[jboss-jira] [JBoss JIRA] Commented: (JGRP-495) Memory leak in NAKACK

[Bela Ban (JIRA)]

    [ http://jira.jboss.com/jira/browse/JGRP-495?page=comments#action_12361168 ] 
            
Bela Ban commented on JGRP-495:
-------------------------------

oob_loopback_msgs seems to have been the culprit; we added messages (even regular ones), but only removed OOB messages, causing this hashmap to grow indefinitely !
Testing with 3 million messages each (2 nodes) works... will need to do more testing, with larger message sets.
Result for 2 nodes, each sending 3 million 1K messages: 17800 messages/sec:


-- results:

192.168.5.2:3464:
num_msgs_expected=6000000, num_msgs_received=6000000 (loss rate=0.0%), received=6GB, time=335687ms, msgs/sec=17873.79, throughput=17.87MB

192.168.5.2:3474 (myself):
num_msgs_expected=6000000, num_msgs_received=6000000 (loss rate=0.0%), received=6GB, time=335969ms, msgs/sec=17858.79, throughput=17.86MB

combined: 17866.29 msgs/sec averaged over all receivers (throughput=17.87MB/sec)


> Memory leak in NAKACK
> ---------------------
>
>                 Key: JGRP-495
>                 URL: http://jira.jboss.com/jira/browse/JGRP-495
>             Project: JGroups
>          Issue Type: Bug
>            Reporter: Bela Ban
>         Assigned To: Bela Ban
>            Priority: Blocker
>             Fix For: 2.5
>
>
> Probably due to the changes introduced in NAKACK/NakReceiverWindow (http://jira.jboss.com/jira/browse/JGRP-281).
> To reproduce, run perf.Test on 2 nodes with 1 million 1K messages each. JProfiler shows that we are leaking Map.Entry instances in NAKACK. They seem to accumulate in xmit_table.

-- 
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators: http://jira.jboss.com/jira/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira