[
https://issues.apache.org/jira/browse/HADOOP-1898?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12535852
]
Owen O'Malley commented on HADOOP-1898:
---------------------------------------
This looks right, but I'd suggest a minor style change in the patch:
{code}
boolean dumpStack = false;
synchronized (ReflectionUtils.class) {
long now = System.currentTimeMillis();
if (new - previousLogTime > minInterval * 1000) {
previousLogTime = now;
dumpStack = true;
}
}
if (dumpStack) {
..
}
{code}
> locking for the ReflectionUtils.logThreadInfo is too conservative
> -----------------------------------------------------------------
>
> Key: HADOOP-1898
> URL: https://issues.apache.org/jira/browse/HADOOP-1898
> Project: Hadoop
> Issue Type: Bug
> Components: util
> Affects Versions: 0.15.0
> Reporter: Owen O'Malley
> Assignee: Amareshwari Sri Ramadasu
> Fix For: 0.16.0
>
> Attachments: patch-1898.txt, patch-1898.txt, patch-1898.txt
>
>
> When the RPC servers get into trouble with their call queues backing up, they
> occasionally dump the call stacks. These are very useful for identifying hot
> spots, but the locking is too conservative and so all of the handlers are
> blocked while the thread call stacks are dumped.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
