[infinispan-dev] Infinispan added to bench fwk but exception when running test

Manik Surtani manik at jboss.org
Mon Apr 27 11:16:27 EDT 2009


On 27 Apr 2009, at 10:47, Manik Surtani wrote:

>
> On 24 Apr 2009, at 21:25, Galder Zamarreno wrote:
>
>>
>>
>> Galder Zamarreno wrote:
>>> manik at jboss.org wrote:
>>>> Looks like a state transfer sync issue. Just to confirm, does  
>>>> this work if state transfer is disabled? Or, if ST is already  
>>>> disabled, could you try enabling?
>>> I'll verify later today. Thx Manik!
>>
>> Eventually, I got this working. I, naively, pointed to the default  
>> JGroups stacks that come inside jgroups.jar but neither udp.xml nor  
>> tcp.xml had FLUSH, so infinispan would get stuck!
>>
>> It's a very silly configuration mistake but it's taken me ages to  
>> figure out. We should make it more obvious if FLUSH is a must  
>> (regardless of whether state transfer is on or not!)
>
> Hmm, that is interesting - FLUSH should not be necessary since we  
> don't use it for anything.  If the startup does get stuck because  
> FLUSH is missing then that is a bug.
>
> Could you:
>
> 1.  Verify that leaving FLUSH out causes startup to get stuck, and  
> adding FLUSH back in fixes it?
> 2.  Add a sanity check at the end of buildChannel() in  
> JGroupsTransport [1] to ensure FLUSH is there otherwise throw a  
> ConfigurationException
> 3.  Create a JIRA to investigate why FLUSH is needed!  It shouldn't  
> be needed, could be some old JBC code in there (perhaps a flush  
> gate) that is causing the problem.

Never mind about these, I have already fixed this in trunk.
--
Manik Surtani
manik at jboss.org
Lead, Infinispan
Lead, JBoss Cache
http://www.infinispan.org
http://www.jbosscache.org







More information about the infinispan-dev mailing list