Uploaded image for project: 'Hadoop HDFS'
  1. Hadoop HDFS
  2. HDFS-11056

Concurrent append and read operations lead to checksum error

VotersStop watchingWatchersCreate sub-taskLinkCloneUpdate Comment AuthorReplace String in CommentUpdate Comment VisibilityDelete Comments
    XMLWordPrintableJSON

Details

    • Reviewed
    • Load last partial chunk checksum properly into memory when converting a finalized/temporary replica to rbw replica. This ensures concurrent reader reads the correct checksum that matches the data before the update.

    Description

      If there are two clients, one of them open-append-close a file continuously, while the other open-read-close the same file continuously, the reader eventually gets a checksum error in the data read.

      On my local Mac, it takes a few minutes to produce the error. This happens to httpfs clients, but there's no reason not believe this happens to any append clients.

      I have a unit test that demonstrates the checksum error. Will attach later.

      Relevant log:

      2016-10-25 15:34:45,153 INFO audit - allowed=true ugi=weichiu (auth:SIMPLE) ip=/127.0.0.1 cmd=open src=/tmp/bar.txt dst=null perm=null proto=rpc
      2016-10-25 15:34:45,155 INFO DataNode - Receiving BP-837130339-172.16.1.88-1477434851452:blk_1073741825_1182 src: /127.0.0.1:51130 dest: /127.0.0.1:50131
      2016-10-25 15:34:45,155 INFO FsDatasetImpl - Appending to FinalizedReplica, blk_1073741825_1182, FINALIZED
      getNumBytes() = 182
      getBytesOnDisk() = 182
      getVisibleLength()= 182
      getVolume() = /Users/weichiu/sandbox/hadoop/hadoop-hdfs-project/hadoop-hdfs-httpfs/target/test-dir/dfs/data/data1
      getBlockURI() = file:/Users/weichiu/sandbox/hadoop/hadoop-hdfs-project/hadoop-hdfs-httpfs/target/test-dir/dfs/data/data1/current/BP-837130339-172.16.1.88-1477434851452/current/finalized/subdir0/subdir0/blk_1073741825
      2016-10-25 15:34:45,167 INFO DataNode - opReadBlock BP-837130339-172.16.1.88-1477434851452:blk_1073741825_1182 received exception java.io.IOException: No data exists for block BP-837130339-172.16.1.88-1477434851452:blk_1073741825_1182
      2016-10-25 15:34:45,167 WARN DataNode - DatanodeRegistration(127.0.0.1:50131, datanodeUuid=41c96335-5e4b-4950-ac22-3d21b353abb8, infoPort=50133, infoSecurePort=0, ipcPort=50134, storageInfo=lv=-57;cid=testClusterID;nsid=1472068852;c=1477434851452):Got exception while serving BP-837130339-172.16.1.88-1477434851452:blk_1073741825_1182 to /127.0.0.1:51121
      java.io.IOException: No data exists for block BP-837130339-172.16.1.88-1477434851452:blk_1073741825_1182
      at org.apache.hadoop.hdfs.server.datanode.fsdataset.impl.FsDatasetImpl.getBlockInputStream(FsDatasetImpl.java:773)
      at org.apache.hadoop.hdfs.server.datanode.BlockSender.<init>(BlockSender.java:400)
      at org.apache.hadoop.hdfs.server.datanode.DataXceiver.readBlock(DataXceiver.java:581)
      at org.apache.hadoop.hdfs.protocol.datatransfer.Receiver.opReadBlock(Receiver.java:150)
      at org.apache.hadoop.hdfs.protocol.datatransfer.Receiver.processOp(Receiver.java:102)
      at org.apache.hadoop.hdfs.server.datanode.DataXceiver.run(DataXceiver.java:289)
      at java.lang.Thread.run(Thread.java:745)
      2016-10-25 15:34:45,168 INFO FSNamesystem - updatePipeline(blk_1073741825_1182, newGS=1183, newLength=182, newNodes=[127.0.0.1:50131], client=DFSClient_NONMAPREDUCE_-1743096965_197)
      2016-10-25 15:34:45,168 ERROR DataNode - 127.0.0.1:50131:DataXceiver error processing READ_BLOCK operation src: /127.0.0.1:51121 dst: /127.0.0.1:50131
      java.io.IOException: No data exists for block BP-837130339-172.16.1.88-1477434851452:blk_1073741825_1182
      at org.apache.hadoop.hdfs.server.datanode.fsdataset.impl.FsDatasetImpl.getBlockInputStream(FsDatasetImpl.java:773)
      at org.apache.hadoop.hdfs.server.datanode.BlockSender.<init>(BlockSender.java:400)
      at org.apache.hadoop.hdfs.server.datanode.DataXceiver.readBlock(DataXceiver.java:581)
      at org.apache.hadoop.hdfs.protocol.datatransfer.Receiver.opReadBlock(Receiver.java:150)
      at org.apache.hadoop.hdfs.protocol.datatransfer.Receiver.processOp(Receiver.java:102)
      at org.apache.hadoop.hdfs.server.datanode.DataXceiver.run(DataXceiver.java:289)
      at java.lang.Thread.run(Thread.java:745)
      2016-10-25 15:34:45,168 INFO FSNamesystem - updatePipeline(blk_1073741825_1182 => blk_1073741825_1183) success
      2016-10-25 15:34:45,170 WARN DFSClient - Found Checksum error for BP-837130339-172.16.1.88-1477434851452:blk_1073741825_1182 from DatanodeInfoWithStorage[127.0.0.1:50131,DS-a1878418-4f7f-4fc9-b3f7-d7ed780b5373,DISK] at 0
      2016-10-25 15:34:45,170 WARN DFSClient - No live nodes contain block BP-837130339-172.16.1.88-1477434851452:blk_1073741825_1182 after checking nodes = [DatanodeInfoWithStorage[127.0.0.1:50131,DS-a1878418-4f7f-4fc9-b3f7-d7ed780b5373,DISK]], ignoredNodes = null
      2016-10-25 15:34:45,170 INFO DFSClient - Could not obtain BP-837130339-172.16.1.88-1477434851452:blk_1073741825_1182 from any node: No live nodes contain current block Block locations: DatanodeInfoWithStorage[127.0.0.1:50131,DS-a1878418-4f7f-4fc9-b3f7-d7ed780b5373,DISK] Dead nodes: DatanodeInfoWithStorage[127.0.0.1:50131,DS-a1878418-4f7f-4fc9-b3f7-d7ed780b5373,DISK]. Will get new block locations from namenode and retry...
      2016-10-25 15:34:45,170 WARN DFSClient - DFS chooseDataNode: got # 1 IOException, will wait for 981.8085941094539 msec.
      2016-10-25 15:34:45,171 INFO clienttrace - src: /127.0.0.1:51130, dest: /127.0.0.1:50131, bytes: 183, op: HDFS_WRITE, cliID: DFSClient_NONMAPREDUCE_-1743096965_197, offset: 0, srvID: 41c96335-5e4b-4950-ac22-3d21b353abb8, blockid: BP-837130339-172.16.1.88-1477434851452:blk_1073741825_1183, duration: 2175363
      2016-10-25 15:34:45,171 INFO DataNode - PacketResponder: BP-837130339-172.16.1.88-1477434851452:blk_1073741825_1183, type=LAST_IN_PIPELINE terminating
      2016-10-25 15:34:45,172 INFO FSNamesystem - BLOCK* blk_1073741825_1183 is COMMITTED but not COMPLETE(numNodes= 0 < minimum = 1) in file /tmp/bar.txt
      2016-10-25 15:34:45,576 INFO StateChange - DIR* completeFile: /tmp/bar.txt is closed by DFSClient_NONMAPREDUCE_-1743096965_197
      2016-10-25 15:34:45,577 INFO httpfsaudit - [/tmp/bar.txt]
      2016-10-25 15:34:45,579 INFO AppendTestUtil - seed=-3144873070946578911, size=1
      2016-10-25 15:34:45,590 INFO audit - allowed=true ugi=weichiu (auth:PROXY) via weichiu (auth:SIMPLE) ip=/127.0.0.1 cmd=append src=/tmp/bar.txt dst=null perm=null proto=rpc
      2016-10-25 15:34:45,593 INFO DataNode - Receiving BP-837130339-172.16.1.88-1477434851452:blk_1073741825_1183 src: /127.0.0.1:51132 dest: /127.0.0.1:50131
      2016-10-25 15:34:45,593 INFO FsDatasetImpl - Appending to FinalizedReplica, blk_1073741825_1183, FINALIZED
      getNumBytes() = 183
      getBytesOnDisk() = 183
      getVisibleLength()= 183
      getVolume() = /Users/weichiu/sandbox/hadoop/hadoop-hdfs-project/hadoop-hdfs-httpfs/target/test-dir/dfs/data/data1
      getBlockURI() = file:/Users/weichiu/sandbox/hadoop/hadoop-hdfs-project/hadoop-hdfs-httpfs/target/test-dir/dfs/data/data1/current/BP-837130339-172.16.1.88-1477434851452/current/finalized/subdir0/subdir0/blk_1073741825
      2016-10-25 15:34:45,603 INFO FSNamesystem - updatePipeline(blk_1073741825_1183, newGS=1184, newLength=183, newNodes=[127.0.0.1:50131], client=DFSClient_NONMAPREDUCE_-1743096965_197)
      2016-10-25 15:34:45,603 INFO FSNamesystem - updatePipeline(blk_1073741825_1183 => blk_1073741825_1184) success
      2016-10-25 15:34:45,605 INFO clienttrace - src: /127.0.0.1:51132, dest: /127.0.0.1:50131, bytes: 184, op: HDFS_WRITE, cliID: DFSClient_NONMAPREDUCE_-1743096965_197, offset: 0, srvID: 41c96335-5e4b-4950-ac22-3d21b353abb8, blockid: BP-837130339-172.16.1.88-1477434851452:blk_1073741825_1184, duration: 1377229
      2016-10-25 15:34:45,605 INFO DataNode - PacketResponder: BP-837130339-172.16.1.88-1477434851452:blk_1073741825_1184, type=LAST_IN_PIPELINE terminating
      2016-10-25 15:34:45,606 INFO FSNamesystem - BLOCK* blk_1073741825_1184 is COMMITTED but not COMPLETE(numNodes= 0 < minimum = 1) in file /tmp/bar.txt
      2016-10-25 15:34:46,009 INFO StateChange - DIR* completeFile: /tmp/bar.txt is closed by DFSClient_NONMAPREDUCE_-1743096965_197
      2016-10-25 15:34:46,010 INFO httpfsaudit - [/tmp/bar.txt]
      2016-10-25 15:34:46,012 INFO AppendTestUtil - seed=-263001291976323720, size=1
      2016-10-25 15:34:46,022 INFO audit - allowed=true ugi=weichiu (auth:PROXY) via weichiu (auth:SIMPLE) ip=/127.0.0.1 cmd=append src=/tmp/bar.txt dst=null perm=null proto=rpc
      2016-10-25 15:34:46,024 INFO DataNode - Receiving BP-837130339-172.16.1.88-1477434851452:blk_1073741825_1184 src: /127.0.0.1:51133 dest: /127.0.0.1:50131
      2016-10-25 15:34:46,024 INFO FsDatasetImpl - Appending to FinalizedReplica, blk_1073741825_1184, FINALIZED
      getNumBytes() = 184
      getBytesOnDisk() = 184
      getVisibleLength()= 184
      getVolume() = /Users/weichiu/sandbox/hadoop/hadoop-hdfs-project/hadoop-hdfs-httpfs/target/test-dir/dfs/data/data1
      getBlockURI() = file:/Users/weichiu/sandbox/hadoop/hadoop-hdfs-project/hadoop-hdfs-httpfs/target/test-dir/dfs/data/data1/current/BP-837130339-172.16.1.88-1477434851452/current/finalized/subdir0/subdir0/blk_1073741825
      2016-10-25 15:34:46,032 INFO FSNamesystem - updatePipeline(blk_1073741825_1184, newGS=1185, newLength=184, newNodes=[127.0.0.1:50131], client=DFSClient_NONMAPREDUCE_-1743096965_197)
      2016-10-25 15:34:46,032 INFO FSNamesystem - updatePipeline(blk_1073741825_1184 => blk_1073741825_1185) success
      2016-10-25 15:34:46,033 INFO clienttrace - src: /127.0.0.1:51133, dest: /127.0.0.1:50131, bytes: 185, op: HDFS_WRITE, cliID: DFSClient_NONMAPREDUCE_-1743096965_197, offset: 0, srvID: 41c96335-5e4b-4950-ac22-3d21b353abb8, blockid: BP-837130339-172.16.1.88-1477434851452:blk_1073741825_1185, duration: 1112564
      2016-10-25 15:34:46,033 INFO DataNode - PacketResponder: BP-837130339-172.16.1.88-1477434851452:blk_1073741825_1185, type=LAST_IN_PIPELINE terminating
      2016-10-25 15:34:46,033 INFO FSNamesystem - BLOCK* blk_1073741825_1185 is COMMITTED but not COMPLETE(numNodes= 0 < minimum = 1) in file /tmp/bar.txt
      2016-10-25 15:34:46,156 INFO audit - allowed=true ugi=weichiu (auth:SIMPLE) ip=/127.0.0.1 cmd=open src=/tmp/bar.txt dst=null perm=null proto=rpc
      2016-10-25 15:34:46,158 INFO StateChange - DIR reportBadBlocks for block: BP-837130339-172.16.1.88-1477434851452:blk_1073741825_1182 on datanode: 127.0.0.1:50131
      Exception in thread "Thread-144" java.lang.RuntimeException: org.apache.hadoop.fs.ChecksumException: Checksum CRC32C not matched for file /tmp/bar.txt at position 0: expected=C893FEDE but computed=69322F90, algorithm=PureJavaCrc32C
      at org.apache.hadoop.fs.http.client.BaseTestHttpFSWith$1ReaderRunnable.run(BaseTestHttpFSWith.java:309)
      at java.lang.Thread.run(Thread.java:745)
      Caused by: org.apache.hadoop.fs.ChecksumException: Checksum CRC32C not matched for file /tmp/bar.txt at position 0: expected=C893FEDE but computed=69322F90, algorithm=PureJavaCrc32C
      at org.apache.hadoop.util.DataChecksum.throwChecksumException(DataChecksum.java:407)
      at org.apache.hadoop.util.DataChecksum.verifyChunked(DataChecksum.java:351)
      at org.apache.hadoop.util.DataChecksum.verifyChunkedSums(DataChecksum.java:311)
      at org.apache.hadoop.hdfs.client.impl.BlockReaderRemote.readNextPacket(BlockReaderRemote.java:216)
      at org.apache.hadoop.hdfs.client.impl.BlockReaderRemote.read(BlockReaderRemote.java:144)
      at org.apache.hadoop.hdfs.ByteArrayStrategy.readFromBlock(ReaderStrategy.java:119)
      at org.apache.hadoop.hdfs.DFSInputStream.readBuffer(DFSInputStream.java:704)
      at org.apache.hadoop.hdfs.DFSInputStream.readWithStrategy(DFSInputStream.java:765)
      at org.apache.hadoop.hdfs.DFSInputStream.read(DFSInputStream.java:814)
      at java.io.DataInputStream.read(DataInputStream.java:149)
      at org.apache.hadoop.fs.http.client.BaseTestHttpFSWith$1ReaderRunnable.run(BaseTestHttpFSWith.java:302)
      ... 1 more
      2016-10-25 15:34:46,437 INFO StateChange - DIR* completeFile: /tmp/bar.txt is closed by DFSClient_NONMAPREDUCE_-1743096965_197
      2016-10-25 15:34:46,437 INFO httpfsaudit - [/tmp/bar.txt]
      2016-10-25 15:34:46,440 INFO AppendTestUtil - seed=8756761565208093670, size=1
      2016-10-25 15:34:46,450 WARN StateChange - DIR* NameSystem.append: append: lastBlock=blk_1073741825_1185 of src=/tmp/bar.txt is not sufficiently replicated yet.
      2016-10-25 15:34:46,450 INFO Server - IPC Server handler 7 on 50130, call Call#25082 Retry#0 org.apache.hadoop.hdfs.protocol.ClientProtocol.append from 127.0.0.1:50147
      java.io.IOException: append: lastBlock=blk_1073741825_1185 of src=/tmp/bar.txt is not sufficiently replicated yet.
      at org.apache.hadoop.hdfs.server.namenode.FSDirAppendOp.appendFile(FSDirAppendOp.java:136)
      at org.apache.hadoop.hdfs.server.namenode.FSNamesystem.appendFile(FSNamesystem.java:2423)
      at org.apache.hadoop.hdfs.server.namenode.NameNodeRpcServer.append(NameNodeRpcServer.java:773)
      at org.apache.hadoop.hdfs.protocolPB.ClientNamenodeProtocolServerSideTranslatorPB.append(ClientNamenodeProtocolServerSideTranslatorPB.java:444)
      at org.apache.hadoop.hdfs.protocol.proto.ClientNamenodeProtocolProtos$ClientNamenodeProtocol$2.callBlockingMethod(ClientNamenodeProtocolProtos.java)
      at org.apache.hadoop.ipc.ProtobufRpcEngine$Server$ProtoBufRpcInvoker.call(ProtobufRpcEngine.java:467)
      at org.apache.hadoop.ipc.RPC$Server.call(RPC.java:990)
      at org.apache.hadoop.ipc.Server$RpcCall.run(Server.java:845)
      at org.apache.hadoop.ipc.Server$RpcCall.run(Server.java:788)
      at java.security.AccessController.doPrivileged(Native Method)
      at javax.security.auth.Subject.doAs(Subject.java:422)
      at org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1795)
      at org.apache.hadoop.ipc.Server$Handler.run(Server.java:2535)

      Exception in thread "Thread-143" java.lang.RuntimeException: java.io.IOException: HTTP status [500], exception [org.apache.hadoop.ipc.RemoteException], message [append: lastBlock=blk_1073741825_1185 of src=/tmp/bar.txt is not sufficiently replicated yet.]
      at org.apache.hadoop.fs.http.client.BaseTestHttpFSWith$1.run(BaseTestHttpFSWith.java:283)
      at java.lang.Thread.run(Thread.java:745)
      Caused by: java.io.IOException: HTTP status [500], exception [org.apache.hadoop.ipc.RemoteException], message [append: lastBlock=blk_1073741825_1185 of src=/tmp/bar.txt is not sufficiently replicated yet.]
      at org.apache.hadoop.util.HttpExceptionUtils.validateResponse(HttpExceptionUtils.java:159)
      at org.apache.hadoop.fs.http.client.HttpFSFileSystem$HttpFSDataOutputStream.close(HttpFSFileSystem.java:470)
      at org.apache.hadoop.fs.http.client.BaseTestHttpFSWith$1.run(BaseTestHttpFSWith.java:279)
      ... 1 more

      org.apache.hadoop.fs.ChecksumException: Checksum CRC32C not matched for file /tmp/bar.txt at position 0: expected=C893FEDE but computed=69322F90, algorithm=PureJavaCrc32C

      at org.apache.hadoop.util.DataChecksum.throwChecksumException(DataChecksum.java:407)
      at org.apache.hadoop.util.DataChecksum.verifyChunked(DataChecksum.java:351)
      at org.apache.hadoop.util.DataChecksum.verifyChunkedSums(DataChecksum.java:311)
      at org.apache.hadoop.hdfs.client.impl.BlockReaderRemote.readNextPacket(BlockReaderRemote.java:216)
      at org.apache.hadoop.hdfs.client.impl.BlockReaderRemote.read(BlockReaderRemote.java:144)
      at org.apache.hadoop.hdfs.ByteArrayStrategy.readFromBlock(ReaderStrategy.java:119)
      at org.apache.hadoop.hdfs.DFSInputStream.readBuffer(DFSInputStream.java:704)
      at org.apache.hadoop.hdfs.DFSInputStream.readWithStrategy(DFSInputStream.java:765)
      at org.apache.hadoop.hdfs.DFSInputStream.read(DFSInputStream.java:814)
      at java.io.DataInputStream.read(DataInputStream.java:149)
      at org.apache.hadoop.fs.http.client.BaseTestHttpFSWith$1ReaderRunnable.run(BaseTestHttpFSWith.java:302)
      at java.lang.Thread.run(Thread.java:745)

      Attachments

        1. HDFS-11056.reproduce.patch
          3 kB
          Wei-Chiu Chuang
        2. HDFS-11056.branch-2.patch
          10 kB
          Wei-Chiu Chuang
        3. HDFS-11056.branch-2.7.patch
          13 kB
          Wei-Chiu Chuang
        4. HDFS-11056.002.patch
          11 kB
          Wei-Chiu Chuang
        5. HDFS-11056.001.patch
          4 kB
          Wei-Chiu Chuang

        Issue Links

        breaks

        Bug - A problem which impairs or prevents the functions of the product. HDFS-11229 HDFS-11056 failed to close meta file

        • Blocker - Blocks development and/or testing work, production could not run
        • Resolved
        contains

        Bug - A problem which impairs or prevents the functions of the product. HDFS-6804 Add test for race condition between transferring block and appending block causes "Unexpected checksum mismatch exception"

        • Major - Major loss of function.
        • Resolved
        is superceded by

        Improvement - An improvement or enhancement to an existing feature or task. HDFS-11187 Optimize disk access for last partial chunk checksum of Finalized replica

        • Major - Major loss of function.
        • Resolved
        relates to

        Bug - A problem which impairs or prevents the functions of the product. HDFS-6804 Add test for race condition between transferring block and appending block causes "Unexpected checksum mismatch exception"

        • Major - Major loss of function.
        • Resolved

        Activity
          This comment will be Viewable by All Users Viewable by All Users
          Cancel

          People

            weichiu Wei-Chiu Chuang
            weichiu Wei-Chiu Chuang
            Votes:
            0 Vote for this issue
            Watchers:
            14 Stop watching this issue

            Dates

              Created:
              Updated:
              Resolved:

              Slack

                Issue deployment