TestHLog.testSplit hangs
------------------------
Key: HBASE-2786
URL: https://issues.apache.org/jira/browse/HBASE-2786
Project: HBase
Issue Type: Bug
Reporter: Jean-Daniel Cryans
Priority: Blocker
Fix For: 0.21.0
This a blocker had it blocks and times out Hudson.
It seems that when we upgraded to latest of 0.20-append we got into a new
situation where we can't recover a file that's empty if the original writer is
still alive:
{noformat}
2010-06-24 10:41:20,645 DEBUG [main] wal.HLog(1281): Splitting hlog 4 of 4:
hdfs://localhost:64456/hbase/testSplit/.logs/hlog.1277401279534, length=0
2010-06-24 10:41:20,645 INFO [main] util.FSUtils(612): Recovering
filehdfs://localhost:64456/hbase/testSplit/.logs/hlog.1277401279534
2010-06-24 10:41:20,647 WARN [IPC Server handler 5 on 64456]
namenode.FSNamesystem(1156): DIR* NameSystem.startFile:
failed to create file /hbase/testSplit/.logs/hlog.1277401279534 for
DFSClient_-1981892617 on client 127.0.0.1 because current leaseholder is trying
to recreate file
...
2010-06-24 10:42:24,919 WARN [IPC Server handler 0 on 64456]
namenode.FSNamesystem(1156): DIR* NameSystem.startFile:
failed to create file /hbase/testSplit/.logs/hlog.1277401279534 for
DFSClient_-1981892617 on client 127.0.0.1 because current leaseholder is trying
to recreate file.
2010-06-24 10:42:24,919 WARN [main] util.FSUtils(631): Waited 64274ms for
lease recovery on
hdfs://localhost:64456/hbase/testSplit/.logs/hlog.1277401279534:org.apache.hadoop.hdfs.protocol.AlreadyBeingCreatedException:
failed to create file /hbase/testSplit/.logs/hlog.1277401279534 for
DFSClient_-1981892617 on client 127.0.0.1 because current leaseholder is trying
to recreate file.
at
org.apache.hadoop.hdfs.server.namenode.FSNamesystem.startFileInternal(FSNamesystem.java:1058)
at
org.apache.hadoop.hdfs.server.namenode.FSNamesystem.appendFile(FSNamesystem.java:1171)
at
org.apache.hadoop.hdfs.server.namenode.NameNode.append(NameNode.java:396)
{noformat}
We could just not roll the latest log and it would probably fix the issue, but
I wonder if we could change something in HDFS instead. Todd?
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
