }

-func (dir *Dir) newFile(name string) fs.Node {
+func (dir *Dir) newFile(name string, fileMode os.FileMode) fs.Node {

 	fileFullPath := util.NewFullPath(dir.FullPath(), name)
-	fileId := fileFullPath.AsInode()
+	fileId := fileFullPath.AsInode(fileMode)
 	dir.wfs.handlesLock.Lock()
 	existingHandle, found := dir.wfs.handles[fileId]
 	dir.wfs.handlesLock.Unlock()
@@ -122,13 +122,17 @@ func (dir *Dir) newFile(name string) fs.Node {

 func (dir *Dir) newDirectory(fullpath util.FullPath) fs.Node {

-	return &Dir{name: fullpath.Name(), wfs: dir.wfs, parent: dir, id: fullpath.AsInode()}
+	return &Dir{name: fullpath.Name(), wfs: dir.wfs, parent: dir, id: fullpath.AsInode(os.ModeDir)}

 }

 func (dir *Dir) Create(ctx context.Context, req *fuse.CreateRequest,
 	resp *fuse.CreateResponse) (fs.Node, fs.Handle, error) {

+	if err := checkName(req.Name); err != nil {
+		return nil, nil, err
+	}
+
 	exclusive := req.Flags&fuse.OpenExclusive != 0
 	isDirectory := req.Mode&os.ModeDir > 0

@@ -144,7 +148,7 @@ func (dir *Dir) Create(ctx context.Context, req *fuse.CreateRequest,
 		return node, node, nil
 	}

-	node = dir.newFile(req.Name)
+	node = dir.newFile(req.Name, req.Mode)
 	file := node.(*File)
 	file.entry = &filer_pb.Entry{
 		Name:        req.Name,
@@ -168,13 +172,19 @@ func (dir *Dir) Create(ctx context.Context, req *fuse.CreateRequest,

 func (dir *Dir) Mknod(ctx context.Context, req *fuse.MknodRequest) (fs.Node, error) {

+	if err := checkName(req.Name); err != nil {
+		return nil, err
+	}
+
+	glog.V(3).Infof("dir %s Mknod %+v", dir.FullPath(), req)
+
 	_, err := dir.doCreateEntry(req.Name, req.Mode, req.Uid, req.Gid, false)

 	if err != nil {
 		return nil, err
 	}
 	var node fs.Node
-	node = dir.newFile(req.Name)
+	node = dir.newFile(req.Name, req.Mode)
 	return node, nil
 }

@@ -226,6 +236,10 @@ func (dir *Dir) doCreateEntry(name string, mode os.FileMode, uid, gid uint32, ex

 func (dir *Dir) Mkdir(ctx context.Context, req *fuse.MkdirRequest) (fs.Node, error) {

+	if err := checkName(req.Name); err != nil {
+		return nil, err
+	}
+
 	glog.V(4).Infof("mkdir %s: %s", dir.FullPath(), req.Name)

 	newEntry := &filer_pb.Entry{
@@ -280,6 +294,10 @@ func (dir *Dir) Mkdir(ctx context.Context, req *fuse.MkdirRequest) (fs.Node, err

 func (dir *Dir) Lookup(ctx context.Context, req *fuse.LookupRequest, resp *fuse.LookupResponse) (node fs.Node, err error) {

+	if err := checkName(req.Name); err != nil {
+		return nil, err
+	}
+
 	dirPath := util.FullPath(dir.FullPath())
 	// glog.V(4).Infof("dir Lookup %s: %s by %s", dirPath, req.Name, req.Header.String())

@@ -310,11 +328,10 @@ func (dir *Dir) Lookup(ctx context.Context, req *fuse.LookupRequest, resp *fuse.
 		if localEntry.IsDirectory() {
 			node = dir.newDirectory(fullFilePath)
 		} else {
-			node = dir.newFile(req.Name)
+			node = dir.newFile(req.Name, localEntry.Attr.Mode)
 		}

 		// resp.EntryValid = time.Second
-		resp.Attr.Inode = fullFilePath.AsInode()
 		resp.Attr.Valid = time.Second
 		resp.Attr.Size = localEntry.FileSize
 		resp.Attr.Mtime = localEntry.Attr.Mtime
@@ -340,10 +357,10 @@ func (dir *Dir) ReadDirAll(ctx context.Context) (ret []fuse.Dirent, err error) {

 	processEachEntryFn := func(entry *filer.Entry, isLast bool) {
 		if entry.IsDirectory() {
-			dirent := fuse.Dirent{Name: entry.Name(), Type: fuse.DT_Dir, Inode: dirPath.Child(entry.Name()).AsInode()}
+			dirent := fuse.Dirent{Name: entry.Name(), Type: fuse.DT_Dir, Inode: dirPath.Child(entry.Name()).AsInode(os.ModeDir)}
 			ret = append(ret, dirent)
 		} else {
-			dirent := fuse.Dirent{Name: entry.Name(), Type: findFileType(uint16(entry.Attr.Mode)), Inode: dirPath.Child(entry.Name()).AsInode()}
+			dirent := fuse.Dirent{Name: entry.Name(), Type: findFileType(uint16(entry.Attr.Mode)), Inode: dirPath.Child(entry.Name()).AsInode(entry.Attr.Mode)}
 			ret = append(ret, dirent)
 		}
 	}
@@ -363,7 +380,7 @@ func (dir *Dir) ReadDirAll(ctx context.Context) (ret []fuse.Dirent, err error) {

 	// create proper . and .. directories
 	ret = append(ret, fuse.Dirent{
-		Inode: dirPath.AsInode(),
+		Inode: dirPath.AsInode(os.ModeDir),
 		Name:  ".",
 		Type:  fuse.DT_Dir,
 	})
@@ -373,7 +390,7 @@ func (dir *Dir) ReadDirAll(ctx context.Context) (ret []fuse.Dirent, err error) {
 	if string(dirPath) == dir.wfs.option.FilerMountRootPath {
 		inode = dir.wfs.option.MountParentInode
 	} else {
-		inode = util.FullPath(dir.parent.FullPath()).AsInode()
+		inode = util.FullPath(dir.parent.FullPath()).AsInode(os.ModeDir)
 	}

 	ret = append(ret, fuse.Dirent{
@@ -407,15 +424,14 @@ func findFileType(mode uint16) fuse.DirentType {

 func (dir *Dir) Remove(ctx context.Context, req *fuse.RemoveRequest) error {

-	if !req.Dir {
-		return dir.removeOneFile(req)
+	parentHasPermission := false
+	parentEntry, err := dir.maybeLoadEntry()
+	if err != nil {
+		return err
+	}
+	if err := checkPermission(parentEntry, req.Uid, req.Gid, true); err == nil {
+		parentHasPermission = true
 	}
-
-	return dir.removeFolder(req)
-
-}
-
-func (dir *Dir) removeOneFile(req *fuse.RemoveRequest) error {

 	dirFullPath := dir.FullPath()
 	filePath := util.NewFullPath(dirFullPath, req.Name)
@@ -423,11 +439,29 @@ func (dir *Dir) removeOneFile(req *fuse.RemoveRequest) error {
 	if err != nil {
 		return err
 	}
+	if !parentHasPermission {
+		if err := checkPermission(entry, req.Uid, req.Gid, true); err != nil {
+			return err
+		}
+	}
+
+	if !req.Dir {
+		return dir.removeOneFile(entry, req)
+	}
+
+	return dir.removeFolder(entry, req)
+
+}
+
+func (dir *Dir) removeOneFile(entry *filer_pb.Entry, req *fuse.RemoveRequest) error {
+
+	dirFullPath := dir.FullPath()
+	filePath := util.NewFullPath(dirFullPath, req.Name)

 	// first, ensure the filer store can correctly delete
 	glog.V(3).Infof("remove file: %v", req)
 	isDeleteData := entry != nil && entry.HardLinkCounter <= 1
-	err = filer_pb.Remove(dir.wfs, dirFullPath, req.Name, isDeleteData, false, false, false, []int32{dir.wfs.signature})
+	err := filer_pb.Remove(dir.wfs, dirFullPath, req.Name, isDeleteData, false, false, false, []int32{dir.wfs.signature})
 	if err != nil {
 		glog.V(3).Infof("not found remove file %s: %v", filePath, err)
 		return fuse.ENOENT
@@ -442,7 +476,7 @@ func (dir *Dir) removeOneFile(req *fuse.RemoveRequest) error {
 	// remove current file handle if any
 	dir.wfs.handlesLock.Lock()
 	defer dir.wfs.handlesLock.Unlock()
-	inodeId := filePath.AsInode()
+	inodeId := filePath.AsInode(0)
 	if fh, ok := dir.wfs.handles[inodeId]; ok {
 		delete(dir.wfs.handles, inodeId)
 		fh.isDeleted = true
@@ -452,15 +486,16 @@ func (dir *Dir) removeOneFile(req *fuse.RemoveRequest) error {

 }

-func (dir *Dir) removeFolder(req *fuse.RemoveRequest) error {
+func (dir *Dir) removeFolder(entry *filer_pb.Entry, req *fuse.RemoveRequest) error {

 	dirFullPath := dir.FullPath()
+
 	glog.V(3).Infof("remove directory entry: %v", req)
 	ignoreRecursiveErr := true // ignore recursion error since the OS should manage it
 	err := filer_pb.Remove(dir.wfs, dirFullPath, req.Name, true, true, ignoreRecursiveErr, false, []int32{dir.wfs.signature})
 	if err != nil {
 		glog.V(0).Infof("remove %s/%s: %v", dirFullPath, req.Name, err)
-		if strings.Contains(err.Error(), "non-empty") {
+		if strings.Contains(err.Error(), filer.MsgFailDelNonEmptyFolder) {
 			return fuse.EEXIST
 		}
 		return fuse.ENOENT
@@ -475,7 +510,7 @@ func (dir *Dir) removeFolder(req *fuse.RemoveRequest) error {

 func (dir *Dir) Setattr(ctx context.Context, req *fuse.SetattrRequest, resp *fuse.SetattrResponse) error {

-	glog.V(4).Infof("%v dir setattr %+v", dir.FullPath(), req)
+	glog.V(4).Infof("%v dir setattr %+v mode=%d", dir.FullPath(), req, req.Mode)

 	entry, err := dir.maybeLoadEntry()
 	if err != nil {
@@ -498,6 +533,8 @@ func (dir *Dir) Setattr(ctx context.Context, req *fuse.SetattrRequest, resp *fus
 		entry.Attributes.Mtime = req.Mtime.Unix()
 	}

+	entry.Attributes.Mtime = time.Now().Unix()
+
 	return dir.saveEntry(entry)

 }
@@ -24,6 +24,10 @@ const (

 func (dir *Dir) Link(ctx context.Context, req *fuse.LinkRequest, old fs.Node) (fs.Node, error) {

+	if err := checkName(req.NewName); err != nil {
+		return nil, err
+	}
+
 	oldFile, ok := old.(*File)
 	if !ok {
 		glog.Errorf("old node is not a file: %+v", old)
@@ -53,6 +57,7 @@ func (dir *Dir) Link(ctx context.Context, req *fuse.LinkRequest, old fs.Node) (f
 	}

 	// CreateLink 1.2 : update new file to hardlink mode
+	oldEntry.Attributes.Mtime = time.Now().Unix()
 	request := &filer_pb.CreateEntryRequest{
 		Directory: dir.FullPath(),
 		Entry: &filer_pb.Entry{
@@ -93,7 +98,7 @@ func (dir *Dir) Link(ctx context.Context, req *fuse.LinkRequest, old fs.Node) (f
 	}

 	// create new file node
-	newNode := dir.newFile(req.NewName)
+	newNode := dir.newFile(req.NewName, 0)
 	newFile := newNode.(*File)

 	return newFile, err
@@ -102,6 +107,10 @@ func (dir *Dir) Link(ctx context.Context, req *fuse.LinkRequest, old fs.Node) (f

 func (dir *Dir) Symlink(ctx context.Context, req *fuse.SymlinkRequest) (fs.Node, error) {

+	if err := checkName(req.NewName); err != nil {
+		return nil, err
+	}
+
 	glog.V(4).Infof("Symlink: %v/%v to %v", dir.FullPath(), req.NewName, req.Target)

 	request := &filer_pb.CreateEntryRequest{
@@ -136,7 +145,7 @@ func (dir *Dir) Symlink(ctx context.Context, req *fuse.SymlinkRequest) (fs.Node,
 		return nil
 	})

-	symlink := dir.newFile(req.NewName)
+	symlink := dir.newFile(req.NewName, os.ModeSymlink)

 	return symlink, err

@@ -3,17 +3,25 @@ package filesys
 import (
 	"context"
 	"github.com/chrislusf/seaweedfs/weed/filer"
-	"github.com/seaweedfs/fuse"
-	"github.com/seaweedfs/fuse/fs"
-	"io"
-
 	"github.com/chrislusf/seaweedfs/weed/glog"
 	"github.com/chrislusf/seaweedfs/weed/pb/filer_pb"
 	"github.com/chrislusf/seaweedfs/weed/util"
+	"github.com/seaweedfs/fuse"
+	"github.com/seaweedfs/fuse/fs"
+	"io"
+	"os"
+	"strings"
 )

 func (dir *Dir) Rename(ctx context.Context, req *fuse.RenameRequest, newDirectory fs.Node) error {

+	if err := checkName(req.NewName); err != nil {
+		return err
+	}
+	if err := checkName(req.OldName); err != nil {
+		return err
+	}
+
 	newDir := newDirectory.(*Dir)

 	newPath := util.NewFullPath(newDir.FullPath(), req.NewName)
@@ -37,7 +45,7 @@ func (dir *Dir) Rename(ctx context.Context, req *fuse.RenameRequest, newDirector
 		stream, err := client.StreamRenameEntry(ctx, request)
 		if err != nil {
 			glog.Errorf("dir AtomicRenameEntry %s => %s : %v", oldPath, newPath, err)
-			return fuse.EXDEV
+			return fuse.EIO
 		}

 		for {
@@ -46,11 +54,19 @@ func (dir *Dir) Rename(ctx context.Context, req *fuse.RenameRequest, newDirector
 				if recvErr == io.EOF {
 					break
 				} else {
+					glog.V(0).Infof("dir Rename %s => %s receive: %v", oldPath, newPath, recvErr)
+					if strings.Contains(recvErr.Error(), "not empty") {
+						return fuse.EEXIST
+					}
+					if strings.Contains(recvErr.Error(), "not directory") {
+						return fuse.ENOTDIR
+					}
 					return recvErr
 				}
 			}

 			if err = dir.handleRenameResponse(ctx, resp); err != nil {
+				glog.V(0).Infof("dir Rename %s => %s : %v", oldPath, newPath, err)
 				return err
 			}

@@ -59,12 +75,8 @@ func (dir *Dir) Rename(ctx context.Context, req *fuse.RenameRequest, newDirector
 		return nil

 	})
-	if err != nil {
-		glog.V(0).Infof("dir Rename %s => %s : %v", oldPath, newPath, err)
-		return fuse.EIO
-	}

-	return nil
+	return err
 }

 func (dir *Dir) handleRenameResponse(ctx context.Context, resp *filer_pb.StreamRenameEntryResponse) error {
@@ -80,11 +92,13 @@ func (dir *Dir) handleRenameResponse(ctx context.Context, resp *filer_pb.StreamR
 		oldParent, newParent := util.FullPath(resp.Directory), util.FullPath(resp.EventNotification.NewParentPath)
 		oldName, newName := resp.EventNotification.OldEntry.Name, resp.EventNotification.NewEntry.Name

+		entryFileMode := newEntry.Attr.Mode
+
 		oldPath := oldParent.Child(oldName)
 		newPath := newParent.Child(newName)
-		oldFsNode := NodeWithId(oldPath.AsInode())
-		newFsNode := NodeWithId(newPath.AsInode())
-		newDirNode, found := dir.wfs.Server.FindInternalNode(NodeWithId(newParent.AsInode()))
+		oldFsNode := NodeWithId(oldPath.AsInode(entryFileMode))
+		newFsNode := NodeWithId(newPath.AsInode(entryFileMode))
+		newDirNode, found := dir.wfs.Server.FindInternalNode(NodeWithId(newParent.AsInode(os.ModeDir)))
 		var newDir *Dir
 		if found {
 			newDir = newDirNode.(*Dir)
@@ -109,15 +123,19 @@ func (dir *Dir) handleRenameResponse(ctx context.Context, resp *filer_pb.StreamR
 		})

 		// change file handle
-		inodeId := oldPath.AsInode()
-		dir.wfs.handlesLock.Lock()
-		if existingHandle, found := dir.wfs.handles[inodeId]; found && existingHandle != nil {
-			glog.V(4).Infof("opened file handle %s => %s", oldPath, newPath)
-			delete(dir.wfs.handles, inodeId)
-			existingHandle.f.entry.Name = newName
-			dir.wfs.handles[newPath.AsInode()] = existingHandle
+		if !newEntry.IsDirectory() {
+			inodeId := oldPath.AsInode(entryFileMode)
+			dir.wfs.handlesLock.Lock()
+			if existingHandle, found := dir.wfs.handles[inodeId]; found && existingHandle != nil {
+				glog.V(4).Infof("opened file handle %s => %s", oldPath, newPath)
+				delete(dir.wfs.handles, inodeId)
+				existingHandle.handle = newPath.AsInode(entryFileMode)
+				existingHandle.f.entry.Name = newName
+				existingHandle.f.id = newPath.AsInode(entryFileMode)
+				dir.wfs.handles[newPath.AsInode(entryFileMode)] = existingHandle
+			}
+			dir.wfs.handlesLock.Unlock()
 		}
-		dir.wfs.handlesLock.Unlock()

 	} else if resp.EventNotification.OldEntry != nil {
 		// without new entry, only old entry name exists. This is the second step to delete old entry
@@ -0,0 +1,100 @@
+package filesys
+
+import (
+	"fmt"
+	"github.com/chrislusf/seaweedfs/weed/filesys/page_writer"
+	"github.com/chrislusf/seaweedfs/weed/glog"
+	"github.com/chrislusf/seaweedfs/weed/pb/filer_pb"
+	"io"
+	"sync"
+	"time"
+)
+
+type ChunkedDirtyPages struct {
+	fh             *FileHandle
+	writeWaitGroup sync.WaitGroup
+	chunkAddLock   sync.Mutex
+	lastErr        error
+	collection     string
+	replication    string
+	uploadPipeline *page_writer.UploadPipeline
+	hasWrites      bool
+}
+
+var (
+	_ = page_writer.DirtyPages(&ChunkedDirtyPages{})
+)
+
+func newMemoryChunkPages(fh *FileHandle, chunkSize int64) *ChunkedDirtyPages {
+
+	dirtyPages := &ChunkedDirtyPages{
+		fh: fh,
+	}
+
+	dirtyPages.uploadPipeline = page_writer.NewUploadPipeline(fh.f.wfs.concurrentWriters, chunkSize, dirtyPages.saveChunkedFileIntevalToStorage, fh.f.wfs.option.ConcurrentWriters)
+
+	return dirtyPages
+}
+
+func (pages *ChunkedDirtyPages) AddPage(offset int64, data []byte) {
+	pages.hasWrites = true
+
+	glog.V(4).Infof("%v memory AddPage [%d, %d)", pages.fh.f.fullpath(), offset, offset+int64(len(data)))
+	pages.uploadPipeline.SaveDataAt(data, offset)
+
+	return
+}
+
+func (pages *ChunkedDirtyPages) FlushData() error {
+	if !pages.hasWrites {
+		return nil
+	}
+	pages.uploadPipeline.FlushAll()
+	if pages.lastErr != nil {
+		return fmt.Errorf("flush data: %v", pages.lastErr)
+	}
+	return nil
+}
+
+func (pages *ChunkedDirtyPages) ReadDirtyDataAt(data []byte, startOffset int64) (maxStop int64) {
+	if !pages.hasWrites {
+		return
+	}
+	return pages.uploadPipeline.MaybeReadDataAt(data, startOffset)
+}
+
+func (pages *ChunkedDirtyPages) GetStorageOptions() (collection, replication string) {
+	return pages.collection, pages.replication
+}
+
+func (pages *ChunkedDirtyPages) saveChunkedFileIntevalToStorage(reader io.Reader, offset int64, size int64, cleanupFn func()) {
+
+	mtime := time.Now().UnixNano()
+	defer cleanupFn()
+
+	chunk, collection, replication, err := pages.fh.f.wfs.saveDataAsChunk(pages.fh.f.fullpath())(reader, pages.fh.f.Name, offset)
+	if err != nil {
+		glog.V(0).Infof("%s saveToStorage [%d,%d): %v", pages.fh.f.fullpath(), offset, offset+size, err)
+		pages.lastErr = err
+		return
+	}
+	chunk.Mtime = mtime
+	pages.collection, pages.replication = collection, replication
+	pages.chunkAddLock.Lock()
+	pages.fh.f.addChunks([]*filer_pb.FileChunk{chunk})
+	pages.fh.entryViewCache = nil
+	glog.V(3).Infof("%s saveToStorage %s [%d,%d)", pages.fh.f.fullpath(), chunk.FileId, offset, offset+size)
+	pages.chunkAddLock.Unlock()
+
+}
+
+func (pages ChunkedDirtyPages) Destroy() {
+	pages.uploadPipeline.Shutdown()
+}
+
+func (pages *ChunkedDirtyPages) LockForRead(startOffset, stopOffset int64) {
+	pages.uploadPipeline.LockForRead(startOffset, stopOffset)
+}
+func (pages *ChunkedDirtyPages) UnlockForRead(startOffset, stopOffset int64) {
+	pages.uploadPipeline.UnlockForRead(startOffset, stopOffset)
+}
@@ -1,138 +0,0 @@
-package filesys
-
-import (
-	"bytes"
-	"fmt"
-	"github.com/chrislusf/seaweedfs/weed/filesys/page_writer"
-	"io"
-	"sync"
-	"time"
-
-	"github.com/chrislusf/seaweedfs/weed/glog"
-	"github.com/chrislusf/seaweedfs/weed/pb/filer_pb"
-)
-
-type ContinuousDirtyPages struct {
-	intervals      *page_writer.ContinuousIntervals
-	f              *File
-	writeWaitGroup sync.WaitGroup
-	chunkAddLock   sync.Mutex
-	lastErr        error
-	collection     string
-	replication    string
-}
-
-func newContinuousDirtyPages(file *File) *ContinuousDirtyPages {
-	dirtyPages := &ContinuousDirtyPages{
-		intervals: &page_writer.ContinuousIntervals{},
-		f:         file,
-	}
-	return dirtyPages
-}
-
-func (pages *ContinuousDirtyPages) AddPage(offset int64, data []byte) {
-
-	glog.V(4).Infof("%s AddPage [%d,%d)", pages.f.fullpath(), offset, offset+int64(len(data)))
-
-	if len(data) > int(pages.f.wfs.option.ChunkSizeLimit) {
-		// this is more than what buffer can hold.
-		pages.flushAndSave(offset, data)
-	}
-
-	pages.intervals.AddInterval(data, offset)
-
-	if pages.intervals.TotalSize() >= pages.f.wfs.option.ChunkSizeLimit {
-		pages.saveExistingLargestPageToStorage()
-	}
-
-	return
-}
-
-func (pages *ContinuousDirtyPages) flushAndSave(offset int64, data []byte) {
-
-	// flush existing
-	pages.saveExistingPagesToStorage()
-
-	// flush the new page
-	pages.saveToStorage(bytes.NewReader(data), offset, int64(len(data)))
-
-	return
-}
-
-func (pages *ContinuousDirtyPages) FlushData() error {
-
-	pages.saveExistingPagesToStorage()
-	pages.writeWaitGroup.Wait()
-	if pages.lastErr != nil {
-		return fmt.Errorf("flush data: %v", pages.lastErr)
-	}
-	return nil
-}
-
-func (pages *ContinuousDirtyPages) saveExistingPagesToStorage() {
-	for pages.saveExistingLargestPageToStorage() {
-	}
-}
-
-func (pages *ContinuousDirtyPages) saveExistingLargestPageToStorage() (hasSavedData bool) {
-
-	maxList := pages.intervals.RemoveLargestIntervalLinkedList()
-	if maxList == nil {
-		return false
-	}
-
-	entry := pages.f.getEntry()
-	if entry == nil {
-		return false
-	}
-
-	fileSize := int64(entry.Attributes.FileSize)
-
-	chunkSize := min(maxList.Size(), fileSize-maxList.Offset())
-	if chunkSize == 0 {
-		return false
-	}
-
-	pages.saveToStorage(maxList.ToReader(), maxList.Offset(), chunkSize)
-
-	return true
-}
-
-func (pages *ContinuousDirtyPages) saveToStorage(reader io.Reader, offset int64, size int64) {
-
-	mtime := time.Now().UnixNano()
-	pages.writeWaitGroup.Add(1)
-	writer := func() {
-		defer pages.writeWaitGroup.Done()
-
-		reader = io.LimitReader(reader, size)
-		chunk, collection, replication, err := pages.f.wfs.saveDataAsChunk(pages.f.fullpath())(reader, pages.f.Name, offset)
-		if err != nil {
-			glog.V(0).Infof("%s saveToStorage [%d,%d): %v", pages.f.fullpath(), offset, offset+size, err)
-			pages.lastErr = err
-			return
-		}
-		chunk.Mtime = mtime
-		pages.collection, pages.replication = collection, replication
-		pages.chunkAddLock.Lock()
-		defer pages.chunkAddLock.Unlock()
-		pages.f.addChunks([]*filer_pb.FileChunk{chunk})
-		glog.V(3).Infof("%s saveToStorage [%d,%d)", pages.f.fullpath(), offset, offset+size)
-	}
-
-	if pages.f.wfs.concurrentWriters != nil {
-		pages.f.wfs.concurrentWriters.Execute(writer)
-	} else {
-		go writer()
-	}
-}
-
-func (pages *ContinuousDirtyPages) ReadDirtyDataAt(data []byte, startOffset int64) (maxStop int64) {
-	return pages.intervals.ReadDataAt(data, startOffset)
-}
-
-func (pages *ContinuousDirtyPages) GetStorageOptions() (collection, replication string) {
-	return pages.collection, pages.replication
-}
-func (pages ContinuousDirtyPages) Destroy() {
-}
@@ -1,106 +0,0 @@
-package filesys
-
-import (
-	"fmt"
-	"github.com/chrislusf/seaweedfs/weed/filesys/page_writer"
-	"github.com/chrislusf/seaweedfs/weed/glog"
-	"github.com/chrislusf/seaweedfs/weed/pb/filer_pb"
-	"io"
-	"sync"
-	"time"
-)
-
-type StreamDirtyPages struct {
-	f              *File
-	writeWaitGroup sync.WaitGroup
-	pageAddLock    sync.Mutex
-	chunkAddLock   sync.Mutex
-	lastErr        error
-	collection     string
-	replication    string
-	chunkedStream  *page_writer.ChunkedStreamWriter
-}
-
-func newStreamDirtyPages(file *File, chunkSize int64) *StreamDirtyPages {
-
-	dirtyPages := &StreamDirtyPages{
-		f:             file,
-		chunkedStream: page_writer.NewChunkedStreamWriter(chunkSize),
-	}
-
-	dirtyPages.chunkedStream.SetSaveToStorageFunction(dirtyPages.saveChunkedFileIntevalToStorage)
-
-	return dirtyPages
-}
-
-func (pages *StreamDirtyPages) AddPage(offset int64, data []byte) {
-
-	pages.pageAddLock.Lock()
-	defer pages.pageAddLock.Unlock()
-
-	glog.V(4).Infof("%v stream AddPage [%d, %d)", pages.f.fullpath(), offset, offset+int64(len(data)))
-	if _, err := pages.chunkedStream.WriteAt(data, offset); err != nil {
-		pages.lastErr = err
-	}
-
-	return
-}
-
-func (pages *StreamDirtyPages) FlushData() error {
-	pages.saveChunkedFileToStorage()
-	pages.writeWaitGroup.Wait()
-	if pages.lastErr != nil {
-		return fmt.Errorf("flush data: %v", pages.lastErr)
-	}
-	pages.chunkedStream.Reset()
-	return nil
-}
-
-func (pages *StreamDirtyPages) ReadDirtyDataAt(data []byte, startOffset int64) (maxStop int64) {
-	return pages.chunkedStream.ReadDataAt(data, startOffset)
-}
-
-func (pages *StreamDirtyPages) GetStorageOptions() (collection, replication string) {
-	return pages.collection, pages.replication
-}
-
-func (pages *StreamDirtyPages) saveChunkedFileToStorage() {
-
-	pages.chunkedStream.FlushAll()
-
-}
-
-func (pages *StreamDirtyPages) saveChunkedFileIntevalToStorage(reader io.Reader, offset int64, size int64, cleanupFn func()) {
-
-	mtime := time.Now().UnixNano()
-	pages.writeWaitGroup.Add(1)
-	writer := func() {
-		defer pages.writeWaitGroup.Done()
-		defer cleanupFn()
-
-		chunk, collection, replication, err := pages.f.wfs.saveDataAsChunk(pages.f.fullpath())(reader, pages.f.Name, offset)
-		if err != nil {
-			glog.V(0).Infof("%s saveToStorage [%d,%d): %v", pages.f.fullpath(), offset, offset+size, err)
-			pages.lastErr = err
-			return
-		}
-		chunk.Mtime = mtime
-		pages.collection, pages.replication = collection, replication
-		pages.chunkAddLock.Lock()
-		pages.f.addChunks([]*filer_pb.FileChunk{chunk})
-		glog.V(3).Infof("%s saveToStorage %s [%d,%d)", pages.f.fullpath(), chunk.FileId, offset, offset+size)
-		pages.chunkAddLock.Unlock()
-
-	}
-
-	if pages.f.wfs.concurrentWriters != nil {
-		pages.f.wfs.concurrentWriters.Execute(writer)
-	} else {
-		go writer()
-	}
-
-}
-
-func (pages StreamDirtyPages) Destroy() {
-	pages.chunkedStream.Reset()
-}
@@ -1,106 +0,0 @@
-package filesys
-
-import (
-	"fmt"
-	"github.com/chrislusf/seaweedfs/weed/filesys/page_writer"
-	"github.com/chrislusf/seaweedfs/weed/glog"
-	"github.com/chrislusf/seaweedfs/weed/pb/filer_pb"
-	"io"
-	"os"
-	"sync"
-	"time"
-)
-
-type TempFileDirtyPages struct {
-	f              *File
-	writeWaitGroup sync.WaitGroup
-	pageAddLock    sync.Mutex
-	chunkAddLock   sync.Mutex
-	lastErr        error
-	collection     string
-	replication    string
-	chunkedFile    *page_writer.ChunkedFileWriter
-}
-
-func newTempFileDirtyPages(file *File, chunkSize int64) *TempFileDirtyPages {
-
-	tempFile := &TempFileDirtyPages{
-		f:           file,
-		chunkedFile: page_writer.NewChunkedFileWriter(file.wfs.option.getTempFilePageDir(), chunkSize),
-	}
-
-	return tempFile
-}
-
-func (pages *TempFileDirtyPages) AddPage(offset int64, data []byte) {
-
-	pages.pageAddLock.Lock()
-	defer pages.pageAddLock.Unlock()
-
-	glog.V(4).Infof("%v tempfile AddPage [%d, %d)", pages.f.fullpath(), offset, offset+int64(len(data)))
-	if _, err := pages.chunkedFile.WriteAt(data, offset); err != nil {
-		pages.lastErr = err
-	}
-
-	return
-}
-
-func (pages *TempFileDirtyPages) FlushData() error {
-	pages.saveChunkedFileToStorage()
-	pages.writeWaitGroup.Wait()
-	if pages.lastErr != nil {
-		return fmt.Errorf("flush data: %v", pages.lastErr)
-	}
-	pages.chunkedFile.Reset()
-	return nil
-}
-
-func (pages *TempFileDirtyPages) ReadDirtyDataAt(data []byte, startOffset int64) (maxStop int64) {
-	return pages.chunkedFile.ReadDataAt(data, startOffset)
-}
-
-func (pages *TempFileDirtyPages) GetStorageOptions() (collection, replication string) {
-	return pages.collection, pages.replication
-}
-
-func (pages *TempFileDirtyPages) saveChunkedFileToStorage() {
-
-	pages.chunkedFile.ProcessEachInterval(func(file *os.File, logicChunkIndex page_writer.LogicChunkIndex, interval *page_writer.ChunkWrittenInterval) {
-		reader := page_writer.NewFileIntervalReader(pages.chunkedFile, logicChunkIndex, interval)
-		pages.saveChunkedFileIntevalToStorage(reader, int64(logicChunkIndex)*pages.chunkedFile.ChunkSize+interval.StartOffset, interval.Size())
-	})
-
-}
-
-func (pages *TempFileDirtyPages) saveChunkedFileIntevalToStorage(reader io.Reader, offset int64, size int64) {
-
-	mtime := time.Now().UnixNano()
-	pages.writeWaitGroup.Add(1)
-	writer := func() {
-		defer pages.writeWaitGroup.Done()
-
-		chunk, collection, replication, err := pages.f.wfs.saveDataAsChunk(pages.f.fullpath())(reader, pages.f.Name, offset)
-		if err != nil {
-			glog.V(0).Infof("%s saveToStorage [%d,%d): %v", pages.f.fullpath(), offset, offset+size, err)
-			pages.lastErr = err
-			return
-		}
-		chunk.Mtime = mtime
-		pages.collection, pages.replication = collection, replication
-		pages.chunkAddLock.Lock()
-		defer pages.chunkAddLock.Unlock()
-		pages.f.addChunks([]*filer_pb.FileChunk{chunk})
-		glog.V(3).Infof("%s saveToStorage %s [%d,%d)", pages.f.fullpath(), chunk.FileId, offset, offset+size)
-	}
-
-	if pages.f.wfs.concurrentWriters != nil {
-		pages.f.wfs.concurrentWriters.Execute(writer)
-	} else {
-		go writer()
-	}
-
-}
-
-func (pages TempFileDirtyPages) Destroy() {
-	pages.chunkedFile.Reset()
-}
@@ -68,11 +68,12 @@ func (file *File) Attr(ctx context.Context, attr *fuse.Attr) (err error) {
 		glog.V(4).Infof("file Attr %s, open:%v, size: %d", file.fullpath(), file.isOpen, attr.Size)
 	}
 	attr.Crtime = time.Unix(entry.Attributes.Crtime, 0)
+	attr.Ctime = time.Unix(entry.Attributes.Mtime, 0)
 	attr.Mtime = time.Unix(entry.Attributes.Mtime, 0)
 	attr.Gid = entry.Attributes.Gid
 	attr.Uid = entry.Attributes.Uid
-	attr.Blocks = attr.Size/blockSize + 1
-	attr.BlockSize = uint32(file.wfs.option.ChunkSizeLimit)
+	attr.BlockSize = blockSize
+	attr.Blocks = (attr.Size + blockSize - 1) / blockSize
 	if entry.HardLinkCounter > 0 {
 		attr.Nlink = uint32(entry.HardLinkCounter)
 	}
@@ -110,7 +111,7 @@ func (file *File) Open(ctx context.Context, req *fuse.OpenRequest, resp *fuse.Op

 func (file *File) Setattr(ctx context.Context, req *fuse.SetattrRequest, resp *fuse.SetattrResponse) error {

-	glog.V(4).Infof("%v file setattr %+v", file.fullpath(), req)
+	glog.V(4).Infof("%v file setattr %+v mode=%d", file.fullpath(), req, req.Mode)

 	entry, err := file.maybeLoadEntry(ctx)
 	if err != nil {
@@ -139,29 +140,42 @@ func (file *File) Setattr(ctx context.Context, req *fuse.SetattrRequest, resp *f
 					}
 				}
 			}
+			// set the new chunks and reset entry cache
 			entry.Chunks = chunks
+			file.wfs.handlesLock.Lock()
+			existingHandle, found := file.wfs.handles[file.Id()]
+			file.wfs.handlesLock.Unlock()
+			if found {
+				existingHandle.entryViewCache = nil
+			}
+
 		}
+		entry.Attributes.Mtime = time.Now().Unix()
 		entry.Attributes.FileSize = req.Size
 		file.dirtyMetadata = true
 	}

 	if req.Valid.Mode() && entry.Attributes.FileMode != uint32(req.Mode) {
 		entry.Attributes.FileMode = uint32(req.Mode)
+		entry.Attributes.Mtime = time.Now().Unix()
 		file.dirtyMetadata = true
 	}

 	if req.Valid.Uid() && entry.Attributes.Uid != req.Uid {
 		entry.Attributes.Uid = req.Uid
+		entry.Attributes.Mtime = time.Now().Unix()
 		file.dirtyMetadata = true
 	}

 	if req.Valid.Gid() && entry.Attributes.Gid != req.Gid {
 		entry.Attributes.Gid = req.Gid
+		entry.Attributes.Mtime = time.Now().Unix()
 		file.dirtyMetadata = true
 	}

 	if req.Valid.Crtime() {
 		entry.Attributes.Crtime = req.Crtime.Unix()
+		entry.Attributes.Mtime = time.Now().Unix()
 		file.dirtyMetadata = true
 	}

@@ -248,17 +262,19 @@ func (file *File) Listxattr(ctx context.Context, req *fuse.ListxattrRequest, res
 }

 func (file *File) Fsync(ctx context.Context, req *fuse.FsyncRequest) error {
-	// fsync works at OS level
+
 	// write the file chunks to the filerGrpcAddress
 	glog.V(4).Infof("%s/%s fsync file %+v", file.dir.FullPath(), file.Name, req)

-	return nil
+	return file.wfs.Fsync(file, req.Header)
+
 }

 func (file *File) Forget() {
 	t := util.NewFullPath(file.dir.FullPath(), file.Name)
 	glog.V(4).Infof("Forget file %s", t)
-	file.wfs.ReleaseHandle(t, fuse.HandleID(t.AsInode()))
+	file.wfs.ReleaseHandle(t, fuse.HandleID(t.AsInode(file.entry.FileMode())))
+
 }

 func (file *File) maybeLoadEntry(ctx context.Context) (entry *filer_pb.Entry, err error) {
@@ -26,25 +26,22 @@ type FileHandle struct {
 	contentType    string
 	handle         uint64
 	sync.Mutex
-	sync.WaitGroup

 	f         *File
-	RequestId fuse.RequestID // unique ID for request
-	NodeId    fuse.NodeID    // file or directory the request is about
-	Uid       uint32         // user ID of process making request
-	Gid       uint32         // group ID of process making request
-	writeOnly bool
+	NodeId    fuse.NodeID // file or directory the request is about
+	Uid       uint32      // user ID of process making request
+	Gid       uint32      // group ID of process making request
 	isDeleted bool
 }

 func newFileHandle(file *File, uid, gid uint32) *FileHandle {
 	fh := &FileHandle{
-		f: file,
-		// dirtyPages: newContinuousDirtyPages(file, writeOnly),
-		dirtyPages: newPageWriter(file, file.wfs.option.ChunkSizeLimit),
-		Uid:        uid,
-		Gid:        gid,
+		f:   file,
+		Uid: uid,
+		Gid: gid,
 	}
+	// dirtyPages: newContinuousDirtyPages(file, writeOnly),
+	fh.dirtyPages = newPageWriter(fh, file.wfs.option.ChunkSizeLimit)
 	entry := fh.f.getEntry()
 	if entry != nil {
 		entry.Attributes.FileSize = filer.FileSize(entry)
@@ -63,9 +60,6 @@ var _ = fs.HandleReleaser(&FileHandle{})

 func (fh *FileHandle) Read(ctx context.Context, req *fuse.ReadRequest, resp *fuse.ReadResponse) error {

-	fh.Add(1)
-	defer fh.Done()
-
 	fh.Lock()
 	defer fh.Unlock()

@@ -81,6 +75,8 @@ func (fh *FileHandle) Read(ctx context.Context, req *fuse.ReadRequest, resp *fus
 		buff = make([]byte, req.Size)
 	}

+	fh.lockForRead(req.Offset, len(buff))
+	defer fh.unlockForRead(req.Offset, len(buff))
 	totalRead, err := fh.readFromChunks(buff, req.Offset)
 	if err == nil || err == io.EOF {
 		maxStop := fh.readFromDirtyPages(buff, req.Offset)
@@ -107,6 +103,13 @@ func (fh *FileHandle) Read(ctx context.Context, req *fuse.ReadRequest, resp *fus
 	return err
 }

+func (fh *FileHandle) lockForRead(startOffset int64, size int) {
+	fh.dirtyPages.LockForRead(startOffset, startOffset+int64(size))
+}
+func (fh *FileHandle) unlockForRead(startOffset int64, size int) {
+	fh.dirtyPages.UnlockForRead(startOffset, startOffset+int64(size))
+}
+
 func (fh *FileHandle) readFromDirtyPages(buff []byte, startOffset int64) (maxStop int64) {
 	maxStop = fh.dirtyPages.ReadDirtyDataAt(buff, startOffset)
 	return
@@ -155,6 +158,10 @@ func (fh *FileHandle) readFromChunks(buff []byte, offset int64) (int64, error) {
 	reader := fh.reader
 	if reader == nil {
 		chunkViews := filer.ViewFromVisibleIntervals(fh.entryViewCache, 0, math.MaxInt64)
+		glog.V(4).Infof("file handle read %s [%d,%d) from %d views", fileFullPath, offset, offset+int64(len(buff)), len(chunkViews))
+		for _, chunkView := range chunkViews {
+			glog.V(4).Infof("  read %s [%d,%d) from chunk %+v", fileFullPath, chunkView.LogicOffset, chunkView.LogicOffset+int64(chunkView.Size), chunkView.FileId)
+		}
 		reader = filer.NewChunkReaderAtFromClient(fh.f.wfs.LookupFn(), chunkViews, fh.f.wfs.chunkCache, fileSize)
 	}
 	fh.reader = reader
@@ -165,7 +172,7 @@ func (fh *FileHandle) readFromChunks(buff []byte, offset int64) (int64, error) {
 		glog.Errorf("file handle read %s: %v", fileFullPath, err)
 	}

-	// glog.V(4).Infof("file handle read %s [%d,%d] %d : %v", fileFullPath, offset, offset+int64(totalRead), totalRead, err)
+	glog.V(4).Infof("file handle read %s [%d,%d] %d : %v", fileFullPath, offset, offset+int64(totalRead), totalRead, err)

 	return int64(totalRead), err
 }
@@ -173,9 +180,6 @@ func (fh *FileHandle) readFromChunks(buff []byte, offset int64) (int64, error) {
 // Write to the file handle
 func (fh *FileHandle) Write(ctx context.Context, req *fuse.WriteRequest, resp *fuse.WriteResponse) error {

-	fh.Add(1)
-	defer fh.Done()
-
 	fh.dirtyPages.writerPattern.MonitorWriteAt(req.Offset, len(req.Data))

 	fh.Lock()
@@ -217,8 +221,6 @@ func (fh *FileHandle) Release(ctx context.Context, req *fuse.ReleaseRequest) err

 	glog.V(4).Infof("Release %v fh %d open=%d", fh.f.fullpath(), fh.handle, fh.f.isOpen)

-	fh.Wait()
-
 	fh.f.wfs.handlesLock.Lock()
 	fh.f.isOpen--
 	fh.f.wfs.handlesLock.Unlock()
@@ -250,9 +252,6 @@ func (fh *FileHandle) Flush(ctx context.Context, req *fuse.FlushRequest) error {
 		return nil
 	}

-	fh.Add(1)
-	defer fh.Done()
-
 	fh.Lock()
 	defer fh.Unlock()

@@ -5,6 +5,7 @@ import (
 	"github.com/chrislusf/seaweedfs/weed/filer"
 	"github.com/chrislusf/seaweedfs/weed/filer/leveldb"
 	"github.com/chrislusf/seaweedfs/weed/glog"
+	"github.com/chrislusf/seaweedfs/weed/pb/filer_pb"
 	"github.com/chrislusf/seaweedfs/weed/util"
 	"github.com/chrislusf/seaweedfs/weed/util/bounded_tree"
 	"os"
@@ -18,16 +19,16 @@ type MetaCache struct {
 	// sync.RWMutex
 	visitedBoundary *bounded_tree.BoundedTree
 	uidGidMapper    *UidGidMapper
-	invalidateFunc  func(util.FullPath)
+	invalidateFunc  func(fullpath util.FullPath, entry *filer_pb.Entry)
 }

-func NewMetaCache(dbFolder string, baseDir util.FullPath, uidGidMapper *UidGidMapper, invalidateFunc func(util.FullPath)) *MetaCache {
+func NewMetaCache(dbFolder string, baseDir util.FullPath, uidGidMapper *UidGidMapper, invalidateFunc func(util.FullPath, *filer_pb.Entry)) *MetaCache {
 	return &MetaCache{
 		localStore:      openMetaStore(dbFolder),
 		visitedBoundary: bounded_tree.NewBoundedTree(baseDir),
 		uidGidMapper:    uidGidMapper,
-		invalidateFunc: func(fullpath util.FullPath) {
-			invalidateFunc(fullpath)
+		invalidateFunc: func(fullpath util.FullPath, entry *filer_pb.Entry) {
+			invalidateFunc(fullpath, entry)
 		},
 	}
 }
@@ -144,3 +145,10 @@ func (mc *MetaCache) Shutdown() {
 func (mc *MetaCache) mapIdFromFilerToLocal(entry *filer.Entry) {
 	entry.Attr.Uid, entry.Attr.Gid = mc.uidGidMapper.FilerToLocal(entry.Attr.Uid, entry.Attr.Gid)
 }
+
+func (mc *MetaCache) Debug() {
+	if debuggable, ok := mc.localStore.(filer.Debuggable); ok {
+		println("start debugging")
+		debuggable.Debug(os.Stderr)
+	}
+}
@@ -40,16 +40,16 @@ func SubscribeMetaEvents(mc *MetaCache, selfSignature int32, client filer_pb.Fil
 		if err == nil {
 			if message.OldEntry != nil && message.NewEntry != nil {
 				oldKey := util.NewFullPath(resp.Directory, message.OldEntry.Name)
-				mc.invalidateFunc(oldKey)
+				mc.invalidateFunc(oldKey, message.OldEntry)
 				if message.OldEntry.Name != message.NewEntry.Name {
 					newKey := util.NewFullPath(dir, message.NewEntry.Name)
-					mc.invalidateFunc(newKey)
+					mc.invalidateFunc(newKey, message.NewEntry)
 				}
 			} else if message.OldEntry == nil && message.NewEntry != nil {
 				// no need to invaalidate
 			} else if message.OldEntry != nil && message.NewEntry == nil {
 				oldKey := util.NewFullPath(resp.Directory, message.OldEntry.Name)
-				mc.invalidateFunc(oldKey)
+				mc.invalidateFunc(oldKey, message.OldEntry)
 			}
 		}

@@ -6,36 +6,33 @@ import (
 )

 type PageWriter struct {
-	f             *File
+	fh            *FileHandle
 	collection    string
 	replication   string
 	chunkSize     int64
 	writerPattern *WriterPattern

 	randomWriter page_writer.DirtyPages
-	streamWriter page_writer.DirtyPages
 }

 var (
 	_ = page_writer.DirtyPages(&PageWriter{})
 )

-func newPageWriter(file *File, chunkSize int64) *PageWriter {
+func newPageWriter(fh *FileHandle, chunkSize int64) *PageWriter {
 	pw := &PageWriter{
-		f:             file,
+		fh:            fh,
 		chunkSize:     chunkSize,
 		writerPattern: NewWriterPattern(chunkSize),
-		randomWriter:  newTempFileDirtyPages(file, chunkSize),
-		streamWriter:  newStreamDirtyPages(file, chunkSize),
-		//streamWriter: newContinuousDirtyPages(file),
-		//streamWriter:  nil,
+		randomWriter:  newMemoryChunkPages(fh, chunkSize),
+		// randomWriter: newTempFileDirtyPages(fh.f, chunkSize),
 	}
 	return pw
 }

 func (pw *PageWriter) AddPage(offset int64, data []byte) {

-	glog.V(4).Infof("%v AddPage [%d, %d) streaming:%v", pw.f.fullpath(), offset, offset+int64(len(data)), pw.writerPattern.IsStreamingMode())
+	glog.V(4).Infof("%v AddPage [%d, %d) streaming:%v", pw.fh.f.fullpath(), offset, offset+int64(len(data)), pw.writerPattern.IsStreamingMode())

 	chunkIndex := offset / pw.chunkSize
 	for i := chunkIndex; len(data) > 0; i++ {
@@ -47,38 +44,22 @@ func (pw *PageWriter) AddPage(offset int64, data []byte) {
 }

 func (pw *PageWriter) addToOneChunk(chunkIndex, offset int64, data []byte) {
-	if chunkIndex > 0 {
-		if pw.writerPattern.IsStreamingMode() && pw.streamWriter != nil {
-			pw.streamWriter.AddPage(offset, data)
-			return
-		}
-	}
 	pw.randomWriter.AddPage(offset, data)
 }

 func (pw *PageWriter) FlushData() error {
 	pw.writerPattern.Reset()
-	if pw.streamWriter != nil {
-		if err := pw.streamWriter.FlushData(); err != nil {
-			return err
-		}
-	}
 	return pw.randomWriter.FlushData()
 }

 func (pw *PageWriter) ReadDirtyDataAt(data []byte, offset int64) (maxStop int64) {
-	glog.V(4).Infof("ReadDirtyDataAt %v [%d, %d)", pw.f.fullpath(), offset, offset+int64(len(data)))
+	glog.V(4).Infof("ReadDirtyDataAt %v [%d, %d)", pw.fh.f.fullpath(), offset, offset+int64(len(data)))

 	chunkIndex := offset / pw.chunkSize
 	for i := chunkIndex; len(data) > 0; i++ {
 		readSize := min(int64(len(data)), (i+1)*pw.chunkSize-offset)

-		if pw.streamWriter != nil {
-			m1 := pw.streamWriter.ReadDirtyDataAt(data[:readSize], offset)
-			maxStop = max(maxStop, m1)
-		}
-		m2 := pw.randomWriter.ReadDirtyDataAt(data[:readSize], offset)
-		maxStop = max(maxStop, m2)
+		maxStop = pw.randomWriter.ReadDirtyDataAt(data[:readSize], offset)

 		offset += readSize
 		data = data[readSize:]
@@ -88,16 +69,18 @@ func (pw *PageWriter) ReadDirtyDataAt(data []byte, offset int64) (maxStop int64)
 }

 func (pw *PageWriter) GetStorageOptions() (collection, replication string) {
-	if pw.writerPattern.IsStreamingMode() && pw.streamWriter != nil {
-		return pw.streamWriter.GetStorageOptions()
-	}
 	return pw.randomWriter.GetStorageOptions()
 }

+func (pw *PageWriter) LockForRead(startOffset, stopOffset int64) {
+	pw.randomWriter.LockForRead(startOffset, stopOffset)
+}
+
+func (pw *PageWriter) UnlockForRead(startOffset, stopOffset int64) {
+	pw.randomWriter.UnlockForRead(startOffset, stopOffset)
+}
+
 func (pw *PageWriter) Destroy() {
-	if pw.streamWriter != nil {
-		pw.streamWriter.Destroy()
-	}
 	pw.randomWriter.Destroy()
 }

@@ -51,6 +51,12 @@ func (list *ChunkWrittenIntervalList) MarkWritten(startOffset, stopOffset int64)
 func (list *ChunkWrittenIntervalList) IsComplete(chunkSize int64) bool {
 	return list.size() == 1 && list.head.next.isComplete(chunkSize)
 }
+func (list *ChunkWrittenIntervalList) WrittenSize() (writtenByteCount int64) {
+	for t := list.head; t != nil; t = t.next {
+		writtenByteCount += t.Size()
+	}
+	return
+}

 func (list *ChunkWrittenIntervalList) addInterval(interval *ChunkWrittenInterval) {

@@ -1,160 +0,0 @@
-package page_writer
-
-import (
-	"github.com/chrislusf/seaweedfs/weed/glog"
-	"io"
-	"os"
-	"sync"
-)
-
-type LogicChunkIndex int
-type ActualChunkIndex int
-
-// ChunkedFileWriter assumes the write requests will come in within chunks
-type ChunkedFileWriter struct {
-	dir                     string
-	file                    *os.File
-	logicToActualChunkIndex map[LogicChunkIndex]ActualChunkIndex
-	chunkUsages             []*ChunkWrittenIntervalList
-	ChunkSize               int64
-	sync.Mutex
-}
-
-var _ = io.WriterAt(&ChunkedFileWriter{})
-
-func NewChunkedFileWriter(dir string, chunkSize int64) *ChunkedFileWriter {
-	return &ChunkedFileWriter{
-		dir:                     dir,
-		file:                    nil,
-		logicToActualChunkIndex: make(map[LogicChunkIndex]ActualChunkIndex),
-		ChunkSize:               chunkSize,
-	}
-}
-
-func (cw *ChunkedFileWriter) WriteAt(p []byte, off int64) (n int, err error) {
-	cw.Lock()
-	defer cw.Unlock()
-
-	if cw.file == nil {
-		cw.file, err = os.CreateTemp(cw.dir, "")
-		if err != nil {
-			glog.Errorf("create temp file: %v", err)
-			return
-		}
-	}
-
-	actualOffset, chunkUsage := cw.toActualWriteOffset(off)
-	n, err = cw.file.WriteAt(p, actualOffset)
-	if err == nil {
-		startOffset := off % cw.ChunkSize
-		chunkUsage.MarkWritten(startOffset, startOffset+int64(n))
-	}
-	return
-}
-
-func (cw *ChunkedFileWriter) ReadDataAt(p []byte, off int64) (maxStop int64) {
-	cw.Lock()
-	defer cw.Unlock()
-
-	if cw.file == nil {
-		return
-	}
-
-	logicChunkIndex := off / cw.ChunkSize
-	actualChunkIndex, chunkUsage := cw.toActualReadOffset(off)
-	if chunkUsage != nil {
-		for t := chunkUsage.head.next; t != chunkUsage.tail; t = t.next {
-			logicStart := max(off, logicChunkIndex*cw.ChunkSize+t.StartOffset)
-			logicStop := min(off+int64(len(p)), logicChunkIndex*cw.ChunkSize+t.stopOffset)
-			if logicStart < logicStop {
-				actualStart := logicStart - logicChunkIndex*cw.ChunkSize + int64(actualChunkIndex)*cw.ChunkSize
-				_, err := cw.file.ReadAt(p[logicStart-off:logicStop-off], actualStart)
-				if err != nil {
-					glog.Errorf("reading temp file: %v", err)
-					break
-				}
-				maxStop = max(maxStop, logicStop)
-			}
-		}
-	}
-	return
-}
-
-func (cw *ChunkedFileWriter) toActualWriteOffset(logicOffset int64) (actualOffset int64, chunkUsage *ChunkWrittenIntervalList) {
-	logicChunkIndex := LogicChunkIndex(logicOffset / cw.ChunkSize)
-	offsetRemainder := logicOffset % cw.ChunkSize
-	existingActualChunkIndex, found := cw.logicToActualChunkIndex[logicChunkIndex]
-	if found {
-		return int64(existingActualChunkIndex)*cw.ChunkSize + offsetRemainder, cw.chunkUsages[existingActualChunkIndex]
-	}
-	cw.logicToActualChunkIndex[logicChunkIndex] = ActualChunkIndex(len(cw.chunkUsages))
-	chunkUsage = newChunkWrittenIntervalList()
-	cw.chunkUsages = append(cw.chunkUsages, chunkUsage)
-	return int64(len(cw.chunkUsages)-1)*cw.ChunkSize + offsetRemainder, chunkUsage
-}
-
-func (cw *ChunkedFileWriter) toActualReadOffset(logicOffset int64) (actualChunkIndex ActualChunkIndex, chunkUsage *ChunkWrittenIntervalList) {
-	logicChunkIndex := LogicChunkIndex(logicOffset / cw.ChunkSize)
-	existingActualChunkIndex, found := cw.logicToActualChunkIndex[logicChunkIndex]
-	if found {
-		return existingActualChunkIndex, cw.chunkUsages[existingActualChunkIndex]
-	}
-	return 0, nil
-}
-
-func (cw *ChunkedFileWriter) ProcessEachInterval(process func(file *os.File, logicChunkIndex LogicChunkIndex, interval *ChunkWrittenInterval)) {
-	for logicChunkIndex, actualChunkIndex := range cw.logicToActualChunkIndex {
-		chunkUsage := cw.chunkUsages[actualChunkIndex]
-		for t := chunkUsage.head.next; t != chunkUsage.tail; t = t.next {
-			process(cw.file, logicChunkIndex, t)
-		}
-	}
-}
-
-// Reset releases used resources
-func (cw *ChunkedFileWriter) Reset() {
-	if cw.file != nil {
-		cw.file.Close()
-		os.Remove(cw.file.Name())
-		cw.file = nil
-	}
-	cw.logicToActualChunkIndex = make(map[LogicChunkIndex]ActualChunkIndex)
-	cw.chunkUsages = cw.chunkUsages[:0]
-}
-
-type FileIntervalReader struct {
-	f           *os.File
-	startOffset int64
-	stopOffset  int64
-	position    int64
-}
-
-var _ = io.Reader(&FileIntervalReader{})
-
-func NewFileIntervalReader(cw *ChunkedFileWriter, logicChunkIndex LogicChunkIndex, interval *ChunkWrittenInterval) *FileIntervalReader {
-	actualChunkIndex, found := cw.logicToActualChunkIndex[logicChunkIndex]
-	if !found {
-		// this should never happen
-		return nil
-	}
-	return &FileIntervalReader{
-		f:           cw.file,
-		startOffset: int64(actualChunkIndex)*cw.ChunkSize + interval.StartOffset,
-		stopOffset:  int64(actualChunkIndex)*cw.ChunkSize + interval.stopOffset,
-		position:    0,
-	}
-}
-
-func (fr *FileIntervalReader) Read(p []byte) (n int, err error) {
-	readSize := minInt(len(p), int(fr.stopOffset-fr.startOffset-fr.position))
-	n, err = fr.f.ReadAt(p[:readSize], fr.startOffset+fr.position)
-	if err == nil || err == io.EOF {
-		fr.position += int64(n)
-		if fr.stopOffset-fr.startOffset-fr.position == 0 {
-			// return a tiny bit faster
-			err = io.EOF
-			return
-		}
-	}
-	return
-}
@@ -1,60 +0,0 @@
-package page_writer
-
-import (
-	"github.com/stretchr/testify/assert"
-	"os"
-	"testing"
-)
-
-func TestChunkedFileWriter_toActualOffset(t *testing.T) {
-	cw := NewChunkedFileWriter("", 16)
-
-	writeToFile(cw, 50, 60)
-	writeToFile(cw, 60, 64)
-
-	writeToFile(cw, 32, 40)
-	writeToFile(cw, 42, 48)
-
-	writeToFile(cw, 48, 50)
-
-	assert.Equal(t, 1, cw.chunkUsages[0].size(), "fully covered")
-	assert.Equal(t, 2, cw.chunkUsages[1].size(), "2 intervals")
-
-}
-
-func writeToFile(cw *ChunkedFileWriter, startOffset int64, stopOffset int64) {
-
-	_, chunkUsage := cw.toActualWriteOffset(startOffset)
-
-	// skip doing actual writing
-
-	innerOffset := startOffset % cw.ChunkSize
-	chunkUsage.MarkWritten(innerOffset, innerOffset+stopOffset-startOffset)
-
-}
-
-func TestWriteChunkedFile(t *testing.T) {
-	x := NewChunkedFileWriter(os.TempDir(), 20)
-	defer x.Reset()
-	y := NewChunkedFileWriter(os.TempDir(), 12)
-	defer y.Reset()
-
-	batchSize := 4
-	buf := make([]byte, batchSize)
-	for i := 0; i < 256; i++ {
-		for x := 0; x < batchSize; x++ {
-			buf[x] = byte(i)
-		}
-		x.WriteAt(buf, int64(i*batchSize))
-		y.WriteAt(buf, int64((255-i)*batchSize))
-	}
-
-	a := make([]byte, 1)
-	b := make([]byte, 1)
-	for i := 0; i < 256*batchSize; i++ {
-		x.ReadDataAt(a, int64(i))
-		y.ReadDataAt(b, int64(256*batchSize-1-i))
-		assert.Equal(t, a[0], b[0], "same read")
-	}
-
-}
@@ -1,119 +0,0 @@
-package page_writer
-
-import (
-	"github.com/chrislusf/seaweedfs/weed/util"
-	"github.com/chrislusf/seaweedfs/weed/util/mem"
-	"io"
-	"sync"
-	"sync/atomic"
-)
-
-type SaveToStorageFunc func(reader io.Reader, offset int64, size int64, cleanupFn func())
-
-// ChunkedStreamWriter assumes the write requests will come in within chunks and in streaming mode
-type ChunkedStreamWriter struct {
-	activeChunks     map[LogicChunkIndex]*MemChunk
-	activeChunksLock sync.Mutex
-	ChunkSize        int64
-	saveToStorageFn  SaveToStorageFunc
-	sync.Mutex
-}
-
-type MemChunk struct {
-	buf   []byte
-	usage *ChunkWrittenIntervalList
-}
-
-var _ = io.WriterAt(&ChunkedStreamWriter{})
-
-func NewChunkedStreamWriter(chunkSize int64) *ChunkedStreamWriter {
-	return &ChunkedStreamWriter{
-		ChunkSize:    chunkSize,
-		activeChunks: make(map[LogicChunkIndex]*MemChunk),
-	}
-}
-
-func (cw *ChunkedStreamWriter) SetSaveToStorageFunction(saveToStorageFn SaveToStorageFunc) {
-	cw.saveToStorageFn = saveToStorageFn
-}
-
-func (cw *ChunkedStreamWriter) WriteAt(p []byte, off int64) (n int, err error) {
-	cw.Lock()
-	defer cw.Unlock()
-
-	logicChunkIndex := LogicChunkIndex(off / cw.ChunkSize)
-	offsetRemainder := off % cw.ChunkSize
-
-	memChunk, found := cw.activeChunks[logicChunkIndex]
-	if !found {
-		memChunk = &MemChunk{
-			buf:   mem.Allocate(int(cw.ChunkSize)),
-			usage: newChunkWrittenIntervalList(),
-		}
-		cw.activeChunks[logicChunkIndex] = memChunk
-	}
-	n = copy(memChunk.buf[offsetRemainder:], p)
-	memChunk.usage.MarkWritten(offsetRemainder, offsetRemainder+int64(n))
-	if memChunk.usage.IsComplete(cw.ChunkSize) {
-		if cw.saveToStorageFn != nil {
-			cw.saveOneChunk(memChunk, logicChunkIndex)
-			delete(cw.activeChunks, logicChunkIndex)
-		}
-	}
-
-	return
-}
-
-func (cw *ChunkedStreamWriter) ReadDataAt(p []byte, off int64) (maxStop int64) {
-	cw.Lock()
-	defer cw.Unlock()
-
-	logicChunkIndex := LogicChunkIndex(off / cw.ChunkSize)
-	memChunkBaseOffset := int64(logicChunkIndex) * cw.ChunkSize
-	memChunk, found := cw.activeChunks[logicChunkIndex]
-	if !found {
-		return
-	}
-
-	for t := memChunk.usage.head.next; t != memChunk.usage.tail; t = t.next {
-		logicStart := max(off, int64(logicChunkIndex)*cw.ChunkSize+t.StartOffset)
-		logicStop := min(off+int64(len(p)), memChunkBaseOffset+t.stopOffset)
-		if logicStart < logicStop {
-			copy(p[logicStart-off:logicStop-off], memChunk.buf[logicStart-memChunkBaseOffset:logicStop-memChunkBaseOffset])
-			maxStop = max(maxStop, logicStop)
-		}
-	}
-	return
-}
-
-func (cw *ChunkedStreamWriter) FlushAll() {
-	cw.Lock()
-	defer cw.Unlock()
-	for logicChunkIndex, memChunk := range cw.activeChunks {
-		if cw.saveToStorageFn != nil {
-			cw.saveOneChunk(memChunk, logicChunkIndex)
-			delete(cw.activeChunks, logicChunkIndex)
-		}
-	}
-}
-
-func (cw *ChunkedStreamWriter) saveOneChunk(memChunk *MemChunk, logicChunkIndex LogicChunkIndex) {
-	var referenceCounter = int32(memChunk.usage.size())
-	for t := memChunk.usage.head.next; t != memChunk.usage.tail; t = t.next {
-		reader := util.NewBytesReader(memChunk.buf[t.StartOffset:t.stopOffset])
-		cw.saveToStorageFn(reader, int64(logicChunkIndex)*cw.ChunkSize+t.StartOffset, t.Size(), func() {
-			atomic.AddInt32(&referenceCounter, -1)
-			if atomic.LoadInt32(&referenceCounter) == 0 {
-				mem.Free(memChunk.buf)
-			}
-		})
-	}
-}
-
-// Reset releases used resources
-func (cw *ChunkedStreamWriter) Reset() {
-	for t, memChunk := range cw.activeChunks {
-		mem.Free(memChunk.buf)
-		delete(cw.activeChunks, t)
-	}
-}
@@ -1,33 +0,0 @@
-package page_writer
-
-import (
-	"github.com/stretchr/testify/assert"
-	"os"
-	"testing"
-)
-
-func TestWriteChunkedStream(t *testing.T) {
-	x := NewChunkedStreamWriter(20)
-	defer x.Reset()
-	y := NewChunkedFileWriter(os.TempDir(), 12)
-	defer y.Reset()
-
-	batchSize := 4
-	buf := make([]byte, batchSize)
-	for i := 0; i < 256; i++ {
-		for x := 0; x < batchSize; x++ {
-			buf[x] = byte(i)
-		}
-		x.WriteAt(buf, int64(i*batchSize))
-		y.WriteAt(buf, int64((255-i)*batchSize))
-	}
-
-	a := make([]byte, 1)
-	b := make([]byte, 1)
-	for i := 0; i < 256*batchSize; i++ {
-		x.ReadDataAt(a, int64(i))
-		y.ReadDataAt(b, int64(256*batchSize-1-i))
-		assert.Equal(t, a[0], b[0], "same read")
-	}
-
-}
@@ -1,222 +0,0 @@
-package page_writer
-
-import (
-	"io"
-
-	"github.com/chrislusf/seaweedfs/weed/util"
-)
-
-type IntervalNode struct {
-	Data   []byte
-	Offset int64
-	Size   int64
-	Next   *IntervalNode
-}
-
-type IntervalLinkedList struct {
-	Head *IntervalNode
-	Tail *IntervalNode
-}
-
-type ContinuousIntervals struct {
-	lists []*IntervalLinkedList
-}
-
-func (list *IntervalLinkedList) Offset() int64 {
-	return list.Head.Offset
-}
-func (list *IntervalLinkedList) Size() int64 {
-	return list.Tail.Offset + list.Tail.Size - list.Head.Offset
-}
-func (list *IntervalLinkedList) addNodeToTail(node *IntervalNode) {
-	// glog.V(4).Infof("add to tail [%d,%d) + [%d,%d) => [%d,%d)", list.Head.Offset, list.Tail.Offset+list.Tail.Size, node.Offset, node.Offset+node.Size, list.Head.Offset, node.Offset+node.Size)
-	list.Tail.Next = node
-	list.Tail = node
-}
-func (list *IntervalLinkedList) addNodeToHead(node *IntervalNode) {
-	// glog.V(4).Infof("add to head [%d,%d) + [%d,%d) => [%d,%d)", node.Offset, node.Offset+node.Size, list.Head.Offset, list.Tail.Offset+list.Tail.Size, node.Offset, list.Tail.Offset+list.Tail.Size)
-	node.Next = list.Head
-	list.Head = node
-}
-
-func (list *IntervalLinkedList) ReadData(buf []byte, start, stop int64) {
-	t := list.Head
-	for {
-
-		nodeStart, nodeStop := max(start, t.Offset), min(stop, t.Offset+t.Size)
-		if nodeStart < nodeStop {
-			// glog.V(0).Infof("copying start=%d stop=%d t=[%d,%d) t.data=%d => bufSize=%d nodeStart=%d, nodeStop=%d", start, stop, t.Offset, t.Offset+t.Size, len(t.Data), len(buf), nodeStart, nodeStop)
-			copy(buf[nodeStart-start:], t.Data[nodeStart-t.Offset:nodeStop-t.Offset])
-		}
-
-		if t.Next == nil {
-			break
-		}
-		t = t.Next
-	}
-}
-
-func (c *ContinuousIntervals) TotalSize() (total int64) {
-	for _, list := range c.lists {
-		total += list.Size()
-	}
-	return
-}
-
-func subList(list *IntervalLinkedList, start, stop int64) *IntervalLinkedList {
-	var nodes []*IntervalNode
-	for t := list.Head; t != nil; t = t.Next {
-		nodeStart, nodeStop := max(start, t.Offset), min(stop, t.Offset+t.Size)
-		if nodeStart >= nodeStop {
-			// skip non overlapping IntervalNode
-			continue
-		}
-		nodes = append(nodes, &IntervalNode{
-			Data:   t.Data[nodeStart-t.Offset : nodeStop-t.Offset],
-			Offset: nodeStart,
-			Size:   nodeStop - nodeStart,
-			Next:   nil,
-		})
-	}
-	for i := 1; i < len(nodes); i++ {
-		nodes[i-1].Next = nodes[i]
-	}
-	return &IntervalLinkedList{
-		Head: nodes[0],
-		Tail: nodes[len(nodes)-1],
-	}
-}
-
-func (c *ContinuousIntervals) AddInterval(data []byte, offset int64) {
-
-	interval := &IntervalNode{Data: data, Offset: offset, Size: int64(len(data))}
-
-	// append to the tail and return
-	if len(c.lists) == 1 {
-		lastSpan := c.lists[0]
-		if lastSpan.Tail.Offset+lastSpan.Tail.Size == offset {
-			lastSpan.addNodeToTail(interval)
-			return
-		}
-	}
-
-	var newLists []*IntervalLinkedList
-	for _, list := range c.lists {
-		// if list is to the left of new interval, add to the new list
-		if list.Tail.Offset+list.Tail.Size <= interval.Offset {
-			newLists = append(newLists, list)
-		}
-		// if list is to the right of new interval, add to the new list
-		if interval.Offset+interval.Size <= list.Head.Offset {
-			newLists = append(newLists, list)
-		}
-		// if new interval overwrite the right part of the list
-		if list.Head.Offset < interval.Offset && interval.Offset < list.Tail.Offset+list.Tail.Size {
-			// create a new list of the left part of existing list
-			newLists = append(newLists, subList(list, list.Offset(), interval.Offset))
-		}
-		// if new interval overwrite the left part of the list
-		if list.Head.Offset < interval.Offset+interval.Size && interval.Offset+interval.Size < list.Tail.Offset+list.Tail.Size {
-			// create a new list of the right part of existing list
-			newLists = append(newLists, subList(list, interval.Offset+interval.Size, list.Tail.Offset+list.Tail.Size))
-		}
-		// skip anything that is fully overwritten by the new interval
-	}
-
-	c.lists = newLists
-	// add the new interval to the lists, connecting neighbor lists
-	var prevList, nextList *IntervalLinkedList
-
-	for _, list := range c.lists {
-		if list.Head.Offset == interval.Offset+interval.Size {
-			nextList = list
-			break
-		}
-	}
-
-	for _, list := range c.lists {
-		if list.Head.Offset+list.Size() == offset {
-			list.addNodeToTail(interval)
-			prevList = list
-			break
-		}
-	}
-
-	if prevList != nil && nextList != nil {
-		// glog.V(4).Infof("connecting [%d,%d) + [%d,%d) => [%d,%d)", prevList.Head.Offset, prevList.Tail.Offset+prevList.Tail.Size, nextList.Head.Offset, nextList.Tail.Offset+nextList.Tail.Size, prevList.Head.Offset, nextList.Tail.Offset+nextList.Tail.Size)
-		prevList.Tail.Next = nextList.Head
-		prevList.Tail = nextList.Tail
-		c.removeList(nextList)
-	} else if nextList != nil {
-		// add to head was not done when checking
-		nextList.addNodeToHead(interval)
-	}
-	if prevList == nil && nextList == nil {
-		c.lists = append(c.lists, &IntervalLinkedList{
-			Head: interval,
-			Tail: interval,
-		})
-	}
-
-	return
-}
-
-func (c *ContinuousIntervals) RemoveLargestIntervalLinkedList() *IntervalLinkedList {
-	var maxSize int64
-	maxIndex := -1
-	for k, list := range c.lists {
-		if maxSize <= list.Size() {
-			maxSize = list.Size()
-			maxIndex = k
-		}
-	}
-	if maxSize <= 0 {
-		return nil
-	}
-
-	t := c.lists[maxIndex]
-	c.lists = append(c.lists[0:maxIndex], c.lists[maxIndex+1:]...)
-	return t
-
-}
-
-func (c *ContinuousIntervals) removeList(target *IntervalLinkedList) {
-	index := -1
-	for k, list := range c.lists {
-		if list.Offset() == target.Offset() {
-			index = k
-		}
-	}
-	if index < 0 {
-		return
-	}
-
-	c.lists = append(c.lists[0:index], c.lists[index+1:]...)
-
-}
-
-func (c *ContinuousIntervals) ReadDataAt(data []byte, startOffset int64) (maxStop int64) {
-	for _, list := range c.lists {
-		start := max(startOffset, list.Offset())
-		stop := min(startOffset+int64(len(data)), list.Offset()+list.Size())
-		if start < stop {
-			list.ReadData(data[start-startOffset:], start, stop)
-			maxStop = max(maxStop, stop)
-		}
-	}
-	return
-}
-
-func (l *IntervalLinkedList) ToReader() io.Reader {
-	var readers []io.Reader
-	t := l.Head
-	readers = append(readers, util.NewBytesReader(t.Data))
-	for t.Next != nil {
-		t = t.Next
-		readers = append(readers, util.NewBytesReader(t.Data))
-	}
-	if len(readers) == 1 {
-		return readers[0]
-	}
-	return io.MultiReader(readers...)
-}
@@ -1,113 +0,0 @@
-package page_writer
-
-import (
-	"bytes"
-	"math/rand"
-	"testing"
-)
-
-func TestContinuousIntervals_AddIntervalAppend(t *testing.T) {
-
-	c := &ContinuousIntervals{}
-
-	// 25, 25, 25
-	c.AddInterval(getBytes(25, 3), 0)
-	//  _,  _, 23, 23, 23, 23
-	c.AddInterval(getBytes(23, 4), 2)
-
-	expectedData(t, c, 0, 25, 25, 23, 23, 23, 23)
-
-}
-
-func TestContinuousIntervals_AddIntervalInnerOverwrite(t *testing.T) {
-
-	c := &ContinuousIntervals{}
-
-	// 25, 25, 25, 25, 25
-	c.AddInterval(getBytes(25, 5), 0)
-	//  _,  _, 23, 23
-	c.AddInterval(getBytes(23, 2), 2)
-
-	expectedData(t, c, 0, 25, 25, 23, 23, 25)
-
-}
-
-func TestContinuousIntervals_AddIntervalFullOverwrite(t *testing.T) {
-
-	c := &ContinuousIntervals{}
-
-	// 1,
-	c.AddInterval(getBytes(1, 1), 0)
-	//  _,  2,
-	c.AddInterval(getBytes(2, 1), 1)
-	//  _,  _,  3,  3, 3
-	c.AddInterval(getBytes(3, 3), 2)
-	//  _,  _,  _,  4, 4, 4
-	c.AddInterval(getBytes(4, 3), 3)
-
-	expectedData(t, c, 0, 1, 2, 3, 4, 4, 4)
-
-}
-
-func TestContinuousIntervals_RealCase1(t *testing.T) {
-
-	c := &ContinuousIntervals{}
-
-	// 25,
-	c.AddInterval(getBytes(25, 1), 0)
-	//  _,  _,  _,  _, 23, 23
-	c.AddInterval(getBytes(23, 2), 4)
-	//  _,  _,  _, 24, 24, 24, 24
-	c.AddInterval(getBytes(24, 4), 3)
-
-	//  _,  22, 22
-	c.AddInterval(getBytes(22, 2), 1)
-
-	expectedData(t, c, 0, 25, 22, 22, 24, 24, 24, 24)
-
-}
-
-func TestRandomWrites(t *testing.T) {
-
-	c := &ContinuousIntervals{}
-
-	data := make([]byte, 1024)
-
-	for i := 0; i < 1024; i++ {
-
-		start, stop := rand.Intn(len(data)), rand.Intn(len(data))
-		if start > stop {
-			start, stop = stop, start
-		}
-
-		rand.Read(data[start : stop+1])
-
-		c.AddInterval(data[start:stop+1], int64(start))
-
-		expectedData(t, c, 0, data...)
-
-	}
-
-}
-
-func expectedData(t *testing.T, c *ContinuousIntervals, offset int, data ...byte) {
-	start, stop := int64(offset), int64(offset+len(data))
-	for _, list := range c.lists {
-		nodeStart, nodeStop := max(start, list.Head.Offset), min(stop, list.Head.Offset+list.Size())
-		if nodeStart < nodeStop {
-			buf := make([]byte, nodeStop-nodeStart)
-			list.ReadData(buf, nodeStart, nodeStop)
-			if bytes.Compare(buf, data[nodeStart-start:nodeStop-start]) != 0 {
-				t.Errorf("expected %v actual %v", data[nodeStart-start:nodeStop-start], buf)
-			}
-		}
-	}
-}
-
-func getBytes(content byte, length int) []byte {
-	data := make([]byte, length)
-	for i := 0; i < length; i++ {
-		data[i] = content
-	}
-	return data
-}
@@ -6,6 +6,8 @@ type DirtyPages interface {
 	ReadDirtyDataAt(data []byte, startOffset int64) (maxStop int64)
 	GetStorageOptions() (collection, replication string)
 	Destroy()
+	LockForRead(startOffset, stopOffset int64)
+	UnlockForRead(startOffset, stopOffset int64)
 }

 func max(x, y int64) int64 {
@@ -0,0 +1,16 @@
+package page_writer
+
+import (
+	"io"
+)
+
+type SaveToStorageFunc func(reader io.Reader, offset int64, size int64, cleanupFn func())
+
+type PageChunk interface {
+	FreeResource()
+	WriteDataAt(src []byte, offset int64) (n int)
+	ReadDataAt(p []byte, off int64) (maxStop int64)
+	IsComplete() bool
+	WrittenSize() int64
+	SaveContent(saveFn SaveToStorageFunc)
+}
@@ -0,0 +1,69 @@
+package page_writer
+
+import (
+	"github.com/chrislusf/seaweedfs/weed/util"
+	"github.com/chrislusf/seaweedfs/weed/util/mem"
+)
+
+var (
+	_ = PageChunk(&MemChunk{})
+)
+
+type MemChunk struct {
+	buf             []byte
+	usage           *ChunkWrittenIntervalList
+	chunkSize       int64
+	logicChunkIndex LogicChunkIndex
+}
+
+func NewMemChunk(logicChunkIndex LogicChunkIndex, chunkSize int64) *MemChunk {
+	return &MemChunk{
+		logicChunkIndex: logicChunkIndex,
+		chunkSize:       chunkSize,
+		buf:             mem.Allocate(int(chunkSize)),
+		usage:           newChunkWrittenIntervalList(),
+	}
+}
+
+func (mc *MemChunk) FreeResource() {
+	mem.Free(mc.buf)
+}
+
+func (mc *MemChunk) WriteDataAt(src []byte, offset int64) (n int) {
+	innerOffset := offset % mc.chunkSize
+	n = copy(mc.buf[innerOffset:], src)
+	mc.usage.MarkWritten(innerOffset, innerOffset+int64(n))
+	return
+}
+
+func (mc *MemChunk) ReadDataAt(p []byte, off int64) (maxStop int64) {
+	memChunkBaseOffset := int64(mc.logicChunkIndex) * mc.chunkSize
+	for t := mc.usage.head.next; t != mc.usage.tail; t = t.next {
+		logicStart := max(off, int64(mc.logicChunkIndex)*mc.chunkSize+t.StartOffset)
+		logicStop := min(off+int64(len(p)), memChunkBaseOffset+t.stopOffset)
+		if logicStart < logicStop {
+			copy(p[logicStart-off:logicStop-off], mc.buf[logicStart-memChunkBaseOffset:logicStop-memChunkBaseOffset])
+			maxStop = max(maxStop, logicStop)
+		}
+	}
+	return
+}
+
+func (mc *MemChunk) IsComplete() bool {
+	return mc.usage.IsComplete(mc.chunkSize)
+}
+
+func (mc *MemChunk) WrittenSize() int64 {
+	return mc.usage.WrittenSize()
+}
+
+func (mc *MemChunk) SaveContent(saveFn SaveToStorageFunc) {
+	if saveFn == nil {
+		return
+	}
+	for t := mc.usage.head.next; t != mc.usage.tail; t = t.next {
+		reader := util.NewBytesReader(mc.buf[t.StartOffset:t.stopOffset])
+		saveFn(reader, int64(mc.logicChunkIndex)*mc.chunkSize+t.StartOffset, t.Size(), func() {
+		})
+	}
+}
@@ -0,0 +1,121 @@
+package page_writer
+
+import (
+	"github.com/chrislusf/seaweedfs/weed/glog"
+	"github.com/chrislusf/seaweedfs/weed/util"
+	"github.com/chrislusf/seaweedfs/weed/util/mem"
+	"os"
+)
+
+var (
+	_ = PageChunk(&SwapFileChunk{})
+)
+
+type ActualChunkIndex int
+
+type SwapFile struct {
+	dir                     string
+	file                    *os.File
+	logicToActualChunkIndex map[LogicChunkIndex]ActualChunkIndex
+	chunkSize               int64
+}
+
+type SwapFileChunk struct {
+	swapfile         *SwapFile
+	usage            *ChunkWrittenIntervalList
+	logicChunkIndex  LogicChunkIndex
+	actualChunkIndex ActualChunkIndex
+}
+
+func NewSwapFile(dir string, chunkSize int64) *SwapFile {
+	return &SwapFile{
+		dir:                     dir,
+		file:                    nil,
+		logicToActualChunkIndex: make(map[LogicChunkIndex]ActualChunkIndex),
+		chunkSize:               chunkSize,
+	}
+}
+func (sf *SwapFile) FreeResource() {
+	if sf.file != nil {
+		sf.file.Close()
+		os.Remove(sf.file.Name())
+	}
+}
+
+func (sf *SwapFile) NewTempFileChunk(logicChunkIndex LogicChunkIndex) (tc *SwapFileChunk) {
+	if sf.file == nil {
+		var err error
+		sf.file, err = os.CreateTemp(sf.dir, "")
+		if err != nil {
+			glog.Errorf("create swap file: %v", err)
+			return nil
+		}
+	}
+	actualChunkIndex, found := sf.logicToActualChunkIndex[logicChunkIndex]
+	if !found {
+		actualChunkIndex = ActualChunkIndex(len(sf.logicToActualChunkIndex))
+		sf.logicToActualChunkIndex[logicChunkIndex] = actualChunkIndex
+	}
+
+	return &SwapFileChunk{
+		swapfile:         sf,
+		usage:            newChunkWrittenIntervalList(),
+		logicChunkIndex:  logicChunkIndex,
+		actualChunkIndex: actualChunkIndex,
+	}
+}
+
+func (sc *SwapFileChunk) FreeResource() {
+}
+
+func (sc *SwapFileChunk) WriteDataAt(src []byte, offset int64) (n int) {
+	innerOffset := offset % sc.swapfile.chunkSize
+	var err error
+	n, err = sc.swapfile.file.WriteAt(src, int64(sc.actualChunkIndex)*sc.swapfile.chunkSize+innerOffset)
+	if err == nil {
+		sc.usage.MarkWritten(innerOffset, innerOffset+int64(n))
+	} else {
+		glog.Errorf("failed to write swap file %s: %v", sc.swapfile.file.Name(), err)
+	}
+	return
+}
+
+func (sc *SwapFileChunk) ReadDataAt(p []byte, off int64) (maxStop int64) {
+	chunkStartOffset := int64(sc.logicChunkIndex) * sc.swapfile.chunkSize
+	for t := sc.usage.head.next; t != sc.usage.tail; t = t.next {
+		logicStart := max(off, chunkStartOffset+t.StartOffset)
+		logicStop := min(off+int64(len(p)), chunkStartOffset+t.stopOffset)
+		if logicStart < logicStop {
+			actualStart := logicStart - chunkStartOffset + int64(sc.actualChunkIndex)*sc.swapfile.chunkSize
+			if _, err := sc.swapfile.file.ReadAt(p[logicStart-off:logicStop-off], actualStart); err != nil {
+				glog.Errorf("failed to reading swap file %s: %v", sc.swapfile.file.Name(), err)
+				break
+			}
+			maxStop = max(maxStop, logicStop)
+		}
+	}
+	return
+}
+
+func (sc *SwapFileChunk) IsComplete() bool {
+	return sc.usage.IsComplete(sc.swapfile.chunkSize)
+}
+
+func (sc *SwapFileChunk) WrittenSize() int64 {
+	return sc.usage.WrittenSize()
+}
+
+func (sc *SwapFileChunk) SaveContent(saveFn SaveToStorageFunc) {
+	if saveFn == nil {
+		return
+	}
+	for t := sc.usage.head.next; t != sc.usage.tail; t = t.next {
+		data := mem.Allocate(int(t.Size()))
+		sc.swapfile.file.ReadAt(data, t.StartOffset+int64(sc.actualChunkIndex)*sc.swapfile.chunkSize)
+		reader := util.NewBytesReader(data)
+		saveFn(reader, int64(sc.logicChunkIndex)*sc.swapfile.chunkSize+t.StartOffset, t.Size(), func() {
+		})
+		mem.Free(data)
+	}
+	sc.usage = newChunkWrittenIntervalList()
+}
@@ -0,0 +1,182 @@
+package page_writer
+
+import (
+	"fmt"
+	"github.com/chrislusf/seaweedfs/weed/glog"
+	"github.com/chrislusf/seaweedfs/weed/util"
+	"sync"
+	"sync/atomic"
+	"time"
+)
+
+type LogicChunkIndex int
+
+type UploadPipeline struct {
+	filepath             util.FullPath
+	ChunkSize            int64
+	writableChunks       map[LogicChunkIndex]PageChunk
+	writableChunksLock   sync.Mutex
+	sealedChunks         map[LogicChunkIndex]*SealedChunk
+	sealedChunksLock     sync.Mutex
+	uploaders            *util.LimitedConcurrentExecutor
+	uploaderCount        int32
+	uploaderCountCond    *sync.Cond
+	saveToStorageFn      SaveToStorageFunc
+	activeReadChunks     map[LogicChunkIndex]int
+	activeReadChunksLock sync.Mutex
+	bufferChunkLimit     int
+}
+
+type SealedChunk struct {
+	chunk            PageChunk
+	referenceCounter int // track uploading or reading processes
+}
+
+func (sc *SealedChunk) FreeReference(messageOnFree string) {
+	sc.referenceCounter--
+	if sc.referenceCounter == 0 {
+		glog.V(4).Infof("Free sealed chunk: %s", messageOnFree)
+		sc.chunk.FreeResource()
+	}
+}
+
+func NewUploadPipeline(writers *util.LimitedConcurrentExecutor, chunkSize int64, saveToStorageFn SaveToStorageFunc, bufferChunkLimit int) *UploadPipeline {
+	return &UploadPipeline{
+		ChunkSize:         chunkSize,
+		writableChunks:    make(map[LogicChunkIndex]PageChunk),
+		sealedChunks:      make(map[LogicChunkIndex]*SealedChunk),
+		uploaders:         writers,
+		uploaderCountCond: sync.NewCond(&sync.Mutex{}),
+		saveToStorageFn:   saveToStorageFn,
+		activeReadChunks:  make(map[LogicChunkIndex]int),
+		bufferChunkLimit:  bufferChunkLimit,
+	}
+}
+
+func (up *UploadPipeline) SaveDataAt(p []byte, off int64) (n int) {
+	up.writableChunksLock.Lock()
+	defer up.writableChunksLock.Unlock()
+
+	logicChunkIndex := LogicChunkIndex(off / up.ChunkSize)
+
+	memChunk, found := up.writableChunks[logicChunkIndex]
+	if !found {
+		if len(up.writableChunks) < up.bufferChunkLimit {
+			memChunk = NewMemChunk(logicChunkIndex, up.ChunkSize)
+		} else {
+			fullestChunkIndex, fullness := LogicChunkIndex(-1), int64(0)
+			for lci, mc := range up.writableChunks {
+				chunkFullness := mc.WrittenSize()
+				if fullness < chunkFullness {
+					fullestChunkIndex = lci
+					fullness = chunkFullness
+				}
+			}
+			up.moveToSealed(up.writableChunks[fullestChunkIndex], fullestChunkIndex)
+			delete(up.writableChunks, fullestChunkIndex)
+			fmt.Printf("flush chunk %d with %d bytes written", logicChunkIndex, fullness)
+			memChunk = NewMemChunk(logicChunkIndex, up.ChunkSize)
+		}
+		up.writableChunks[logicChunkIndex] = memChunk
+	}
+	n = memChunk.WriteDataAt(p, off)
+	up.maybeMoveToSealed(memChunk, logicChunkIndex)
+
+	return
+}
+
+func (up *UploadPipeline) MaybeReadDataAt(p []byte, off int64) (maxStop int64) {
+	logicChunkIndex := LogicChunkIndex(off / up.ChunkSize)
+
+	// read from sealed chunks first
+	up.sealedChunksLock.Lock()
+	sealedChunk, found := up.sealedChunks[logicChunkIndex]
+	if found {
+		sealedChunk.referenceCounter++
+	}
+	up.sealedChunksLock.Unlock()
+	if found {
+		maxStop = sealedChunk.chunk.ReadDataAt(p, off)
+		glog.V(4).Infof("%s read sealed memchunk [%d,%d)", up.filepath, off, maxStop)
+		sealedChunk.FreeReference(fmt.Sprintf("%s finish reading chunk %d", up.filepath, logicChunkIndex))
+	}
+
+	// read from writable chunks last
+	up.writableChunksLock.Lock()
+	defer up.writableChunksLock.Unlock()
+	writableChunk, found := up.writableChunks[logicChunkIndex]
+	if !found {
+		return
+	}
+	writableMaxStop := writableChunk.ReadDataAt(p, off)
+	glog.V(4).Infof("%s read writable memchunk [%d,%d)", up.filepath, off, writableMaxStop)
+	maxStop = max(maxStop, writableMaxStop)
+
+	return
+}
+
+func (up *UploadPipeline) FlushAll() {
+	up.writableChunksLock.Lock()
+	defer up.writableChunksLock.Unlock()
+
+	for logicChunkIndex, memChunk := range up.writableChunks {
+		up.moveToSealed(memChunk, logicChunkIndex)
+	}
+
+	up.waitForCurrentWritersToComplete()
+}
+
+func (up *UploadPipeline) maybeMoveToSealed(memChunk PageChunk, logicChunkIndex LogicChunkIndex) {
+	if memChunk.IsComplete() {
+		up.moveToSealed(memChunk, logicChunkIndex)
+	}
+}
+
+func (up *UploadPipeline) moveToSealed(memChunk PageChunk, logicChunkIndex LogicChunkIndex) {
+	atomic.AddInt32(&up.uploaderCount, 1)
+	glog.V(4).Infof("%s uploaderCount %d ++> %d", up.filepath, up.uploaderCount-1, up.uploaderCount)
+
+	up.sealedChunksLock.Lock()
+
+	if oldMemChunk, found := up.sealedChunks[logicChunkIndex]; found {
+		oldMemChunk.FreeReference(fmt.Sprintf("%s replace chunk %d", up.filepath, logicChunkIndex))
+	}
+	sealedChunk := &SealedChunk{
+		chunk:            memChunk,
+		referenceCounter: 1, // default 1 is for uploading process
+	}
+	up.sealedChunks[logicChunkIndex] = sealedChunk
+	delete(up.writableChunks, logicChunkIndex)
+
+	up.sealedChunksLock.Unlock()
+
+	up.uploaders.Execute(func() {
+		// first add to the file chunks
+		sealedChunk.chunk.SaveContent(up.saveToStorageFn)
+
+		// notify waiting process
+		atomic.AddInt32(&up.uploaderCount, -1)
+		glog.V(4).Infof("%s uploaderCount %d --> %d", up.filepath, up.uploaderCount+1, up.uploaderCount)
+		// Lock and Unlock are not required,
+		// but it may signal multiple times during one wakeup,
+		// and the waiting goroutine may miss some of them!
+		up.uploaderCountCond.L.Lock()
+		up.uploaderCountCond.Broadcast()
+		up.uploaderCountCond.L.Unlock()
+
+		// wait for readers
+		for up.IsLocked(logicChunkIndex) {
+			time.Sleep(59 * time.Millisecond)
+		}
+
+		// then remove from sealed chunks
+		up.sealedChunksLock.Lock()
+		defer up.sealedChunksLock.Unlock()
+		delete(up.sealedChunks, logicChunkIndex)
+		sealedChunk.FreeReference(fmt.Sprintf("%s finished uploading chunk %d", up.filepath, logicChunkIndex))
+
+	})
+}
+
+func (up *UploadPipeline) Shutdown() {
+}
@@ -0,0 +1,63 @@
+package page_writer
+
+import (
+	"sync/atomic"
+)
+
+func (up *UploadPipeline) LockForRead(startOffset, stopOffset int64) {
+	startLogicChunkIndex := LogicChunkIndex(startOffset / up.ChunkSize)
+	stopLogicChunkIndex := LogicChunkIndex(stopOffset / up.ChunkSize)
+	if stopOffset%up.ChunkSize > 0 {
+		stopLogicChunkIndex += 1
+	}
+	up.activeReadChunksLock.Lock()
+	defer up.activeReadChunksLock.Unlock()
+	for i := startLogicChunkIndex; i < stopLogicChunkIndex; i++ {
+		if count, found := up.activeReadChunks[i]; found {
+			up.activeReadChunks[i] = count + 1
+		} else {
+			up.activeReadChunks[i] = 1
+		}
+	}
+}
+
+func (up *UploadPipeline) UnlockForRead(startOffset, stopOffset int64) {
+	startLogicChunkIndex := LogicChunkIndex(startOffset / up.ChunkSize)
+	stopLogicChunkIndex := LogicChunkIndex(stopOffset / up.ChunkSize)
+	if stopOffset%up.ChunkSize > 0 {
+		stopLogicChunkIndex += 1
+	}
+	up.activeReadChunksLock.Lock()
+	defer up.activeReadChunksLock.Unlock()
+	for i := startLogicChunkIndex; i < stopLogicChunkIndex; i++ {
+		if count, found := up.activeReadChunks[i]; found {
+			if count == 1 {
+				delete(up.activeReadChunks, i)
+			} else {
+				up.activeReadChunks[i] = count - 1
+			}
+		}
+	}
+}
+
+func (up *UploadPipeline) IsLocked(logicChunkIndex LogicChunkIndex) bool {
+	up.activeReadChunksLock.Lock()
+	defer up.activeReadChunksLock.Unlock()
+	if count, found := up.activeReadChunks[logicChunkIndex]; found {
+		return count > 0
+	}
+	return false
+}
+
+func (up *UploadPipeline) waitForCurrentWritersToComplete() {
+	up.uploaderCountCond.L.Lock()
+	t := int32(100)
+	for {
+		t = atomic.LoadInt32(&up.uploaderCount)
+		if t <= 0 {
+			break
+		}
+		up.uploaderCountCond.Wait()
+	}
+	up.uploaderCountCond.L.Unlock()
+}
@@ -0,0 +1,47 @@
+package page_writer
+
+import (
+	"github.com/chrislusf/seaweedfs/weed/util"
+	"testing"
+)
+
+func TestUploadPipeline(t *testing.T) {
+
+	uploadPipeline := NewUploadPipeline(nil, 2*1024*1024, nil, 16)
+
+	writeRange(uploadPipeline, 0, 131072)
+	writeRange(uploadPipeline, 131072, 262144)
+	writeRange(uploadPipeline, 262144, 1025536)
+
+	confirmRange(t, uploadPipeline, 0, 1025536)
+
+	writeRange(uploadPipeline, 1025536, 1296896)
+
+	confirmRange(t, uploadPipeline, 1025536, 1296896)
+
+	writeRange(uploadPipeline, 1296896, 2162688)
+
+	confirmRange(t, uploadPipeline, 1296896, 2162688)
+
+	confirmRange(t, uploadPipeline, 1296896, 2162688)
+}
+
+// startOff and stopOff must be divided by 4
+func writeRange(uploadPipeline *UploadPipeline, startOff, stopOff int64) {
+	p := make([]byte, 4)
+	for i := startOff / 4; i < stopOff/4; i += 4 {
+		util.Uint32toBytes(p, uint32(i))
+		uploadPipeline.SaveDataAt(p, i)
+	}
+}
+
+func confirmRange(t *testing.T, uploadPipeline *UploadPipeline, startOff, stopOff int64) {
+	p := make([]byte, 4)
+	for i := startOff; i < stopOff/4; i += 4 {
+		uploadPipeline.MaybeReadDataAt(p, i)
+		x := util.BytesToUint32(p)
+		if x != uint32(i) {
+			t.Errorf("expecting %d found %d at offset [%d,%d)", i, x, i, i+4)
+		}
+	}
+}
@@ -0,0 +1,63 @@
+package filesys
+
+import (
+	"github.com/chrislusf/seaweedfs/weed/pb/filer_pb"
+	"github.com/seaweedfs/fuse"
+)
+
+func checkPermission(entry *filer_pb.Entry, uid, gid uint32, isWrite bool) error {
+	if uid == 0 || gid == 0 {
+		return nil
+	}
+	if entry == nil {
+		return nil
+	}
+	if entry.Attributes == nil {
+		return nil
+	}
+	attr := entry.Attributes
+	if attr.Uid == uid {
+		if isWrite {
+			if attr.FileMode&0200 > 0 {
+				return nil
+			} else {
+				return fuse.EPERM
+			}
+		} else {
+			if attr.FileMode&0400 > 0 {
+				return nil
+			} else {
+				return fuse.EPERM
+			}
+		}
+	} else if attr.Gid == gid {
+		if isWrite {
+			if attr.FileMode&0020 > 0 {
+				return nil
+			} else {
+				return fuse.EPERM
+			}
+		} else {
+			if attr.FileMode&0040 > 0 {
+				return nil
+			} else {
+				return fuse.EPERM
+			}
+		}
+	} else {
+		if isWrite {
+			if attr.FileMode&0002 > 0 {
+				return nil
+			} else {
+				return fuse.EPERM
+			}
+		} else {
+			if attr.FileMode&0004 > 0 {
+				return nil
+			} else {
+				return fuse.EPERM
+			}
+		}
+	}
+
+}
@@ -109,15 +109,15 @@ func NewSeaweedFileSystem(option *Option) *WFS {
 		wfs.chunkCache = chunk_cache.NewTieredChunkCache(256, option.getUniqueCacheDir(), option.CacheSizeMB, 1024*1024)
 	}

-	wfs.metaCache = meta_cache.NewMetaCache(path.Join(option.getUniqueCacheDir(), "meta"), util.FullPath(option.FilerMountRootPath), option.UidGidMapper, func(filePath util.FullPath) {
+	wfs.metaCache = meta_cache.NewMetaCache(path.Join(option.getUniqueCacheDir(), "meta"), util.FullPath(option.FilerMountRootPath), option.UidGidMapper, func(filePath util.FullPath, entry *filer_pb.Entry) {

-		fsNode := NodeWithId(filePath.AsInode())
+		fsNode := NodeWithId(filePath.AsInode(entry.FileMode()))
 		if err := wfs.Server.InvalidateNodeData(fsNode); err != nil {
 			glog.V(4).Infof("InvalidateNodeData %s : %v", filePath, err)
 		}

 		dir, name := filePath.DirAndName()
-		parent := NodeWithId(util.FullPath(dir).AsInode())
+		parent := NodeWithId(util.FullPath(dir).AsInode(os.ModeDir))
 		if dir == option.FilerMountRootPath {
 			parent = NodeWithId(1)
 		}
@@ -179,6 +179,28 @@ func (wfs *WFS) AcquireHandle(file *File, uid, gid uint32) (fileHandle *FileHand
 	return
 }

+func (wfs *WFS) Fsync(file *File, header fuse.Header) error {
+
+	inodeId := file.Id()
+
+	wfs.handlesLock.Lock()
+	existingHandle, found := wfs.handles[inodeId]
+	wfs.handlesLock.Unlock()
+
+	if found && existingHandle != nil {
+
+		existingHandle.Lock()
+		defer existingHandle.Unlock()
+
+		if existingHandle.f.isOpen > 0 {
+			return existingHandle.doFlush(context.Background(), header)
+		}
+
+	}
+
+	return nil
+}
+
 func (wfs *WFS) ReleaseHandle(fullpath util.FullPath, handleId fuse.HandleID) {
 	wfs.handlesLock.Lock()
 	defer wfs.handlesLock.Unlock()
@@ -3,6 +3,7 @@ package filesys
 import (
 	"context"
 	"strings"
+	"syscall"

 	"github.com/seaweedfs/fuse"

@@ -143,3 +144,10 @@ func (wfs *WFS) maybeLoadEntry(dir, name string) (entry *filer_pb.Entry, err err
 	}
 	return cachedEntry.ToProtoEntry(), cacheErr
 }
+
+func checkName(name string) error {
+	if len(name) >= 256 {
+		return syscall.ENAMETOOLONG
+	}
+	return nil
+}
@@ -1,6 +1,7 @@
 package images

 import (
+	"github.com/chrislusf/seaweedfs/weed/util"
 	"os"
 	"testing"
 )
@@ -12,7 +13,7 @@ func TestXYZ(t *testing.T) {

 	fixed_data := FixJpgOrientation(dat)

-	os.WriteFile("fixed1.jpg", fixed_data, 0644)
+	util.WriteFile("fixed1.jpg", fixed_data, 0644)

 	os.Remove("fixed1.jpg")

@@ -2,6 +2,7 @@ package images

 import (
 	"bytes"
+	"github.com/chrislusf/seaweedfs/weed/util"
 	"os"
 	"testing"
 )
@@ -15,7 +16,7 @@ func TestResizing(t *testing.T) {
 	buf := new(bytes.Buffer)
 	buf.ReadFrom(resized)

-	os.WriteFile("resized1.png", buf.Bytes(), 0644)
+	util.WriteFile("resized1.png", buf.Bytes(), 0644)

 	os.Remove("resized1.png")

@@ -0,0 +1,42 @@
+package mount
+
+import (
+	"bytes"
+	"github.com/chrislusf/seaweedfs/weed/pb/filer_pb"
+	"github.com/hanwen/go-fuse/v2/fs"
+	"strings"
+)
+
+type Directory struct {
+	fs.Inode
+
+	name   string
+	wfs    *WFS
+	entry  *filer_pb.Entry
+	parent *Directory
+	id     uint64
+}
+
+func (dir *Directory) FullPath() string {
+	var parts []string
+	for p := dir; p != nil; p = p.parent {
+		if strings.HasPrefix(p.name, "/") {
+			if len(p.name) > 1 {
+				parts = append(parts, p.name[1:])
+			}
+		} else {
+			parts = append(parts, p.name)
+		}
+	}
+
+	if len(parts) == 0 {
+		return "/"
+	}
+
+	var buf bytes.Buffer
+	for i := len(parts) - 1; i >= 0; i-- {
+		buf.WriteString("/")
+		buf.WriteString(parts[i])
+	}
+	return buf.String()
+}
@@ -0,0 +1,84 @@
+package mount
+
+import (
+	"context"
+	"github.com/chrislusf/seaweedfs/weed/filer"
+	"github.com/chrislusf/seaweedfs/weed/glog"
+	"github.com/chrislusf/seaweedfs/weed/mount/meta_cache"
+	"github.com/chrislusf/seaweedfs/weed/util"
+	"github.com/hanwen/go-fuse/v2/fs"
+	"github.com/hanwen/go-fuse/v2/fuse"
+	"math"
+	"os"
+	"syscall"
+)
+
+var _ = fs.NodeReaddirer(&Directory{})
+var _ = fs.NodeGetattrer(&Directory{})
+
+func (dir *Directory) Getattr(ctx context.Context, fh fs.FileHandle, out *fuse.AttrOut) syscall.Errno {
+	out.Mode = 0755
+	return 0
+}
+
+func (dir *Directory) Readdir(ctx context.Context) (fs.DirStream, syscall.Errno) {
+
+	dirPath := util.FullPath(dir.FullPath())
+	glog.V(4).Infof("Readdir %s", dirPath)
+
+	sourceChan := make(chan fuse.DirEntry, 64)
+
+	stream := newDirectoryListStream(sourceChan)
+
+	processEachEntryFn := func(entry *filer.Entry, isLast bool) {
+		sourceChan <- fuse.DirEntry{
+			Mode: uint32(entry.Mode),
+			Name: entry.Name(),
+			Ino:  dirPath.Child(entry.Name()).AsInode(os.ModeDir),
+		}
+	}
+
+	if err := meta_cache.EnsureVisited(dir.wfs.metaCache, dir.wfs, dirPath); err != nil {
+		glog.Errorf("dir ReadDirAll %s: %v", dirPath, err)
+		return nil, fs.ToErrno(os.ErrInvalid)
+	}
+	go func() {
+		dir.wfs.metaCache.ListDirectoryEntries(context.Background(), dirPath, "", false, int64(math.MaxInt32), func(entry *filer.Entry) bool {
+			processEachEntryFn(entry, false)
+			return true
+		})
+		close(sourceChan)
+	}()
+
+	return stream, fs.OK
+}
+
+var _ = fs.DirStream(&DirectoryListStream{})
+
+type DirectoryListStream struct {
+	next       fuse.DirEntry
+	sourceChan chan fuse.DirEntry
+	isStarted  bool
+	hasNext    bool
+}
+
+func newDirectoryListStream(ch chan fuse.DirEntry) *DirectoryListStream {
+	return &DirectoryListStream{
+		sourceChan: ch,
+	}
+}
+
+func (i *DirectoryListStream) HasNext() bool {
+	if !i.isStarted {
+		i.next, i.hasNext = <-i.sourceChan
+		i.isStarted = true
+	}
+	return i.hasNext
+}
+func (i *DirectoryListStream) Next() (fuse.DirEntry, syscall.Errno) {
+	t := i.next
+	i.next, i.hasNext = <-i.sourceChan
+	return t, fs.OK
+}
+func (i *DirectoryListStream) Close() {
+}
@@ -0,0 +1,99 @@
+package mount
+
+import (
+	"fmt"
+	"github.com/chrislusf/seaweedfs/weed/filesys/page_writer"
+	"github.com/chrislusf/seaweedfs/weed/glog"
+	"github.com/chrislusf/seaweedfs/weed/pb/filer_pb"
+	"io"
+	"sync"
+	"time"
+)
+
+type ChunkedDirtyPages struct {
+	fh             *FileHandle
+	writeWaitGroup sync.WaitGroup
+	lastErr        error
+	collection     string
+	replication    string
+	uploadPipeline *page_writer.UploadPipeline
+	hasWrites      bool
+}
+
+var (
+	_ = page_writer.DirtyPages(&ChunkedDirtyPages{})
+)
+
+func newMemoryChunkPages(fh *FileHandle, chunkSize int64) *ChunkedDirtyPages {
+
+	dirtyPages := &ChunkedDirtyPages{
+		fh: fh,
+	}
+
+	dirtyPages.uploadPipeline = page_writer.NewUploadPipeline(fh.wfs.concurrentWriters, chunkSize, dirtyPages.saveChunkedFileIntevalToStorage, fh.wfs.option.ConcurrentWriters)
+
+	return dirtyPages
+}
+
+func (pages *ChunkedDirtyPages) AddPage(offset int64, data []byte) {
+	pages.hasWrites = true
+
+	glog.V(4).Infof("%v memory AddPage [%d, %d)", pages.fh, offset, offset+int64(len(data)))
+	pages.uploadPipeline.SaveDataAt(data, offset)
+
+	return
+}
+
+func (pages *ChunkedDirtyPages) FlushData() error {
+	if !pages.hasWrites {
+		return nil
+	}
+	pages.uploadPipeline.FlushAll()
+	if pages.lastErr != nil {
+		return fmt.Errorf("flush data: %v", pages.lastErr)
+	}
+	return nil
+}
+
+func (pages *ChunkedDirtyPages) ReadDirtyDataAt(data []byte, startOffset int64) (maxStop int64) {
+	if !pages.hasWrites {
+		return
+	}
+	return pages.uploadPipeline.MaybeReadDataAt(data, startOffset)
+}
+
+func (pages *ChunkedDirtyPages) GetStorageOptions() (collection, replication string) {
+	return pages.collection, pages.replication
+}
+
+func (pages *ChunkedDirtyPages) saveChunkedFileIntevalToStorage(reader io.Reader, offset int64, size int64, cleanupFn func()) {
+
+	mtime := time.Now().UnixNano()
+	defer cleanupFn()
+
+	fileFullPath := pages.fh.FullPath()
+	fileName := fileFullPath.Name()
+	chunk, collection, replication, err := pages.fh.wfs.saveDataAsChunk(fileFullPath)(reader, fileName, offset)
+	if err != nil {
+		glog.V(0).Infof("%v saveToStorage [%d,%d): %v", fileFullPath, offset, offset+size, err)
+		pages.lastErr = err
+		return
+	}
+	chunk.Mtime = mtime
+	pages.collection, pages.replication = collection, replication
+	pages.fh.addChunks([]*filer_pb.FileChunk{chunk})
+	pages.fh.entryViewCache = nil
+	glog.V(3).Infof("%v saveToStorage %s [%d,%d)", fileFullPath, chunk.FileId, offset, offset+size)
+
+}
+
+func (pages ChunkedDirtyPages) Destroy() {
+	pages.uploadPipeline.Shutdown()
+}
+
+func (pages *ChunkedDirtyPages) LockForRead(startOffset, stopOffset int64) {
+	pages.uploadPipeline.LockForRead(startOffset, stopOffset)
+}
+func (pages *ChunkedDirtyPages) UnlockForRead(startOffset, stopOffset int64) {
+	pages.uploadPipeline.UnlockForRead(startOffset, stopOffset)
+}
@@ -0,0 +1,95 @@
+package mount
+
+import (
+	"github.com/chrislusf/seaweedfs/weed/filer"
+	"github.com/chrislusf/seaweedfs/weed/glog"
+	"github.com/chrislusf/seaweedfs/weed/pb/filer_pb"
+	"github.com/chrislusf/seaweedfs/weed/util"
+	"io"
+	"sort"
+	"sync"
+)
+
+type FileHandleId uint64
+
+type FileHandle struct {
+	fh           FileHandleId
+	counter      int64
+	entry        *filer_pb.Entry
+	chunkAddLock sync.Mutex
+	inode        uint64
+	wfs          *WFS
+
+	// cache file has been written to
+	dirtyMetadata  bool
+	dirtyPages     *PageWriter
+	entryViewCache []filer.VisibleInterval
+	reader         io.ReaderAt
+	contentType    string
+	handle         uint64
+	sync.Mutex
+
+	isDeleted bool
+}
+
+func newFileHandle(wfs *WFS, handleId FileHandleId, inode uint64, entry *filer_pb.Entry) *FileHandle {
+	fh := &FileHandle{
+		fh:      handleId,
+		counter: 1,
+		inode:   inode,
+		wfs:     wfs,
+	}
+	// dirtyPages: newContinuousDirtyPages(file, writeOnly),
+	fh.dirtyPages = newPageWriter(fh, wfs.option.ChunkSizeLimit)
+	if entry != nil {
+		entry.Attributes.FileSize = filer.FileSize(entry)
+	}
+
+	return fh
+}
+
+func (fh *FileHandle) FullPath() util.FullPath {
+	return fh.wfs.inodeToPath.GetPath(fh.inode)
+}
+
+func (fh *FileHandle) addChunks(chunks []*filer_pb.FileChunk) {
+
+	// find the earliest incoming chunk
+	newChunks := chunks
+	earliestChunk := newChunks[0]
+	for i := 1; i < len(newChunks); i++ {
+		if lessThan(earliestChunk, newChunks[i]) {
+			earliestChunk = newChunks[i]
+		}
+	}
+
+	if fh.entry == nil {
+		return
+	}
+
+	// pick out-of-order chunks from existing chunks
+	for _, chunk := range fh.entry.Chunks {
+		if lessThan(earliestChunk, chunk) {
+			chunks = append(chunks, chunk)
+		}
+	}
+
+	// sort incoming chunks
+	sort.Slice(chunks, func(i, j int) bool {
+		return lessThan(chunks[i], chunks[j])
+	})
+
+	glog.V(4).Infof("%s existing %d chunks adds %d more", fh.FullPath(), len(fh.entry.Chunks), len(chunks))
+
+	fh.chunkAddLock.Lock()
+	fh.entry.Chunks = append(fh.entry.Chunks, newChunks...)
+	fh.entryViewCache = nil
+	fh.chunkAddLock.Unlock()
+}
+
+func lessThan(a, b *filer_pb.FileChunk) bool {
+	if a.Mtime == b.Mtime {
+		return a.Fid.FileKey < b.Fid.FileKey
+	}
+	return a.Mtime < b.Mtime
+}
@@ -0,0 +1,84 @@
+package mount
+
+import (
+	"github.com/chrislusf/seaweedfs/weed/pb/filer_pb"
+	"sync"
+)
+
+type FileHandleToInode struct {
+	sync.RWMutex
+	nextFh   FileHandleId
+	inode2fh map[uint64]*FileHandle
+	fh2inode map[FileHandleId]uint64
+}
+
+func NewFileHandleToInode() *FileHandleToInode {
+	return &FileHandleToInode{
+		inode2fh: make(map[uint64]*FileHandle),
+		fh2inode: make(map[FileHandleId]uint64),
+		nextFh:   0,
+	}
+}
+
+func (i *FileHandleToInode) GetFileHandle(fh FileHandleId) *FileHandle {
+	i.RLock()
+	defer i.RUnlock()
+	inode, found := i.fh2inode[fh]
+	if found {
+		return i.inode2fh[inode]
+	}
+	return nil
+}
+
+func (i *FileHandleToInode) FindFileHandle(inode uint64) (fh *FileHandle, found bool) {
+	i.RLock()
+	defer i.RUnlock()
+	fh, found = i.inode2fh[inode]
+	return
+}
+
+func (i *FileHandleToInode) AcquireFileHandle(wfs *WFS, inode uint64, entry *filer_pb.Entry) *FileHandle {
+	i.Lock()
+	defer i.Unlock()
+	fh, found := i.inode2fh[inode]
+	if !found {
+		fh = newFileHandle(wfs, i.nextFh, inode, entry)
+		i.nextFh++
+		i.inode2fh[inode] = fh
+		i.fh2inode[fh.fh] = inode
+	} else {
+		fh.counter++
+	}
+	return fh
+}
+
+func (i *FileHandleToInode) ReleaseByInode(inode uint64) {
+	i.Lock()
+	defer i.Unlock()
+	fh, found := i.inode2fh[inode]
+	if found {
+		fh.counter--
+		if fh.counter <= 0 {
+			delete(i.inode2fh, inode)
+			delete(i.fh2inode, fh.fh)
+		}
+	}
+}
+func (i *FileHandleToInode) ReleaseByHandle(fh FileHandleId) {
+	i.Lock()
+	defer i.Unlock()
+	inode, found := i.fh2inode[fh]
+	if found {
+		fhHandle, fhFound := i.inode2fh[inode]
+		if !fhFound {
+			delete(i.fh2inode, fh)
+		} else {
+			fhHandle.counter--
+			if fhHandle.counter <= 0 {
+				delete(i.inode2fh, inode)
+				delete(i.fh2inode, fhHandle.fh)
+			}
+		}
+
+	}
+}
@@ -0,0 +1,114 @@
+package mount
+
+import (
+	"context"
+	"fmt"
+	"github.com/chrislusf/seaweedfs/weed/filer"
+	"github.com/chrislusf/seaweedfs/weed/glog"
+	"github.com/chrislusf/seaweedfs/weed/pb/filer_pb"
+	"io"
+	"math"
+)
+
+func (fh *FileHandle) lockForRead(startOffset int64, size int) {
+	fh.dirtyPages.LockForRead(startOffset, startOffset+int64(size))
+}
+func (fh *FileHandle) unlockForRead(startOffset int64, size int) {
+	fh.dirtyPages.UnlockForRead(startOffset, startOffset+int64(size))
+}
+
+func (fh *FileHandle) readFromDirtyPages(buff []byte, startOffset int64) (maxStop int64) {
+	maxStop = fh.dirtyPages.ReadDirtyDataAt(buff, startOffset)
+	return
+}
+
+func (fh *FileHandle) readFromChunks(buff []byte, offset int64) (int64, error) {
+
+	fileFullPath := fh.FullPath()
+
+	entry := fh.entry
+	if entry == nil {
+		return 0, io.EOF
+	}
+
+	if entry.IsInRemoteOnly() {
+		glog.V(4).Infof("download remote entry %s", fileFullPath)
+		newEntry, err := fh.downloadRemoteEntry(entry)
+		if err != nil {
+			glog.V(1).Infof("download remote entry %s: %v", fileFullPath, err)
+			return 0, err
+		}
+		entry = newEntry
+	}
+
+	fileSize := int64(filer.FileSize(entry))
+
+	if fileSize == 0 {
+		glog.V(1).Infof("empty fh %v", fileFullPath)
+		return 0, io.EOF
+	}
+
+	if offset+int64(len(buff)) <= int64(len(entry.Content)) {
+		totalRead := copy(buff, entry.Content[offset:])
+		glog.V(4).Infof("file handle read cached %s [%d,%d] %d", fileFullPath, offset, offset+int64(totalRead), totalRead)
+		return int64(totalRead), nil
+	}
+
+	var chunkResolveErr error
+	if fh.entryViewCache == nil {
+		fh.entryViewCache, chunkResolveErr = filer.NonOverlappingVisibleIntervals(fh.wfs.LookupFn(), entry.Chunks, 0, math.MaxInt64)
+		if chunkResolveErr != nil {
+			return 0, fmt.Errorf("fail to resolve chunk manifest: %v", chunkResolveErr)
+		}
+		fh.reader = nil
+	}
+
+	reader := fh.reader
+	if reader == nil {
+		chunkViews := filer.ViewFromVisibleIntervals(fh.entryViewCache, 0, math.MaxInt64)
+		glog.V(4).Infof("file handle read %s [%d,%d) from %d views", fileFullPath, offset, offset+int64(len(buff)), len(chunkViews))
+		for _, chunkView := range chunkViews {
+			glog.V(4).Infof("  read %s [%d,%d) from chunk %+v", fileFullPath, chunkView.LogicOffset, chunkView.LogicOffset+int64(chunkView.Size), chunkView.FileId)
+		}
+		reader = filer.NewChunkReaderAtFromClient(fh.wfs.LookupFn(), chunkViews, fh.wfs.chunkCache, fileSize)
+	}
+	fh.reader = reader
+
+	totalRead, err := reader.ReadAt(buff, offset)
+
+	if err != nil && err != io.EOF {
+		glog.Errorf("file handle read %s: %v", fileFullPath, err)
+	}
+
+	glog.V(4).Infof("file handle read %s [%d,%d] %d : %v", fileFullPath, offset, offset+int64(totalRead), totalRead, err)
+
+	return int64(totalRead), err
+}
+
+func (fh *FileHandle) downloadRemoteEntry(entry *filer_pb.Entry) (*filer_pb.Entry, error) {
+
+	fileFullPath := fh.FullPath()
+	dir, _ := fileFullPath.DirAndName()
+
+	err := fh.wfs.WithFilerClient(false, func(client filer_pb.SeaweedFilerClient) error {
+
+		request := &filer_pb.CacheRemoteObjectToLocalClusterRequest{
+			Directory: string(dir),
+			Name:      entry.Name,
+		}
+
+		glog.V(4).Infof("download entry: %v", request)
+		resp, err := client.CacheRemoteObjectToLocalCluster(context.Background(), request)
+		if err != nil {
+			return fmt.Errorf("CacheRemoteObjectToLocalCluster file %s: %v", fileFullPath, err)
+		}
+
+		entry = resp.Entry
+
+		fh.wfs.metaCache.InsertEntry(context.Background(), filer.FromPbEntry(request.Directory, resp.Entry))
+
+		return nil
+	})
+
+	return entry, err
+}
@@ -0,0 +1,161 @@
+package mount
+
+import (
+	"github.com/chrislusf/seaweedfs/weed/glog"
+	"github.com/chrislusf/seaweedfs/weed/util"
+	"sync"
+)
+
+type InodeToPath struct {
+	sync.RWMutex
+	nextInodeId uint64
+	inode2path  map[uint64]*InodeEntry
+	path2inode  map[util.FullPath]uint64
+}
+type InodeEntry struct {
+	util.FullPath
+	nlookup          uint64
+	isDirectory      bool
+	isChildrenCached bool
+}
+
+func NewInodeToPath() *InodeToPath {
+	t := &InodeToPath{
+		inode2path:  make(map[uint64]*InodeEntry),
+		path2inode:  make(map[util.FullPath]uint64),
+		nextInodeId: 2, // the root inode id is 1
+	}
+	t.inode2path[1] = &InodeEntry{"/", 1, true, false}
+	t.path2inode["/"] = 1
+	return t
+}
+
+func (i *InodeToPath) Lookup(path util.FullPath, isDirectory bool) uint64 {
+	i.Lock()
+	defer i.Unlock()
+	inode, found := i.path2inode[path]
+	if !found {
+		inode = i.nextInodeId
+		i.nextInodeId++
+		i.path2inode[path] = inode
+		i.inode2path[inode] = &InodeEntry{path, 1, isDirectory, false}
+	} else {
+		i.inode2path[inode].nlookup++
+	}
+	return inode
+}
+
+func (i *InodeToPath) GetInode(path util.FullPath) uint64 {
+	if path == "/" {
+		return 1
+	}
+	i.Lock()
+	defer i.Unlock()
+	inode, found := i.path2inode[path]
+	if !found {
+		// glog.Fatalf("GetInode unknown inode for %s", path)
+		// this could be the parent for mount point
+	}
+	return inode
+}
+
+func (i *InodeToPath) GetPath(inode uint64) util.FullPath {
+	i.RLock()
+	defer i.RUnlock()
+	path, found := i.inode2path[inode]
+	if !found {
+		glog.Fatalf("not found inode %d", inode)
+	}
+	return path.FullPath
+}
+
+func (i *InodeToPath) HasPath(path util.FullPath) bool {
+	i.RLock()
+	defer i.RUnlock()
+	_, found := i.path2inode[path]
+	return found
+}
+
+func (i *InodeToPath) MarkChildrenCached(fullpath util.FullPath) {
+	i.RLock()
+	defer i.RUnlock()
+	inode, found := i.path2inode[fullpath]
+	if !found {
+		glog.Fatalf("MarkChildrenCached not found inode %v", fullpath)
+	}
+	path, found := i.inode2path[inode]
+	path.isChildrenCached = true
+}
+
+func (i *InodeToPath) IsChildrenCached(fullpath util.FullPath) bool {
+	i.RLock()
+	defer i.RUnlock()
+	inode, found := i.path2inode[fullpath]
+	if !found {
+		return false
+	}
+	path, found := i.inode2path[inode]
+	if found {
+		return path.isChildrenCached
+	}
+	return false
+}
+
+func (i *InodeToPath) HasInode(inode uint64) bool {
+	if inode == 1 {
+		return true
+	}
+	i.RLock()
+	defer i.RUnlock()
+	_, found := i.inode2path[inode]
+	return found
+}
+
+func (i *InodeToPath) RemovePath(path util.FullPath) {
+	i.Lock()
+	defer i.Unlock()
+	inode, found := i.path2inode[path]
+	if found {
+		delete(i.path2inode, path)
+		delete(i.inode2path, inode)
+	}
+}
+
+func (i *InodeToPath) MovePath(sourcePath, targetPath util.FullPath) {
+	i.Lock()
+	defer i.Unlock()
+	sourceInode, sourceFound := i.path2inode[sourcePath]
+	targetInode, targetFound := i.path2inode[targetPath]
+	if sourceFound {
+		delete(i.path2inode, sourcePath)
+		i.path2inode[targetPath] = sourceInode
+	} else {
+		// it is possible some source folder items has not been visited before
+		// so no need to worry about their source inodes
+		return
+	}
+	i.inode2path[sourceInode].FullPath = targetPath
+	if targetFound {
+		delete(i.inode2path, targetInode)
+	} else {
+		i.inode2path[sourceInode].nlookup++
+	}
+}
+
+func (i *InodeToPath) Forget(inode, nlookup uint64, onForgetDir func(dir util.FullPath)) {
+	i.Lock()
+	path, found := i.inode2path[inode]
+	if found {
+		path.nlookup -= nlookup
+		if path.nlookup <= 0 {
+			delete(i.path2inode, path.FullPath)
+			delete(i.inode2path, inode)
+		}
+	}
+	i.Unlock()
+	if found {
+		if path.isDirectory && onForgetDir != nil {
+			onForgetDir(path.FullPath)
+		}
+	}
+}
@@ -0,0 +1,32 @@
+package meta_cache
+
+import "github.com/chrislusf/seaweedfs/weed/util"
+
+var (
+	_ = util.Configuration(&cacheConfig{})
+)
+
+// implementing util.Configuraion
+type cacheConfig struct {
+	dir string
+}
+
+func (c cacheConfig) GetString(key string) string {
+	return c.dir
+}
+
+func (c cacheConfig) GetBool(key string) bool {
+	panic("implement me")
+}
+
+func (c cacheConfig) GetInt(key string) int {
+	panic("implement me")
+}
+
+func (c cacheConfig) GetStringSlice(key string) []string {
+	panic("implement me")
+}
+
+func (c cacheConfig) SetDefault(key string, value interface{}) {
+	panic("implement me")
+}
@@ -0,0 +1,101 @@
+package meta_cache
+
+import (
+	"fmt"
+	"strconv"
+	"strings"
+)
+
+type UidGidMapper struct {
+	uidMapper *IdMapper
+	gidMapper *IdMapper
+}
+
+type IdMapper struct {
+	localToFiler map[uint32]uint32
+	filerToLocal map[uint32]uint32
+}
+
+// UidGidMapper translates local uid/gid to filer uid/gid
+// The local storage always persists the same as the filer.
+// The local->filer translation happens when updating the filer first and later saving to meta_cache.
+// And filer->local happens when reading from the meta_cache.
+func NewUidGidMapper(uidPairsStr, gidPairStr string) (*UidGidMapper, error) {
+	uidMapper, err := newIdMapper(uidPairsStr)
+	if err != nil {
+		return nil, err
+	}
+	gidMapper, err := newIdMapper(gidPairStr)
+	if err != nil {
+		return nil, err
+	}
+
+	return &UidGidMapper{
+		uidMapper: uidMapper,
+		gidMapper: gidMapper,
+	}, nil
+}
+
+func (m *UidGidMapper) LocalToFiler(uid, gid uint32) (uint32, uint32) {
+	return m.uidMapper.LocalToFiler(uid), m.gidMapper.LocalToFiler(gid)
+}
+func (m *UidGidMapper) FilerToLocal(uid, gid uint32) (uint32, uint32) {
+	return m.uidMapper.FilerToLocal(uid), m.gidMapper.FilerToLocal(gid)
+}
+
+func (m *IdMapper) LocalToFiler(id uint32) uint32 {
+	value, found := m.localToFiler[id]
+	if found {
+		return value
+	}
+	return id
+}
+func (m *IdMapper) FilerToLocal(id uint32) uint32 {
+	value, found := m.filerToLocal[id]
+	if found {
+		return value
+	}
+	return id
+}
+
+func newIdMapper(pairsStr string) (*IdMapper, error) {
+
+	localToFiler, filerToLocal, err := parseUint32Pairs(pairsStr)
+	if err != nil {
+		return nil, err
+	}
+
+	return &IdMapper{
+		localToFiler: localToFiler,
+		filerToLocal: filerToLocal,
+	}, nil
+
+}
+
+func parseUint32Pairs(pairsStr string) (localToFiler, filerToLocal map[uint32]uint32, err error) {
+
+	if pairsStr == "" {
+		return
+	}
+
+	localToFiler = make(map[uint32]uint32)
+	filerToLocal = make(map[uint32]uint32)
+	for _, pairStr := range strings.Split(pairsStr, ",") {
+		pair := strings.Split(pairStr, ":")
+		localUidStr, filerUidStr := pair[0], pair[1]
+		localUid, localUidErr := strconv.Atoi(localUidStr)
+		if localUidErr != nil {
+			err = fmt.Errorf("failed to parse local %s: %v", localUidStr, localUidErr)
+			return
+		}
+		filerUid, filerUidErr := strconv.Atoi(filerUidStr)
+		if filerUidErr != nil {
+			err = fmt.Errorf("failed to parse remote %s: %v", filerUidStr, filerUidErr)
+			return
+		}
+		localToFiler[uint32(localUid)] = uint32(filerUid)
+		filerToLocal[uint32(filerUid)] = uint32(localUid)
+	}
+
+	return
+}
@@ -0,0 +1,160 @@
+package meta_cache
+
+import (
+	"context"
+	"github.com/chrislusf/seaweedfs/weed/filer"
+	"github.com/chrislusf/seaweedfs/weed/filer/leveldb"
+	"github.com/chrislusf/seaweedfs/weed/glog"
+	"github.com/chrislusf/seaweedfs/weed/pb/filer_pb"
+	"github.com/chrislusf/seaweedfs/weed/util"
+	"os"
+)
+
+// need to have logic similar to FilerStoreWrapper
+// e.g. fill fileId field for chunks
+
+type MetaCache struct {
+	localStore filer.VirtualFilerStore
+	// sync.RWMutex
+	uidGidMapper   *UidGidMapper
+	markCachedFn   func(fullpath util.FullPath)
+	isCachedFn     func(fullpath util.FullPath) bool
+	invalidateFunc func(fullpath util.FullPath, entry *filer_pb.Entry)
+}
+
+func NewMetaCache(dbFolder string, uidGidMapper *UidGidMapper, markCachedFn func(path util.FullPath), isCachedFn func(path util.FullPath) bool, invalidateFunc func(util.FullPath, *filer_pb.Entry)) *MetaCache {
+	return &MetaCache{
+		localStore:   openMetaStore(dbFolder),
+		markCachedFn: markCachedFn,
+		isCachedFn:   isCachedFn,
+		uidGidMapper: uidGidMapper,
+		invalidateFunc: func(fullpath util.FullPath, entry *filer_pb.Entry) {
+			invalidateFunc(fullpath, entry)
+		},
+	}
+}
+
+func openMetaStore(dbFolder string) filer.VirtualFilerStore {
+
+	os.RemoveAll(dbFolder)
+	os.MkdirAll(dbFolder, 0755)
+
+	store := &leveldb.LevelDBStore{}
+	config := &cacheConfig{
+		dir: dbFolder,
+	}
+
+	if err := store.Initialize(config, ""); err != nil {
+		glog.Fatalf("Failed to initialize metadata cache store for %s: %+v", store.GetName(), err)
+	}
+
+	return filer.NewFilerStoreWrapper(store)
+
+}
+
+func (mc *MetaCache) InsertEntry(ctx context.Context, entry *filer.Entry) error {
+	//mc.Lock()
+	//defer mc.Unlock()
+	return mc.doInsertEntry(ctx, entry)
+}
+
+func (mc *MetaCache) doInsertEntry(ctx context.Context, entry *filer.Entry) error {
+	return mc.localStore.InsertEntry(ctx, entry)
+}
+
+func (mc *MetaCache) AtomicUpdateEntryFromFiler(ctx context.Context, oldPath util.FullPath, newEntry *filer.Entry) error {
+	//mc.Lock()
+	//defer mc.Unlock()
+
+	oldDir, _ := oldPath.DirAndName()
+	if mc.isCachedFn(util.FullPath(oldDir)) {
+		if oldPath != "" {
+			if newEntry != nil && oldPath == newEntry.FullPath {
+				// skip the unnecessary deletion
+				// leave the update to the following InsertEntry operation
+			} else {
+				glog.V(3).Infof("DeleteEntry %s", oldPath)
+				if err := mc.localStore.DeleteEntry(ctx, oldPath); err != nil {
+					return err
+				}
+			}
+		}
+	} else {
+		// println("unknown old directory:", oldDir)
+	}
+
+	if newEntry != nil {
+		newDir, _ := newEntry.DirAndName()
+		if mc.isCachedFn(util.FullPath(newDir)) {
+			glog.V(3).Infof("InsertEntry %s/%s", newDir, newEntry.Name())
+			if err := mc.localStore.InsertEntry(ctx, newEntry); err != nil {
+				return err
+			}
+		}
+	}
+	return nil
+}
+
+func (mc *MetaCache) UpdateEntry(ctx context.Context, entry *filer.Entry) error {
+	//mc.Lock()
+	//defer mc.Unlock()
+	return mc.localStore.UpdateEntry(ctx, entry)
+}
+
+func (mc *MetaCache) FindEntry(ctx context.Context, fp util.FullPath) (entry *filer.Entry, err error) {
+	//mc.RLock()
+	//defer mc.RUnlock()
+	entry, err = mc.localStore.FindEntry(ctx, fp)
+	if err != nil {
+		return nil, err
+	}
+	mc.mapIdFromFilerToLocal(entry)
+	return
+}
+
+func (mc *MetaCache) DeleteEntry(ctx context.Context, fp util.FullPath) (err error) {
+	//mc.Lock()
+	//defer mc.Unlock()
+	return mc.localStore.DeleteEntry(ctx, fp)
+}
+func (mc *MetaCache) DeleteFolderChildren(ctx context.Context, fp util.FullPath) (err error) {
+	//mc.Lock()
+	//defer mc.Unlock()
+	return mc.localStore.DeleteFolderChildren(ctx, fp)
+}
+
+func (mc *MetaCache) ListDirectoryEntries(ctx context.Context, dirPath util.FullPath, startFileName string, includeStartFile bool, limit int64, eachEntryFunc filer.ListEachEntryFunc) error {
+	//mc.RLock()
+	//defer mc.RUnlock()
+
+	if !mc.isCachedFn(dirPath) {
+		// if this request comes after renaming, it should be fine
+		glog.Warningf("unsynchronized dir: %v", dirPath)
+	}
+
+	_, err := mc.localStore.ListDirectoryEntries(ctx, dirPath, startFileName, includeStartFile, limit, func(entry *filer.Entry) bool {
+		mc.mapIdFromFilerToLocal(entry)
+		return eachEntryFunc(entry)
+	})
+	if err != nil {
+		return err
+	}
+	return err
+}
+
+func (mc *MetaCache) Shutdown() {
+	//mc.Lock()
+	//defer mc.Unlock()
+	mc.localStore.Shutdown()
+}
+
+func (mc *MetaCache) mapIdFromFilerToLocal(entry *filer.Entry) {
+	entry.Attr.Uid, entry.Attr.Gid = mc.uidGidMapper.FilerToLocal(entry.Attr.Uid, entry.Attr.Gid)
+}
+
+func (mc *MetaCache) Debug() {
+	if debuggable, ok := mc.localStore.(filer.Debuggable); ok {
+		println("start debugging")
+		debuggable.Debug(os.Stderr)
+	}
+}
@@ -0,0 +1,67 @@
+package meta_cache
+
+import (
+	"context"
+	"fmt"
+
+	"github.com/chrislusf/seaweedfs/weed/filer"
+	"github.com/chrislusf/seaweedfs/weed/glog"
+	"github.com/chrislusf/seaweedfs/weed/pb/filer_pb"
+	"github.com/chrislusf/seaweedfs/weed/util"
+)
+
+func EnsureVisited(mc *MetaCache, client filer_pb.FilerClient, dirPath util.FullPath) error {
+
+	for {
+
+		// the directory children are already cached
+		// so no need for this and upper directories
+		if mc.isCachedFn(dirPath) {
+			return nil
+		}
+
+		if err := doEnsureVisited(mc, client, dirPath); err != nil {
+			return err
+		}
+
+		// continue to parent directory
+		if dirPath != "/" {
+			parent, _ := dirPath.DirAndName()
+			dirPath = util.FullPath(parent)
+		} else {
+			break
+		}
+	}
+
+	return nil
+
+}
+
+func doEnsureVisited(mc *MetaCache, client filer_pb.FilerClient, path util.FullPath) error {
+
+	glog.V(4).Infof("ReadDirAllEntries %s ...", path)
+
+	err := util.Retry("ReadDirAllEntries", func() error {
+		return filer_pb.ReadDirAllEntries(client, path, "", func(pbEntry *filer_pb.Entry, isLast bool) error {
+			entry := filer.FromPbEntry(string(path), pbEntry)
+			if IsHiddenSystemEntry(string(path), entry.Name()) {
+				return nil
+			}
+			if err := mc.doInsertEntry(context.Background(), entry); err != nil {
+				glog.V(0).Infof("read %s: %v", entry.FullPath, err)
+				return err
+			}
+			return nil
+		})
+	})
+
+	if err != nil {
+		err = fmt.Errorf("list %s: %v", path, err)
+	}
+	mc.markCachedFn(path)
+	return err
+}
+
+func IsHiddenSystemEntry(dir, name string) bool {
+	return dir == "/" && (name == "topics" || name == "etc")
+}
@@ -0,0 +1,68 @@
+package meta_cache
+
+import (
+	"context"
+	"github.com/chrislusf/seaweedfs/weed/filer"
+	"github.com/chrislusf/seaweedfs/weed/glog"
+	"github.com/chrislusf/seaweedfs/weed/pb"
+	"github.com/chrislusf/seaweedfs/weed/pb/filer_pb"
+	"github.com/chrislusf/seaweedfs/weed/util"
+)
+
+func SubscribeMetaEvents(mc *MetaCache, selfSignature int32, client filer_pb.FilerClient, dir string, lastTsNs int64) error {
+
+	processEventFn := func(resp *filer_pb.SubscribeMetadataResponse) error {
+		message := resp.EventNotification
+
+		for _, sig := range message.Signatures {
+			if sig == selfSignature && selfSignature != 0 {
+				return nil
+			}
+		}
+
+		dir := resp.Directory
+		var oldPath util.FullPath
+		var newEntry *filer.Entry
+		if message.OldEntry != nil {
+			oldPath = util.NewFullPath(dir, message.OldEntry.Name)
+			glog.V(4).Infof("deleting %v", oldPath)
+		}
+
+		if message.NewEntry != nil {
+			if message.NewParentPath != "" {
+				dir = message.NewParentPath
+			}
+			key := util.NewFullPath(dir, message.NewEntry.Name)
+			glog.V(4).Infof("creating %v", key)
+			newEntry = filer.FromPbEntry(dir, message.NewEntry)
+		}
+		err := mc.AtomicUpdateEntryFromFiler(context.Background(), oldPath, newEntry)
+		if err == nil {
+			if message.OldEntry != nil && message.NewEntry != nil {
+				oldKey := util.NewFullPath(resp.Directory, message.OldEntry.Name)
+				mc.invalidateFunc(oldKey, message.OldEntry)
+				if message.OldEntry.Name != message.NewEntry.Name {
+					newKey := util.NewFullPath(dir, message.NewEntry.Name)
+					mc.invalidateFunc(newKey, message.NewEntry)
+				}
+			} else if message.OldEntry == nil && message.NewEntry != nil {
+				// no need to invaalidate
+			} else if message.OldEntry != nil && message.NewEntry == nil {
+				oldKey := util.NewFullPath(resp.Directory, message.OldEntry.Name)
+				mc.invalidateFunc(oldKey, message.OldEntry)
+			}
+		}
+
+		return err
+
+	}
+
+	util.RetryForever("followMetaUpdates", func() error {
+		return pb.WithFilerClientFollowMetadata(client, "mount", selfSignature, dir, &lastTsNs, selfSignature, processEventFn, true)
+	}, func(err error) bool {
+		glog.Errorf("follow metadata updates: %v", err)
+		return true
+	})
+
+	return nil
+}
@@ -0,0 +1,95 @@
+package mount
+
+import (
+	"github.com/chrislusf/seaweedfs/weed/filesys/page_writer"
+	"github.com/chrislusf/seaweedfs/weed/glog"
+)
+
+type PageWriter struct {
+	fh          *FileHandle
+	collection  string
+	replication string
+	chunkSize   int64
+
+	randomWriter page_writer.DirtyPages
+}
+
+var (
+	_ = page_writer.DirtyPages(&PageWriter{})
+)
+
+func newPageWriter(fh *FileHandle, chunkSize int64) *PageWriter {
+	pw := &PageWriter{
+		fh:           fh,
+		chunkSize:    chunkSize,
+		randomWriter: newMemoryChunkPages(fh, chunkSize),
+		// randomWriter: newTempFileDirtyPages(fh.f, chunkSize),
+	}
+	return pw
+}
+
+func (pw *PageWriter) AddPage(offset int64, data []byte) {
+
+	glog.V(4).Infof("%v AddPage [%d, %d)", pw.fh.fh, offset, offset+int64(len(data)))
+
+	chunkIndex := offset / pw.chunkSize
+	for i := chunkIndex; len(data) > 0; i++ {
+		writeSize := min(int64(len(data)), (i+1)*pw.chunkSize-offset)
+		pw.addToOneChunk(i, offset, data[:writeSize])
+		offset += writeSize
+		data = data[writeSize:]
+	}
+}
+
+func (pw *PageWriter) addToOneChunk(chunkIndex, offset int64, data []byte) {
+	pw.randomWriter.AddPage(offset, data)
+}
+
+func (pw *PageWriter) FlushData() error {
+	return pw.randomWriter.FlushData()
+}
+
+func (pw *PageWriter) ReadDirtyDataAt(data []byte, offset int64) (maxStop int64) {
+	glog.V(4).Infof("ReadDirtyDataAt %v [%d, %d)", pw.fh.fh, offset, offset+int64(len(data)))
+
+	chunkIndex := offset / pw.chunkSize
+	for i := chunkIndex; len(data) > 0; i++ {
+		readSize := min(int64(len(data)), (i+1)*pw.chunkSize-offset)
+
+		maxStop = pw.randomWriter.ReadDirtyDataAt(data[:readSize], offset)
+
+		offset += readSize
+		data = data[readSize:]
+	}
+
+	return
+}
+
+func (pw *PageWriter) GetStorageOptions() (collection, replication string) {
+	return pw.randomWriter.GetStorageOptions()
+}
+
+func (pw *PageWriter) LockForRead(startOffset, stopOffset int64) {
+	pw.randomWriter.LockForRead(startOffset, stopOffset)
+}
+
+func (pw *PageWriter) UnlockForRead(startOffset, stopOffset int64) {
+	pw.randomWriter.UnlockForRead(startOffset, stopOffset)
+}
+
+func (pw *PageWriter) Destroy() {
+	pw.randomWriter.Destroy()
+}
+
+func max(x, y int64) int64 {
+	if x > y {
+		return x
+	}
+	return y
+}
+func min(x, y int64) int64 {
+	if x < y {
+		return x
+	}
+	return y
+}
@@ -0,0 +1,115 @@
+package page_writer
+
+import "math"
+
+// ChunkWrittenInterval mark one written interval within one page chunk
+type ChunkWrittenInterval struct {
+	StartOffset int64
+	stopOffset  int64
+	prev        *ChunkWrittenInterval
+	next        *ChunkWrittenInterval
+}
+
+func (interval *ChunkWrittenInterval) Size() int64 {
+	return interval.stopOffset - interval.StartOffset
+}
+
+func (interval *ChunkWrittenInterval) isComplete(chunkSize int64) bool {
+	return interval.stopOffset-interval.StartOffset == chunkSize
+}
+
+// ChunkWrittenIntervalList mark written intervals within one page chunk
+type ChunkWrittenIntervalList struct {
+	head *ChunkWrittenInterval
+	tail *ChunkWrittenInterval
+}
+
+func newChunkWrittenIntervalList() *ChunkWrittenIntervalList {
+	list := &ChunkWrittenIntervalList{
+		head: &ChunkWrittenInterval{
+			StartOffset: -1,
+			stopOffset:  -1,
+		},
+		tail: &ChunkWrittenInterval{
+			StartOffset: math.MaxInt64,
+			stopOffset:  math.MaxInt64,
+		},
+	}
+	list.head.next = list.tail
+	list.tail.prev = list.head
+	return list
+}
+
+func (list *ChunkWrittenIntervalList) MarkWritten(startOffset, stopOffset int64) {
+	interval := &ChunkWrittenInterval{
+		StartOffset: startOffset,
+		stopOffset:  stopOffset,
+	}
+	list.addInterval(interval)
+}
+
+func (list *ChunkWrittenIntervalList) IsComplete(chunkSize int64) bool {
+	return list.size() == 1 && list.head.next.isComplete(chunkSize)
+}
+func (list *ChunkWrittenIntervalList) WrittenSize() (writtenByteCount int64) {
+	for t := list.head; t != nil; t = t.next {
+		writtenByteCount += t.Size()
+	}
+	return
+}
+
+func (list *ChunkWrittenIntervalList) addInterval(interval *ChunkWrittenInterval) {
+
+	p := list.head
+	for ; p.next != nil && p.next.StartOffset <= interval.StartOffset; p = p.next {
+	}
+	q := list.tail
+	for ; q.prev != nil && q.prev.stopOffset >= interval.stopOffset; q = q.prev {
+	}
+
+	if interval.StartOffset <= p.stopOffset && q.StartOffset <= interval.stopOffset {
+		// merge p and q together
+		p.stopOffset = q.stopOffset
+		unlinkNodesBetween(p, q.next)
+		return
+	}
+	if interval.StartOffset <= p.stopOffset {
+		// merge new interval into p
+		p.stopOffset = interval.stopOffset
+		unlinkNodesBetween(p, q)
+		return
+	}
+	if q.StartOffset <= interval.stopOffset {
+		// merge new interval into q
+		q.StartOffset = interval.StartOffset
+		unlinkNodesBetween(p, q)
+		return
+	}
+
+	// add the new interval between p and q
+	unlinkNodesBetween(p, q)
+	p.next = interval
+	interval.prev = p
+	q.prev = interval
+	interval.next = q
+
+}
+
+// unlinkNodesBetween remove all nodes after start and before stop, exclusive
+func unlinkNodesBetween(start *ChunkWrittenInterval, stop *ChunkWrittenInterval) {
+	if start.next == stop {
+		return
+	}
+	start.next.prev = nil
+	start.next = stop
+	stop.prev.next = nil
+	stop.prev = start
+}
+
+func (list *ChunkWrittenIntervalList) size() int {
+	var count int
+	for t := list.head; t != nil; t = t.next {
+		count++
+	}
+	return count - 2
+}
@@ -0,0 +1,49 @@
+package page_writer
+
+import (
+	"github.com/stretchr/testify/assert"
+	"testing"
+)
+
+func Test_PageChunkWrittenIntervalList(t *testing.T) {
+	list := newChunkWrittenIntervalList()
+
+	assert.Equal(t, 0, list.size(), "empty list")
+
+	list.MarkWritten(0, 5)
+	assert.Equal(t, 1, list.size(), "one interval")
+
+	list.MarkWritten(0, 5)
+	assert.Equal(t, 1, list.size(), "duplicated interval2")
+
+	list.MarkWritten(95, 100)
+	assert.Equal(t, 2, list.size(), "two intervals")
+
+	list.MarkWritten(50, 60)
+	assert.Equal(t, 3, list.size(), "three intervals")
+
+	list.MarkWritten(50, 55)
+	assert.Equal(t, 3, list.size(), "three intervals merge")
+
+	list.MarkWritten(40, 50)
+	assert.Equal(t, 3, list.size(), "three intervals grow forward")
+
+	list.MarkWritten(50, 65)
+	assert.Equal(t, 3, list.size(), "three intervals grow backward")
+
+	list.MarkWritten(70, 80)
+	assert.Equal(t, 4, list.size(), "four intervals")
+
+	list.MarkWritten(60, 70)
+	assert.Equal(t, 3, list.size(), "three intervals merged")
+
+	list.MarkWritten(59, 71)
+	assert.Equal(t, 3, list.size(), "covered three intervals")
+
+	list.MarkWritten(5, 59)
+	assert.Equal(t, 2, list.size(), "covered two intervals")
+
+	list.MarkWritten(70, 99)
+	assert.Equal(t, 1, list.size(), "covered one intervals")
+
+}
@@ -0,0 +1,30 @@
+package page_writer
+
+type DirtyPages interface {
+	AddPage(offset int64, data []byte)
+	FlushData() error
+	ReadDirtyDataAt(data []byte, startOffset int64) (maxStop int64)
+	GetStorageOptions() (collection, replication string)
+	Destroy()
+	LockForRead(startOffset, stopOffset int64)
+	UnlockForRead(startOffset, stopOffset int64)
+}
+
+func max(x, y int64) int64 {
+	if x > y {
+		return x
+	}
+	return y
+}
+func min(x, y int64) int64 {
+	if x < y {
+		return x
+	}
+	return y
+}
+func minInt(x, y int) int {
+	if x < y {
+		return x
+	}
+	return y
+}
@@ -0,0 +1,16 @@
+package page_writer
+
+import (
+	"io"
+)
+
+type SaveToStorageFunc func(reader io.Reader, offset int64, size int64, cleanupFn func())
+
+type PageChunk interface {
+	FreeResource()
+	WriteDataAt(src []byte, offset int64) (n int)
+	ReadDataAt(p []byte, off int64) (maxStop int64)
+	IsComplete() bool
+	WrittenSize() int64
+	SaveContent(saveFn SaveToStorageFunc)
+}
@@ -0,0 +1,69 @@
+package page_writer
+
+import (
+	"github.com/chrislusf/seaweedfs/weed/util"
+	"github.com/chrislusf/seaweedfs/weed/util/mem"
+)
+
+var (
+	_ = PageChunk(&MemChunk{})
+)
+
+type MemChunk struct {
+	buf             []byte
+	usage           *ChunkWrittenIntervalList
+	chunkSize       int64
+	logicChunkIndex LogicChunkIndex
+}
+
+func NewMemChunk(logicChunkIndex LogicChunkIndex, chunkSize int64) *MemChunk {
+	return &MemChunk{
+		logicChunkIndex: logicChunkIndex,
+		chunkSize:       chunkSize,
+		buf:             mem.Allocate(int(chunkSize)),
+		usage:           newChunkWrittenIntervalList(),
+	}
+}
+
+func (mc *MemChunk) FreeResource() {
+	mem.Free(mc.buf)
+}
+
+func (mc *MemChunk) WriteDataAt(src []byte, offset int64) (n int) {
+	innerOffset := offset % mc.chunkSize
+	n = copy(mc.buf[innerOffset:], src)
+	mc.usage.MarkWritten(innerOffset, innerOffset+int64(n))
+	return
+}
+
+func (mc *MemChunk) ReadDataAt(p []byte, off int64) (maxStop int64) {
+	memChunkBaseOffset := int64(mc.logicChunkIndex) * mc.chunkSize
+	for t := mc.usage.head.next; t != mc.usage.tail; t = t.next {
+		logicStart := max(off, int64(mc.logicChunkIndex)*mc.chunkSize+t.StartOffset)
+		logicStop := min(off+int64(len(p)), memChunkBaseOffset+t.stopOffset)
+		if logicStart < logicStop {
+			copy(p[logicStart-off:logicStop-off], mc.buf[logicStart-memChunkBaseOffset:logicStop-memChunkBaseOffset])
+			maxStop = max(maxStop, logicStop)
+		}
+	}
+	return
+}
+
+func (mc *MemChunk) IsComplete() bool {
+	return mc.usage.IsComplete(mc.chunkSize)
+}
+
+func (mc *MemChunk) WrittenSize() int64 {
+	return mc.usage.WrittenSize()
+}
+
+func (mc *MemChunk) SaveContent(saveFn SaveToStorageFunc) {
+	if saveFn == nil {
+		return
+	}
+	for t := mc.usage.head.next; t != mc.usage.tail; t = t.next {
+		reader := util.NewBytesReader(mc.buf[t.StartOffset:t.stopOffset])
+		saveFn(reader, int64(mc.logicChunkIndex)*mc.chunkSize+t.StartOffset, t.Size(), func() {
+		})
+	}
+}
@@ -0,0 +1,121 @@
+package page_writer
+
+import (
+	"github.com/chrislusf/seaweedfs/weed/glog"
+	"github.com/chrislusf/seaweedfs/weed/util"
+	"github.com/chrislusf/seaweedfs/weed/util/mem"
+	"os"
+)
+
+var (
+	_ = PageChunk(&SwapFileChunk{})
+)
+
+type ActualChunkIndex int
+
+type SwapFile struct {
+	dir                     string
+	file                    *os.File
+	logicToActualChunkIndex map[LogicChunkIndex]ActualChunkIndex
+	chunkSize               int64
+}
+
+type SwapFileChunk struct {
+	swapfile         *SwapFile
+	usage            *ChunkWrittenIntervalList
+	logicChunkIndex  LogicChunkIndex
+	actualChunkIndex ActualChunkIndex
+}
+
+func NewSwapFile(dir string, chunkSize int64) *SwapFile {
+	return &SwapFile{
+		dir:                     dir,
+		file:                    nil,
+		logicToActualChunkIndex: make(map[LogicChunkIndex]ActualChunkIndex),
+		chunkSize:               chunkSize,
+	}
+}
+func (sf *SwapFile) FreeResource() {
+	if sf.file != nil {
+		sf.file.Close()
+		os.Remove(sf.file.Name())
+	}
+}
+
+func (sf *SwapFile) NewTempFileChunk(logicChunkIndex LogicChunkIndex) (tc *SwapFileChunk) {
+	if sf.file == nil {
+		var err error
+		sf.file, err = os.CreateTemp(sf.dir, "")
+		if err != nil {
+			glog.Errorf("create swap file: %v", err)
+			return nil
+		}
+	}
+	actualChunkIndex, found := sf.logicToActualChunkIndex[logicChunkIndex]
+	if !found {
+		actualChunkIndex = ActualChunkIndex(len(sf.logicToActualChunkIndex))
+		sf.logicToActualChunkIndex[logicChunkIndex] = actualChunkIndex
+	}
+
+	return &SwapFileChunk{
+		swapfile:         sf,
+		usage:            newChunkWrittenIntervalList(),
+		logicChunkIndex:  logicChunkIndex,
+		actualChunkIndex: actualChunkIndex,
+	}
+}
+
+func (sc *SwapFileChunk) FreeResource() {
+}
+
+func (sc *SwapFileChunk) WriteDataAt(src []byte, offset int64) (n int) {
+	innerOffset := offset % sc.swapfile.chunkSize
+	var err error
+	n, err = sc.swapfile.file.WriteAt(src, int64(sc.actualChunkIndex)*sc.swapfile.chunkSize+innerOffset)
+	if err == nil {
+		sc.usage.MarkWritten(innerOffset, innerOffset+int64(n))
+	} else {
+		glog.Errorf("failed to write swap file %s: %v", sc.swapfile.file.Name(), err)
+	}
+	return
+}
+
+func (sc *SwapFileChunk) ReadDataAt(p []byte, off int64) (maxStop int64) {
+	chunkStartOffset := int64(sc.logicChunkIndex) * sc.swapfile.chunkSize
+	for t := sc.usage.head.next; t != sc.usage.tail; t = t.next {
+		logicStart := max(off, chunkStartOffset+t.StartOffset)
+		logicStop := min(off+int64(len(p)), chunkStartOffset+t.stopOffset)
+		if logicStart < logicStop {
+			actualStart := logicStart - chunkStartOffset + int64(sc.actualChunkIndex)*sc.swapfile.chunkSize
+			if _, err := sc.swapfile.file.ReadAt(p[logicStart-off:logicStop-off], actualStart); err != nil {
+				glog.Errorf("failed to reading swap file %s: %v", sc.swapfile.file.Name(), err)
+				break
+			}
+			maxStop = max(maxStop, logicStop)
+		}
+	}
+	return
+}
+
+func (sc *SwapFileChunk) IsComplete() bool {
+	return sc.usage.IsComplete(sc.swapfile.chunkSize)
+}
+
+func (sc *SwapFileChunk) WrittenSize() int64 {
+	return sc.usage.WrittenSize()
+}
+
+func (sc *SwapFileChunk) SaveContent(saveFn SaveToStorageFunc) {
+	if saveFn == nil {
+		return
+	}
+	for t := sc.usage.head.next; t != sc.usage.tail; t = t.next {
+		data := mem.Allocate(int(t.Size()))
+		sc.swapfile.file.ReadAt(data, t.StartOffset+int64(sc.actualChunkIndex)*sc.swapfile.chunkSize)
+		reader := util.NewBytesReader(data)
+		saveFn(reader, int64(sc.logicChunkIndex)*sc.swapfile.chunkSize+t.StartOffset, t.Size(), func() {
+		})
+		mem.Free(data)
+	}
+	sc.usage = newChunkWrittenIntervalList()
+}
@@ -0,0 +1,182 @@
+package page_writer
+
+import (
+	"fmt"
+	"github.com/chrislusf/seaweedfs/weed/glog"
+	"github.com/chrislusf/seaweedfs/weed/util"
+	"sync"
+	"sync/atomic"
+	"time"
+)
+
+type LogicChunkIndex int
+
+type UploadPipeline struct {
+	filepath             util.FullPath
+	ChunkSize            int64
+	writableChunks       map[LogicChunkIndex]PageChunk
+	writableChunksLock   sync.Mutex
+	sealedChunks         map[LogicChunkIndex]*SealedChunk
+	sealedChunksLock     sync.Mutex
+	uploaders            *util.LimitedConcurrentExecutor
+	uploaderCount        int32
+	uploaderCountCond    *sync.Cond
+	saveToStorageFn      SaveToStorageFunc
+	activeReadChunks     map[LogicChunkIndex]int
+	activeReadChunksLock sync.Mutex
+	bufferChunkLimit     int
+}
+
+type SealedChunk struct {
+	chunk            PageChunk
+	referenceCounter int // track uploading or reading processes
+}
+
+func (sc *SealedChunk) FreeReference(messageOnFree string) {
+	sc.referenceCounter--
+	if sc.referenceCounter == 0 {
+		glog.V(4).Infof("Free sealed chunk: %s", messageOnFree)
+		sc.chunk.FreeResource()
+	}
+}
+
+func NewUploadPipeline(writers *util.LimitedConcurrentExecutor, chunkSize int64, saveToStorageFn SaveToStorageFunc, bufferChunkLimit int) *UploadPipeline {
+	return &UploadPipeline{
+		ChunkSize:         chunkSize,
+		writableChunks:    make(map[LogicChunkIndex]PageChunk),
+		sealedChunks:      make(map[LogicChunkIndex]*SealedChunk),
+		uploaders:         writers,
+		uploaderCountCond: sync.NewCond(&sync.Mutex{}),
+		saveToStorageFn:   saveToStorageFn,
+		activeReadChunks:  make(map[LogicChunkIndex]int),
+		bufferChunkLimit:  bufferChunkLimit,
+	}
+}
+
+func (up *UploadPipeline) SaveDataAt(p []byte, off int64) (n int) {
+	up.writableChunksLock.Lock()
+	defer up.writableChunksLock.Unlock()
+
+	logicChunkIndex := LogicChunkIndex(off / up.ChunkSize)
+
+	memChunk, found := up.writableChunks[logicChunkIndex]
+	if !found {
+		if len(up.writableChunks) < up.bufferChunkLimit {
+			memChunk = NewMemChunk(logicChunkIndex, up.ChunkSize)
+		} else {
+			fullestChunkIndex, fullness := LogicChunkIndex(-1), int64(0)
+			for lci, mc := range up.writableChunks {
+				chunkFullness := mc.WrittenSize()
+				if fullness < chunkFullness {
+					fullestChunkIndex = lci
+					fullness = chunkFullness
+				}
+			}
+			up.moveToSealed(up.writableChunks[fullestChunkIndex], fullestChunkIndex)
+			delete(up.writableChunks, fullestChunkIndex)
+			fmt.Printf("flush chunk %d with %d bytes written", logicChunkIndex, fullness)
+			memChunk = NewMemChunk(logicChunkIndex, up.ChunkSize)
+		}
+		up.writableChunks[logicChunkIndex] = memChunk
+	}
+	n = memChunk.WriteDataAt(p, off)
+	up.maybeMoveToSealed(memChunk, logicChunkIndex)
+
+	return
+}
+
+func (up *UploadPipeline) MaybeReadDataAt(p []byte, off int64) (maxStop int64) {
+	logicChunkIndex := LogicChunkIndex(off / up.ChunkSize)
+
+	// read from sealed chunks first
+	up.sealedChunksLock.Lock()
+	sealedChunk, found := up.sealedChunks[logicChunkIndex]
+	if found {
+		sealedChunk.referenceCounter++
+	}
+	up.sealedChunksLock.Unlock()
+	if found {
+		maxStop = sealedChunk.chunk.ReadDataAt(p, off)
+		glog.V(4).Infof("%s read sealed memchunk [%d,%d)", up.filepath, off, maxStop)
+		sealedChunk.FreeReference(fmt.Sprintf("%s finish reading chunk %d", up.filepath, logicChunkIndex))
+	}
+
+	// read from writable chunks last
+	up.writableChunksLock.Lock()
+	defer up.writableChunksLock.Unlock()
+	writableChunk, found := up.writableChunks[logicChunkIndex]
+	if !found {
+		return
+	}
+	writableMaxStop := writableChunk.ReadDataAt(p, off)
+	glog.V(4).Infof("%s read writable memchunk [%d,%d)", up.filepath, off, writableMaxStop)
+	maxStop = max(maxStop, writableMaxStop)
+
+	return
+}
+
+func (up *UploadPipeline) FlushAll() {
+	up.writableChunksLock.Lock()
+	defer up.writableChunksLock.Unlock()
+
+	for logicChunkIndex, memChunk := range up.writableChunks {
+		up.moveToSealed(memChunk, logicChunkIndex)
+	}
+
+	up.waitForCurrentWritersToComplete()
+}
+
+func (up *UploadPipeline) maybeMoveToSealed(memChunk PageChunk, logicChunkIndex LogicChunkIndex) {
+	if memChunk.IsComplete() {
+		up.moveToSealed(memChunk, logicChunkIndex)
+	}
+}
+
+func (up *UploadPipeline) moveToSealed(memChunk PageChunk, logicChunkIndex LogicChunkIndex) {
+	atomic.AddInt32(&up.uploaderCount, 1)
+	glog.V(4).Infof("%s uploaderCount %d ++> %d", up.filepath, up.uploaderCount-1, up.uploaderCount)
+
+	up.sealedChunksLock.Lock()
+
+	if oldMemChunk, found := up.sealedChunks[logicChunkIndex]; found {
+		oldMemChunk.FreeReference(fmt.Sprintf("%s replace chunk %d", up.filepath, logicChunkIndex))
+	}
+	sealedChunk := &SealedChunk{
+		chunk:            memChunk,
+		referenceCounter: 1, // default 1 is for uploading process
+	}
+	up.sealedChunks[logicChunkIndex] = sealedChunk
+	delete(up.writableChunks, logicChunkIndex)
+
+	up.sealedChunksLock.Unlock()
+
+	up.uploaders.Execute(func() {
+		// first add to the file chunks
+		sealedChunk.chunk.SaveContent(up.saveToStorageFn)
+
+		// notify waiting process
+		atomic.AddInt32(&up.uploaderCount, -1)
+		glog.V(4).Infof("%s uploaderCount %d --> %d", up.filepath, up.uploaderCount+1, up.uploaderCount)
+		// Lock and Unlock are not required,
+		// but it may signal multiple times during one wakeup,
+		// and the waiting goroutine may miss some of them!
+		up.uploaderCountCond.L.Lock()
+		up.uploaderCountCond.Broadcast()
+		up.uploaderCountCond.L.Unlock()
+
+		// wait for readers
+		for up.IsLocked(logicChunkIndex) {
+			time.Sleep(59 * time.Millisecond)
+		}
+
+		// then remove from sealed chunks
+		up.sealedChunksLock.Lock()
+		defer up.sealedChunksLock.Unlock()
+		delete(up.sealedChunks, logicChunkIndex)
+		sealedChunk.FreeReference(fmt.Sprintf("%s finished uploading chunk %d", up.filepath, logicChunkIndex))
+
+	})
+}
+
+func (up *UploadPipeline) Shutdown() {
+}
@@ -0,0 +1,63 @@
+package page_writer
+
+import (
+	"sync/atomic"
+)
+
+func (up *UploadPipeline) LockForRead(startOffset, stopOffset int64) {
+	startLogicChunkIndex := LogicChunkIndex(startOffset / up.ChunkSize)
+	stopLogicChunkIndex := LogicChunkIndex(stopOffset / up.ChunkSize)
+	if stopOffset%up.ChunkSize > 0 {
+		stopLogicChunkIndex += 1
+	}
+	up.activeReadChunksLock.Lock()
+	defer up.activeReadChunksLock.Unlock()
+	for i := startLogicChunkIndex; i < stopLogicChunkIndex; i++ {
+		if count, found := up.activeReadChunks[i]; found {
+			up.activeReadChunks[i] = count + 1
+		} else {
+			up.activeReadChunks[i] = 1
+		}
+	}
+}
+
+func (up *UploadPipeline) UnlockForRead(startOffset, stopOffset int64) {
+	startLogicChunkIndex := LogicChunkIndex(startOffset / up.ChunkSize)
+	stopLogicChunkIndex := LogicChunkIndex(stopOffset / up.ChunkSize)
+	if stopOffset%up.ChunkSize > 0 {
+		stopLogicChunkIndex += 1
+	}
+	up.activeReadChunksLock.Lock()
+	defer up.activeReadChunksLock.Unlock()
+	for i := startLogicChunkIndex; i < stopLogicChunkIndex; i++ {
+		if count, found := up.activeReadChunks[i]; found {
+			if count == 1 {
+				delete(up.activeReadChunks, i)
+			} else {
+				up.activeReadChunks[i] = count - 1
+			}
+		}
+	}
+}
+
+func (up *UploadPipeline) IsLocked(logicChunkIndex LogicChunkIndex) bool {
+	up.activeReadChunksLock.Lock()
+	defer up.activeReadChunksLock.Unlock()
+	if count, found := up.activeReadChunks[logicChunkIndex]; found {
+		return count > 0
+	}
+	return false
+}
+
+func (up *UploadPipeline) waitForCurrentWritersToComplete() {
+	up.uploaderCountCond.L.Lock()
+	t := int32(100)
+	for {
+		t = atomic.LoadInt32(&up.uploaderCount)
+		if t <= 0 {
+			break
+		}
+		up.uploaderCountCond.Wait()
+	}
+	up.uploaderCountCond.L.Unlock()
+}
@@ -0,0 +1,47 @@
+package page_writer
+
+import (
+	"github.com/chrislusf/seaweedfs/weed/util"
+	"testing"
+)
+
+func TestUploadPipeline(t *testing.T) {
+
+	uploadPipeline := NewUploadPipeline(nil, 2*1024*1024, nil, 16)
+
+	writeRange(uploadPipeline, 0, 131072)
+	writeRange(uploadPipeline, 131072, 262144)
+	writeRange(uploadPipeline, 262144, 1025536)
+
+	confirmRange(t, uploadPipeline, 0, 1025536)
+
+	writeRange(uploadPipeline, 1025536, 1296896)
+
+	confirmRange(t, uploadPipeline, 1025536, 1296896)
+
+	writeRange(uploadPipeline, 1296896, 2162688)
+
+	confirmRange(t, uploadPipeline, 1296896, 2162688)
+
+	confirmRange(t, uploadPipeline, 1296896, 2162688)
+}
+
+// startOff and stopOff must be divided by 4
+func writeRange(uploadPipeline *UploadPipeline, startOff, stopOff int64) {
+	p := make([]byte, 4)
+	for i := startOff / 4; i < stopOff/4; i += 4 {
+		util.Uint32toBytes(p, uint32(i))
+		uploadPipeline.SaveDataAt(p, i)
+	}
+}
+
+func confirmRange(t *testing.T, uploadPipeline *UploadPipeline, startOff, stopOff int64) {
+	p := make([]byte, 4)
+	for i := startOff; i < stopOff/4; i += 4 {
+		uploadPipeline.MaybeReadDataAt(p, i)
+		x := util.BytesToUint32(p)
+		if x != uint32(i) {
+			t.Errorf("expecting %d found %d at offset [%d,%d)", i, x, i, i+4)
+		}
+	}
+}
@@ -0,0 +1,6 @@
+package unmount
+
+// Unmount tries to unmount the filesystem mounted at dir.
+func Unmount(dir string) error {
+	return unmount(dir)
+}
@@ -0,0 +1,21 @@
+package unmount
+
+import (
+	"bytes"
+	"errors"
+	"os/exec"
+)
+
+func unmount(dir string) error {
+	cmd := exec.Command("fusermount", "-u", dir)
+	output, err := cmd.CombinedOutput()
+	if err != nil {
+		if len(output) > 0 {
+			output = bytes.TrimRight(output, "\n")
+			msg := err.Error() + ": " + string(output)
+			err = errors.New(msg)
+		}
+		return err
+	}
+	return nil
+}
@@ -0,0 +1,18 @@
+//go:build !linux
+// +build !linux
+
+package unmount
+
+import (
+	"os"
+	"syscall"
+)
+
+func unmount(dir string) error {
+	err := syscall.Unmount(dir, 0)
+	if err != nil {
+		err = &os.PathError{Op: "unmount", Path: dir, Err: err}
+		return err
+	}
+	return nil
+}
--- a/Show More
+++ b/Show More