[ https://issues.apache.org/jira/browse/HDFS-1523?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12965569#action_12965569 ]
Konstantin Boudnik commented on HDFS-1523: ------------------------------------------ Here's the scenario of the test: - read the file in chunks of 134217728 (128Mb) - after last full read there are 513 bytes to be read - 512 bytes of those have to be read from the first block - 1 byte is going to be read from the last block (second one) When test passes before reading last 513 bytes call to FSNameSystem.getBlockLocationsInternal returns last block of the file (size=1) {noformat} 2010-11-30 19:44:49,439 DEBUG namenode.FSNamesystem (FSNamesystem.java:getBlockLocationsInternal(866)) - blocks = [blk_-6779333650185181528_1001, blk_-3599 865432887782445_1001] 2010-11-30 19:44:49,440 DEBUG namenode.FSNamesystem (FSNamesystem.java:getBlockLocationsInternal(881)) - last = blk_-3599865432887782445_1001 2010-11-30 19:44:49,457 INFO FSNamesystem.audit (FSNamesystem.java:logAuditEvent(148)) - ugi=cos ip=/127.0.0.1 cmd=open src=/home/cos/work/ H0.23/git/hdfs/build/test/data/2147484160.dat dst=null perm=null 2010-11-30 19:44:49,459 DEBUG hdfs.DFSClient (DFSInputStream.java:openInfo(113)) - newInfo = LocatedBlocks{ fileLength=2147484161 underConstruction=false blocks=[LocatedBlock{blk_-6779333650185181528_1001; getBlockSize()=2147484160; corrupt=false; offset=0; locs=[127.0.0.1:35608]}] lastLocatedBlock=LocatedBlock{blk_-3599865432887782445_1001; getBlockSize()=1; corrupt=false; offset=2147484160; locs=[127.0.0.1:35608]} isLastBlockComplete=true} ... 2010-11-30 19:45:23,880 INFO DataNode.clienttrace (BlockSender.java:sendBlock(491)) - src: /127.0.0.1:35608, dest: /127.0.0.1:51640, bytes: 2164261380, op : HDFS_READ, cliID: DFSClient_1273505070, offset: 0, srvID: DS-212336177-192.168.102.126-35608-1291175019763, blockid: blk_-6779333650185181528_1001, durat ion: 34361753463 2010-11-30 19:45:24,030 DEBUG namenode.FSNamesystem (FSNamesystem.java:getBlockLocationsInternal(866)) - blocks = [blk_-6779333650185181528_1001, blk_-3599 865432887782445_1001] 2010-11-30 19:45:24,030 DEBUG namenode.FSNamesystem (FSNamesystem.java:getBlockLocationsInternal(881)) - last = blk_-3599865432887782445_1001 2010-11-30 19:45:24,031 INFO FSNamesystem.audit (FSNamesystem.java:logAuditEvent(148)) - ugi=cos ip=/127.0.0.1 cmd=open src=/home/cos/work/H0.23/git/hdfs/build/test/data/2147484160.dat dst=null perm=null 2010-11-30 19:45:24,032 DEBUG datanode.DataNode (DataXceiver.java:<init>(86)) - Number of active connections is: 2 2010-11-30 19:45:24,099 DEBUG datanode.DataNode (DataXceiver.java:run(135)) - DatanodeRegistration(127.0.0.1:35608, storageID=DS-212336177-192.168.102.126-35608-1291175019763, infoPort=46218, ipcPort=38099):Number of active connections is: 3 2010-11-30 19:45:24,099 DEBUG datanode.DataNode (BlockSender.java:<init>(140)) - block=blk_-3599865432887782445_1001, replica=FinalizedReplica, blk_-3599865432887782445_1001, FINALIZED getNumBytes() = 1 getBytesOnDisk() = 1 getVisibleLength()= 1 getVolume() = /home/cos/work/H0.23/git/hdfs/build/test/data/dfs/data/data2/current/finalized getBlockFile() = /home/cos/work/H0.23/git/hdfs/build/test/data/dfs/data/data2/current/finalized/blk_-3599865432887782445 unlinked=false 2010-11-30 19:45:24,101 DEBUG datanode.DataNode (BlockSender.java:<init>(231)) - replica=FinalizedReplica, blk_-3599865432887782445_1001, FINALIZED getNumBytes() = 1 getBytesOnDisk() = 1 getVisibleLength()= 1 getVolume() = /home/cos/work/H0.23/git/hdfs/build/test/data/dfs/data/data2/current/finalized getBlockFile() = /home/cos/work/H0.23/git/hdfs/build/test/data/dfs/data/data2/current/finalized/blk_-3599865432887782445 unlinked=false 2010-11-30 19:45:24,103 INFO DataNode.clienttrace (BlockSender.java:sendBlock(491)) - src: /127.0.0.1:35608, dest: /127.0.0.1:51644, bytes: 5, op: HDFS_READ, cliID: DFSClient_1273505070, offset: 0, srvID: DS-212336177-192.168.102.126-35608-1291175019763, blockid: blk_-3599865432887782445_1001, duration: 1854472 {noformat} In case of failure: {noformat} 2010-11-30 19:35:49,426 DEBUG namenode.FSNamesystem (FSNamesystem.java:getBlockLocationsInternal(866)) - blocks = [blk_1170274882140601397_1001, blk_289191 6181488413346_1001] 2010-11-30 19:35:49,426 DEBUG namenode.FSNamesystem (FSNamesystem.java:getBlockLocationsInternal(881)) - last = blk_2891916181488413346_1001 2010-11-30 19:35:49,427 INFO FSNamesystem.audit (FSNamesystem.java:logAuditEvent(148)) - ugi=cos ip=/127.0.0.1 cmd=open src=/home/cos/work/ hadoop/git/hdfs/build/test/data/2147484160.dat dst=null perm=null 2010-11-30 19:35:49,428 DEBUG hdfs.DFSClient (DFSInputStream.java:openInfo(113)) - newInfo = LocatedBlocks{ fileLength=2147484161 underConstruction=false blocks=[LocatedBlock{blk_1170274882140601397_1001; getBlockSize()=2147484160; corrupt=false; offset=0; locs=[127.0.0.1:35644]}] lastLocatedBlock=LocatedBlock{blk_2891916181488413346_1001; getBlockSize()=1; corrupt=false; offset=2147484160; locs=[127.0.0.1:35644]} isLastBlockComplete=true} ... 2010-11-30 19:36:16,761 INFO DataNode.clienttrace (BlockSender.java:sendBlock(491)) - src: /127.0.0.1:35644, dest: /127.0.0.1:52290, bytes: 2164194816, op : HDFS_READ, cliID: DFSClient_635470834, offset: 0, srvID: DS-514949605-127.0.0.1-35644-1291174495289, blockid: blk_1170274882140601397_1001, duration: 273 20364567 2010-11-30 19:36:16,761 DEBUG datanode.DataNode (DataXceiver.java:run(135)) - DatanodeRegistration(127.0.0.1:35644, storageID=DS-514949605-127.0.0.1-35644- 1291174495289, infoPort=56924, ipcPort=60439):Number of active connections is: 2 2010-11-30 19:36:16,794 DEBUG datanode.DataNode (DataXceiver.java:<init>(86)) - Number of active connections is: 1 2010-11-30 19:36:16,794 DEBUG datanode.DataNode (BlockSender.java:<init>(140)) - block=blk_1170274882140601397_1001, replica=FinalizedReplica, blk_11702748 82140601397_1001, FINALIZED getNumBytes() = 2147484160 getBytesOnDisk() = 2147484160 getVisibleLength()= 2147484160 getVolume() = /home/cos/work/hadoop/git/hdfs/build/test/data/dfs/data/data1/current/finalized getBlockFile() = /home/cos/work/hadoop/git/hdfs/build/test/data/dfs/data/data1/current/finalized/blk_1170274882140601397 unlinked=false 2010-11-30 19:36:16,795 DEBUG datanode.DataNode (BlockSender.java:<init>(231)) - replica=FinalizedReplica, blk_1170274882140601397_1001, FINALIZED getNumBytes() = 2147484160 getBytesOnDisk() = 2147484160 getVisibleLength()= 2147484160 getVolume() = /home/cos/work/hadoop/git/hdfs/build/test/data/dfs/data/data1/current/finalized getBlockFile() = /home/cos/work/hadoop/git/hdfs/build/test/data/dfs/data/data1/current/finalized/blk_1170274882140601397 unlinked=false 2010-11-30 19:36:17,276 INFO DataNode.clienttrace (BlockSender.java:sendBlock(491)) - src: /127.0.0.1:35644, dest: /127.0.0.1:52296, bytes: 135200256, op: HDFS_READ, cliID: DFSClient_635470834, offset: 2013265920, srvID: DS-514949605-127.0.0.1-35644-1291174495289, blockid: blk_1170274882140601397_1001, durat ion: 480762241 2010-11-30 19:36:17,276 DEBUG datanode.DataNode (DataXceiver.java:run(135)) - DatanodeRegistration(127.0.0.1:35644, storageID=DS-514949605-127.0.0.1-35644- 1291174495289, infoPort=56924, ipcPort=60439):Number of active connections is: 2 2010-11-30 19:36:17,290 WARN hdfs.DFSClient (DFSInputStream.java:readBuffer(486)) - Exception while reading from blk_1170274882140601397_1001 of /home/cos /work/hadoop/git/hdfs/build/test/data/2147484160.dat from 127.0.0.1:35644: java.io.IOException: Premature EOF from inputStream at org.apache.hadoop.io.IOUtils.readFully(IOUtils.java:118) at org.apache.hadoop.hdfs.BlockReader.readChunk(BlockReader.java:275) {noformat} so it seems like the test fails because wrong block is being read or something. > TestLargeBlock is failing on trunk > ---------------------------------- > > Key: HDFS-1523 > URL: https://issues.apache.org/jira/browse/HDFS-1523 > Project: Hadoop HDFS > Issue Type: Bug > Components: test > Affects Versions: 0.22.0 > Reporter: Konstantin Boudnik > > TestLargeBlock is failing for more than a week not on 0.22 and trunk with > {noformat} > java.io.IOException: Premeture EOF from inputStream > at org.apache.hadoop.io.IOUtils.readFully(IOUtils.java:118) > at org.apache.hadoop.hdfs.BlockReader.readChunk(BlockReader.java:275) > {noformat} -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.