[
https://issues.apache.org/jira/browse/HBASE-11517?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14064090#comment-14064090
]
Mikhail Antonov commented on HBASE-11517:
-----------------------------------------
Hm. [~enis] as I understand closing of table admins and tables (as well as
adding timeouts) was added by [~stack] in that intermediate patch in order to
get more debug information and stabilize the test. In my patch I reverted those
changes as I believed that was not the root cause. We're not eclipsing the
problem by not closeing hbase admins, we're bypassing the problem by delaying
shutdown of second minicluster up until afterClass call (the same technique as
used in TestPerTableCFReplication, for example?). What do you think?
The problem is that in MiniHBaseCluster#shutdown() we call
HConnectionManager.deleteAllConnections(false); which deleted all hconnection
in the JVM. Seems like all tests which run more than 1 minicluster now tend to
shutdown all of them at the very end.
So what's the consensus to proceed?
> TestReplicaWithCluster turns zombie
> -----------------------------------
>
> Key: HBASE-11517
> URL: https://issues.apache.org/jira/browse/HBASE-11517
> Project: HBase
> Issue Type: Bug
> Reporter: stack
> Assignee: Mikhail Antonov
> Fix For: 0.99.0, 2.0.0
>
> Attachments: 10930v4.txt, 11517.timeouts.txt, 11517v2.txt,
> 11517v2.txt, HBASE-11517_v1-mantonov.patch
>
>
> Happened a few times for me fixing unrelated findbugs. Here is example:
> https://builds.apache.org/job/PreCommit-HBASE-Build/10065//consoleFull See
> how it is hanging creating a table:
> "pool-1-thread-1" prio=10 tid=0x00007f1714657000 nid=0x4b7f waiting on
> condition [0x00007f16e9f80000]
> java.lang.Thread.State: TIMED_WAITING (sleeping)
> at java.lang.Thread.sleep(Native Method)
> at
> org.apache.hadoop.hbase.client.HBaseAdmin.createTable(HBaseAdmin.java:539)
> at
> org.apache.hadoop.hbase.client.HBaseAdmin.createTable(HBaseAdmin.java:424)
> at
> org.apache.hadoop.hbase.HBaseTestingUtility.createTable(HBaseTestingUtility.java:1185)
> at
> org.apache.hadoop.hbase.client.TestReplicaWithCluster.testCreateDeleteTable(TestReplicaWithCluster.java:138)
--
This message was sent by Atlassian JIRA
(v6.2#6252)