[
https://issues.apache.org/jira/browse/HBASE-6736?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13462897#comment-13462897
]
nkeywal commented on HBASE-6736:
--------------------------------
There are multiple synchro issues.
One of them is
{code}
@Override
protected void chore() {
// [...]
for (Map.Entry<String, Task> e : tasks.entrySet()) {
{code}
As we're iterating over a set that can be modified we can have reliability
issues, cf. javadoc: "If the map is modified while an iteration over the set is
in progress (except through the iterator's own remove operation, or through the
setValue operation on a map entry returned by the iterator) the results of the
iteration are undefined."
> Distributed Split: a split tasks can be mark as DONE but keep unassigned
> ------------------------------------------------------------------------
>
> Key: HBASE-6736
> URL: https://issues.apache.org/jira/browse/HBASE-6736
> Project: HBase
> Issue Type: Bug
> Components: master
> Affects Versions: 0.96.0
> Reporter: nkeywal
>
> Real cluster, scenario mentioned on HBASE-5843.
> Got it once out of 5 tests on 0.96
> Didn't get it on 0.94 after 3 tests.
> It seems we have a race condition on split logs: the task was nearly
> simultaneously marked as done and resubmitted. Then it remained in the
> unassigned state.
> 2012-09-04 17:27:06,237 DEBUG org.apache.hadoop.hbase.master.SplitLogManager:
> total tasks = 1 unassigned = 0
> 2012-09-04 17:27:06,237 INFO org.apache.hadoop.hbase.master.SplitLogManager:
> resubmitted 1 out of 1 tasks
> 2012-09-04 17:27:06,237 DEBUG org.apache.hadoop.hbase.master.SplitLogManager:
> task not yet acquired
> /hbase/splitlog/hdfs%3A%2F%2FBOX1%3A9000%2Fhbase%2F.logs%2FBOX0%2C60020%2C1346772046399-splitting%2FBOX0%252C60020%252C1346772046399.1346772046609
> ver = 7
> 2012-09-04 17:27:06,314 INFO org.apache.hadoop.hbase.master.SplitLogManager:
> task /hbase/splitlog/RESCAN0000000002 entered state: DONE
> BOX1,60000,1346771990737
> 2012-09-04 17:27:06,337 DEBUG
> org.apache.hadoop.hbase.master.SplitLogManager$DeleteAsyncCallback: deleted
> /hbase/splitlog/RESCAN0000000002
> 2012-09-04 17:27:06,337 DEBUG org.apache.hadoop.hbase.master.SplitLogManager:
> deleted task without in memory state /hbase/splitlog/RESCAN0000000002
> 2012-09-04 17:27:07,226 DEBUG org.apache.hadoop.hbase.master.SplitLogManager:
> total tasks = 1 unassigned = 1
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
