[jboss-jira] [JBoss JIRA] (WFLY-2644) Java heap space exceeded looping connect/disconnect cycle
[
https://issues.jboss.org/browse/WFLY-2644?page=com.atlassian.jira.plugin....
]
David Lloyd commented on WFLY-2644:
-----------------------------------
I did some researching on possible causes of this. I think the only piece that could be
causing this is in XNIO, in the JSSE SSL implementation of XnioSsl, which allocates a
buffer pool on each creation but never releases it; if you use a few buffers from the
provider and then leak it, the thread locals seem to stick around forever.
The problem is that thus far [~aloubyansky] wasn't able to show that disabling SSL
fixed the problem so I can't be 100% sure this is the cause. But given that it
appears to be the only code path that uses buffer pools in Remoting, I think this must be
the problem.
Given the nature of JSSE, I think it might make sense to just have a single static pool of
17K buffers that is shared for all JsseXnioSsl instances. This may solve the issue.
Java heap space exceeded looping connect/disconnect cycle
---------------------------------------------------------
Key: WFLY-2644
URL: https://issues.jboss.org/browse/WFLY-2644
Project: WildFly
Issue Type: Bug
Security Level: Public(Everyone can see)
Components: CLI
Affects Versions: 8.0.0.Beta1
Reporter: Alexey Loubyansky
Assignee: Alexey Loubyansky
Fix For: 8.0.0.CR1
CLIModelControllerClient implementation is not cleaning up something related to
connection handling which leads to OOME repeating a cycle of ensureConnected(...) and
close(). On my laptop it happens after around 1400 iterations.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira