[
https://issues.apache.org/jira/browse/ZOOKEEPER-2185?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14535074#comment-14535074
]
Chris Nauroth commented on ZOOKEEPER-2185:
------------------------------------------
We probably want to put this in the "Supervision" section, because running
under a supervisor should be a pre-requisite for the kill trigger option.
http://zookeeper.apache.org/doc/r3.4.6/zookeeperAdmin.html#sc_supervision
> Document recommendation to use -XX:HeapDumpOnOutOfMemoryError and
> -XX:OnOutOfMemoryError="kill %p" for server.
> --------------------------------------------------------------------------------------------------------------
>
> Key: ZOOKEEPER-2185
> URL: https://issues.apache.org/jira/browse/ZOOKEEPER-2185
> Project: ZooKeeper
> Issue Type: Improvement
> Components: documentation
> Reporter: Chris Nauroth
> Assignee: Chris Nauroth
> Priority: Minor
>
> Continuing to run a server process after it runs out of memory can lead to
> unexpected behavior. This issue proposes that we update documentation to
> recommend adding these JVM options:
> # {{-XX:HeapDumpOnOutOfMemoryError}} for help with post-mortem analysis of
> why the process ran out of memory.
> # {{-XX:OnOutOfMemoryError="kill %p"}} to kill the JVM process, under the
> assumption that a process monitor will restart it.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
