[infinispan-issues] [JBoss JIRA] Commented: (ISPN-1255) RequestIgnoredException if a node

Manik Surtani (JIRA) jira-events at lists.jboss.org
Wed Jul 20 08:10:24 EDT 2011


    [ https://issues.jboss.org/browse/ISPN-1255?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12615393#comment-12615393 ] 

Manik Surtani commented on ISPN-1255:
-------------------------------------

This could be because a single-threaded retry queue is used to handle queued RPCs while a node is starting and cannot handle RPCs.  It would make sense to minimise the number of RPCs enqueued by only enqueueing updates (and not things like remote executions), but this needs more thought.

In the meanwhile, does increasing your sync repl timeout help?



> RequestIgnoredException if a node 
> ----------------------------------
>
>                 Key: ISPN-1255
>                 URL: https://issues.jboss.org/browse/ISPN-1255
>             Project: Infinispan
>          Issue Type: Bug
>    Affects Versions: 5.0.0.CR7
>            Reporter: Erik Salter
>            Assignee: Vladimir Blagojevic
>             Fix For: 5.0.0.FINAL
>
>         Attachments: cacheTest.zip, server_node1.log, server_node2.log
>
>
> My application exposes its distributed operations via a REST-based infrastructure.  To minimize the delta between JBoss starting and the cache starting, I used the new Distributed Executor to "sticky" a task to the data owner of a set of keys (with the same hash code). 
> NOTE:  Rehash still causes problems seen in ISPN-1106.  (Attached new logs)
> I see a lot of the following error from the DistributedExecutorService when the new node's cache doesn't start in a timely manner: 
> Reason: java.lang.IllegalStateException: Invalid response {Satriani-52149(PHL)=RequestIgnoredResponse}
> In addition, I see:
> org.infinispan.util.concurrent.TimeoutException: Timed out waiting for valid responses!
> It takes the cache about 2+ minutes at low throughput rate (30 tx/s) to recover.  For high throughput rate, the cluster doesn't recover. 

--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira

        


More information about the infinispan-issues mailing list