Hello,
We have 3 servers on debian 4.0, and on each server an apache 2.0 and a JBoss 4.0.5, each apache sending some of the request to a JBoss via AJP (mod_jk 1.2) . An application deployed in a 3 node JBoss cluster. Each JBoss receives requests preferentially from the apache on the same server (we put a heavy lbfactor for the jboss on the same server).
The cluster is OK, and the session are well replicated, every JBoss is seeing the others, and no problem with stopping and starting anyone of them without losing the users sessions data.
But after running for a few hours with about 30 users, one of the server (never the same) stops responding to requests from new clients, but appears to continue to responde to requests from some clients... at least for a while.
There is no OutOfMemoryError in the log file, not even any Exception.
The only solution is to stop Jboss which is not responding and restart it...
For info, we had some errors saying to increase the maxThreads in the ajp connector. It was set to the default value (200), we increased it to 400, kept having the error, increased it to 500 and now the error does not appear anymore. I read we should increase also the MaximumPoolSize in jbos-service.xml, which we did (from the default value 10 to even 50, now it is set to 20), with no improvement...
Thank you for any idea that might help us...
