[jboss-jira] [JBoss JIRA] (JGRP-1743) NAKACK*: log a warning if xmit_table gets large

[Dennis Reed (JIRA)]

    [ https://issues.jboss.org/browse/JGRP-1743?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12923777#comment-12923777 ] 

Dennis Reed commented on JGRP-1743:
-----------------------------------

Opening in response to an outage with a major customer, where a misconfiguration (multiple JDG caches accidentally clustered together,
some using NAKACK and some using NAKACK2, so no messages were ever cleared) caused an OutOfMemory situation when 
NAKACK/NAKACK2 had GB's of memory worth of messages.

We've also seen large retention before in NAKACK due to a bug.

NAKACK should never have a huge number of messages in a normal situation, so this seems like a pretty easy place to put a warning
that something is wrong if the number of messages gets too high.
                
> NAKACK*: log a warning if xmit_table gets large
> -----------------------------------------------
>
>                 Key: JGRP-1743
>                 URL: https://issues.jboss.org/browse/JGRP-1743
>             Project: JGroups
>          Issue Type: Feature Request
>    Affects Versions: 3.4.1
>            Reporter: Dennis Reed
>            Assignee: Bela Ban
>            Priority: Minor
>
> There have been some cases where NAKACK* accumulates large number of messages, causing OutOfMemoryErrors.
> Large numbers of messages in NAKACK* would almost always indicate a problem, which the user should be aware of.
> Logging a warning if the number of messages in NAKACK passes a configurable threshold would let the user investigate and take action sooner before a catastrophic OutOfMemoryError, or other issues occur.

--
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