[jbosscache-dev] Stress testing non-blocking state transfer

[Manik Surtani]

This is a quick status update for those interested.

We've been running some tests with NBST and overcame initial  
weirdnesses with the HTTP session replication stress test, mainly  
around config issues (not using MVCC, not using  
STREAMING_STATE_TRANSFER in JGroups).  These are now tested for in the  
JBC configs.

Anyway, the latest run went somewhat well, in that the basic state  
transfer worked.  The test involved async replication, with 4 nodes.   
Here is what happened:

1.  4 nodes started.  State transferred accordingly, no problems.
2.  Load applied to all servers via a load balancer.
3.  Node 4 killed.
4.  Node 3 killed.
5.  Node 2 was observed to throw an OOM at this point.

Examining what happened on Node 2 at this stage, stay tuned for more  
info.

Cheers
--
Manik Surtani
Lead, JBoss Cache
http://www.jbosscache.org
manik at jboss.org




-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.jboss.org/pipermail/jbosscache-dev/attachments/20090225/a281f5ba/attachment.html