[jboss-jira] [JBoss JIRA] Commented: (JBCACHE-1332) Lock in eviction.Region:putNodeEvent()

[Nicolas Lebreton (JIRA)]

    [ http://jira.jboss.com/jira/browse/JBCACHE-1332?page=comments#action_12411132 ] 
            
Nicolas Lebreton commented on JBCACHE-1332:
-------------------------------------------

Hi Manik, 

Yesterday morning, I deactivated the 2nd level cache to confirm that this issue is linked to JBoss Cache. Locks always occurred between 24 to 48h after JBoss recycling. We will wait few more days to be sure.

About your update,
My understanding was that a fix for the issue 1316 was added to 1.4.1 SP9. I downloaded this service pack, and did not see anything about this issue. 
Moreover, I have seen in the JBoss compatibility matrix that JBoss 4.0.5GA is compatible with 1.4.1.SP8, but I guess that this is just a forgetting.

Cheers, 
Nicolas

> Lock in eviction.Region:putNodeEvent()
> --------------------------------------
>
>                 Key: JBCACHE-1332
>                 URL: http://jira.jboss.com/jira/browse/JBCACHE-1332
>             Project: JBoss Cache
>          Issue Type: Bug
>      Security Level: Public(Everyone can see) 
>          Components: Eviction
>    Affects Versions: 1.4.1.SP1
>         Environment: JBoss 4.05
> Hibernate 3.2.6.ga
>            Reporter: Nicolas Lebreton
>         Assigned To: Manik Surtani
>            Priority: Critical
>             Fix For: 1.4.X
>
>         Attachments: 2008-04-10-molene.nexotelematics.com_susan_node2_Percentage db connections in use.jpg, 2008-04-25 locktree.svg, jboss-service.xml, ThreadDumps.zip
>
>
> JBoss TreeCache is used as cache provider. Please find attached jboss-service.xml configuration file.
> The second level cache is actitated, and several read-only classes are cacheable:
> 	<cache usage="read-only"/>
> The query cache is activated, and few queries are cacheable: 
> 	<query name="ActorType.listAll.named" cacheable="true">
> 	
> We encounter an issue. It can be reproduced on an environment with only few users connected (less than 10). For those users, screen is refreshed every 30 seconds through Ajax requests.
> Application is running in a cluster. On the other hand, this issue occurs with only one node started. A JBoss is started. Less than 24 hours later, application is becoming unresponsive. JBoss then needs to be recycled. And it works.
> Before stoping the node, I have taken several thread dumps separated by at least 30 seconds. Please find attached 10th, 11th, 25th's output files. We noticed thread locks in these 3 snapshots. The stack trace of the culprit was always the same (please see attached outputs for more details):
> # java.lang.Object:wait
> # waiting on-->EDU.oswego.cs.dl.util.concurrent.BoundedLinkedQueue(0x9331fd60)
> # java.lang.Object:wait(Object.java:474)
> # EDU.oswego.cs.dl.util.concurrent.BoundedLinkedQueue:put
> # locked-->EDU.oswego.cs.dl.util.concurrent.BoundedLinkedQueue(0x9331fd60)
> # locked-->java.lang.Object(0x9331fe78)
> # org.jboss.cache.eviction.Region:putNodeEvent(Region.java:141)
> # org.jboss.cache.interceptors.EvictionInterceptor:doEventUpdatesOnRegionManager(EvictionInterceptor.java:145)
> # org.jboss.cache.interceptors.EvictionInterceptor:updateNode(EvictionInterceptor.java:118)
> # org.jboss.cache.interceptors.EvictionInterceptor:invoke(EvictionInterceptor.java:93)
> Nicolas

-- 
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