[
https://issues.apache.org/jira/browse/HBASE-11445?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Jeffrey Zhong updated HBASE-11445:
----------------------------------
Attachment: hbase-11445.patch
[~jmhsieh] Could you check the patch? It seems caused by a race condition if
all ProcedureMembers create subprocedures around the same time. Thanks.
> TestZKProcedure#testMultiCohortWithMemberTimeoutDuringPrepare is flaky
> ----------------------------------------------------------------------
>
> Key: HBASE-11445
> URL: https://issues.apache.org/jira/browse/HBASE-11445
> Project: HBase
> Issue Type: Bug
> Components: snapshots
> Reporter: Jeffrey Zhong
> Fix For: 0.99.0, 0.98.3
>
> Attachments: failure.txt, hbase-11445.patch
>
>
> Recently there is a failure from Jenkins
> build:https://builds.apache.org/job/HBase-0.98/364/testReport/junit/org.apache.hadoop.hbase.procedure/TestZKProcedure/testMultiCohortWithMemberTimeoutDuringPrepare/.
> Below are related log message and Member: 'one' joining twice:
> {noformat}
> 2014-06-29 19:26:34,101 DEBUG [member: 'three' subprocedure-pool11-thread-1]
> procedure.ZKProcedureMemberRpcs(237): Member: 'one' joining acquired barrier
> for procedure (op) in zk
> 2014-06-29 19:26:34,101 DEBUG [member: 'one' subprocedure-pool9-thread-1]
> procedure.Subprocedure(162): Subprocedure 'op' locally acquired
> 2014-06-29 19:26:34,101 DEBUG [member: 'one' subprocedure-pool9-thread-1]
> procedure.ZKProcedureMemberRpcs(237): Member: 'one' joining acquired barrier
> for procedure (op) in zk
> {noformat}
--
This message was sent by Atlassian JIRA
(v6.2#6252)
