[
https://issues.apache.org/jira/browse/HDFS-8607?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Walter Su updated HDFS-8607:
----------------------------
Attachment: HDFS-8607.01.patch
{noformat}
2015-06-15 17:02:35,962 WARN hdfs.DFSClient
(DFSInputStream.java:actualGetFromOneDataNode(1218)) - Connection failure:
Failed to connect to /127.0.0.1:58508 for file
/srcdat/nine/eight/5842207179401855738 for block
BP-1622355698-9.96.1.34-1434358952058:blk_1073741833_1009:java.io.IOException:
Got error, status message opReadBlock
BP-1622355698-9.96.1.34-1434358952058:blk_1073741833_1009 received exception
java.io.IOException: BlockId 1073741833 is not valid., for OP_READ_BLOCK,
self=/127.0.0.1:43788, remote=/127.0.0.1:58508, for file
/srcdat/nine/eight/5842207179401855738, for pool
BP-1622355698-9.96.1.34-1434358952058 block 1073741833_1009
java.io.IOException: Got error, status message opReadBlock
BP-1622355698-9.96.1.34-1434358952058:blk_1073741833_1009 received exception
java.io.IOException: BlockId 1073741833 is not valid., for OP_READ_BLOCK,
self=/127.0.0.1:43788, remote=/127.0.0.1:58508, for file
/srcdat/nine/eight/5842207179401855738, for pool
BP-1622355698-9.96.1.34-1434358952058 block 1073741833_1009
at
org.apache.hadoop.hdfs.protocol.datatransfer.DataTransferProtoUtil.checkBlockOpStatus(DataTransferProtoUtil.java:140)
at
org.apache.hadoop.hdfs.RemoteBlockReader2.checkSuccess(RemoteBlockReader2.java:453)
at
org.apache.hadoop.hdfs.RemoteBlockReader2.newBlockReader(RemoteBlockReader2.java:421)
at
org.apache.hadoop.hdfs.BlockReaderFactory.getRemoteBlockReader(BlockReaderFactory.java:819)
at
org.apache.hadoop.hdfs.BlockReaderFactory.getRemoteBlockReaderFromTcp(BlockReaderFactory.java:698)
at
org.apache.hadoop.hdfs.BlockReaderFactory.build(BlockReaderFactory.java:358)
at
org.apache.hadoop.hdfs.DFSInputStream.getBlockReader(DFSInputStream.java:655)
at
org.apache.hadoop.hdfs.DFSInputStream.actualGetFromOneDataNode(DFSInputStream.java:1178)
at
org.apache.hadoop.hdfs.DFSInputStream.actualGetFromOneDataNode(DFSInputStream.java:1141)
at
org.apache.hadoop.hdfs.DFSInputStream.fetchBlockByteRange(DFSInputStream.java:1099)
...
{noformat}
With the patch, we got the expected IOException and know it's properly handled.
> TestFileCorruption doesn't work as expected
> -------------------------------------------
>
> Key: HDFS-8607
> URL: https://issues.apache.org/jira/browse/HDFS-8607
> Project: Hadoop HDFS
> Issue Type: Bug
> Reporter: Walter Su
> Assignee: Walter Su
> Attachments: HDFS-8607.01.patch
>
>
> Although it passes, it's useless.
> {code}
> 77 File[] blocks = data_dir.listFiles();
> 78 assertTrue("Blocks do not exist in data-dir", (blocks != null) &&
> (blocks.length > 0));
> 79 for (int idx = 0; idx < blocks.length; idx++) {
> 80 if (!blocks[idx].getName().startsWith(Block.BLOCK_FILE_PREFIX)) {
> 81 continue;
> 82 }
> 83 System.out.println("Deliberately removing file
> "+blocks[idx].getName());
> 84 assertTrue("Cannot remove file.", blocks[idx].delete());
> 85 }
> {code}
> blocks are located at finalized/subdir0/subdir0, but line 77 only returns
> "subdir0" because {{File.listFiles()}} is not recursive. So line 83~84 will
> never be excuted.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
