Hi -
We have implemented mod_jk load balance and transport solution for our client. There is one configuration that is yet to be 100% correct from my angle.
We have 5 Apache webservers talking to 12 Jboss (4.0.4) servers using mod_jk (1.2.19).
What happens is that the JBoss is 100% available, but sometimes under jkmanager, I see a specific Jboss in ERR state.
My retry at the socket is 2 (so even if the first socket was closed due to connection_timeout at JBoss Layer or any other timeout, mod_jk it will try again).
Although I feel this config is correct, I still see JBoss's in ERR state (at 1 time, I might see 1 in ERR), so the frequency is not too high.
Is there anything I can do, to make JBoss's not go into ERR at all. My config is as follows
worker.template.socket_timeout=5
worker.template.retries=2
worker.template.socket_keepalive=true
worker.template.connection_pool_timeout=300
worker.template.connect_timeout=1000
worker.template.prepost_timeout=1000
worker.template.reply_timeout=180000 (its a high value, we will bring it down when the website is stable)
worker.lbworker.recover_time=300
worker.retries=3
At APJ Level (in JBoss, I have connection_timeout set to 1 hour)
Once I solve this issue, I would have implemented a very High Volume website using Mod_jk.
