[
https://issues.apache.org/jira/browse/HBASE-15613?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15270086#comment-15270086
]
Enis Soztutar commented on HBASE-15613:
---------------------------------------
The problem is this:
We are running with priority handlers of 5 in tests. SimpleRpcController shards
5 handlers into two queues, giving 1 queue 3 handlers, and the other 2
handlers. After HBASE-15295, ACL updates also execute in the high-priority
handler threads, and in the ACL coprocessor execution we are doing meta
lookups. So we block the handler while doing the meta lookups. These meta
requests also go to the same handler queues (which only have 2 or 3 handlers)
and they are already blocked ( {{PriorityRpcServer.handler}}'s with {{queue=1}}
are blocked, those with {{queue=0}} are not). This causes a deadlock in 50% of
the time, and thus causes the UT timeout.
See HBASE-15765.
> TestNamespaceCommand times out
> ------------------------------
>
> Key: HBASE-15613
> URL: https://issues.apache.org/jira/browse/HBASE-15613
> Project: HBase
> Issue Type: Bug
> Reporter: Enis Soztutar
> Assignee: Enis Soztutar
>
> We think that the root cause maybe HBASE-15295. Will inspect more.
> See
> https://issues.apache.org/jira/browse/HBASE-15537?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
