[infinispan-issues] [JBoss JIRA] (ISPN-2197) Cache view instalation takes too long

[Dan Berindei (JIRA)]

     [ https://issues.jboss.org/browse/ISPN-2197?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Dan Berindei resolved ISPN-2197.
--------------------------------

    Fix Version/s: 5.1.7.Final
                   5.2.0.Alpha3
       Resolution: Done


I believe this was caused by ISPN-2186: when the coordinator shuts down, it tries to install a new cache view that doesn't include itself. Because there are multiple nodes shutting down at the same time, two or more nodes can see themselves as the coordinator and will try to install conflicting views, failing and repeating the process for a very long time.

The ISPN-2186 fix changes this by automatically rejecting outside cache views installation requests on the current JGroups coordinator. This way each one of the "coordinators" can shut down without interfering with the others.

                
> Cache view instalation takes too long
> -------------------------------------
>
>                 Key: ISPN-2197
>                 URL: https://issues.jboss.org/browse/ISPN-2197
>             Project: Infinispan
>          Issue Type: Bug
>          Components: State transfer
>            Reporter: Ondrej Nevelik
>            Assignee: Dan Berindei
>             Fix For: 5.1.7.Final, 5.2.0.Final, 5.2.0.Alpha3
>
>
> ISPN000172: Failed to prepare view CacheView{viewId=14, members=[edg-perf05-63964]} for cache  t, rolling back to view CacheView{viewId=11, members=[edg-perf05-63964]}
> java.util.concurrent.ExecutionException: java.lang.IllegalStateException: t: Received cache view prepare request after the local node has already shut down.
> When testing configuration [1] with lots of caches, it takes too long for the caches to shut down - ~20minutes! See the log [2] for more details.. More specifically we are calling cacheManager.stop() method and then waiting for the cacheManager.getMembers() to be <= 0.
> The Infinispan used is the one in JDG 6.0.1.ER2.
> [1] https://svn.devel.redhat.com/repos/jboss-qa/load-testing/etc/jdg-60/radargun/comparison/dist-sync-tx-multicache.xml
> [2] https://jenkins.mw.lab.eng.bos.redhat.com/hudson/view/EDG6/view/JDG-RADARGUN/job/jdg-60-radargun-perf-multicache/24/console-edg-perf05/

--
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