Hi Arosso A few problems could cause this. We found it to be one of 3
problems: 1. Network related most often (seeing any broken pipes?), 2.
Sometimes PoolingException: In-use connections equal max-pool-size and
expired max-wait-time. We suspect this may also be caused by a network dip
causing our connections to become invalid and run forever trying to create
new ones. 3. Long running processes on the DB (I think we have fixed most of
these by now). In all of the above cases, the common timing seems to be a
process that takes longer than 15 minutes (900 seconds). This seems to
correlate to (asadmin dotted names): 1.
server.thread-pools.thread-pool.http-thread-pool.idle-thread-timeout-seconds=900
2.
server.network-config.protocols.protocol.http-listener-1.http.request-timeout-seconds=900
(changed this to -1) 3.
server.network-config.protocols.protocol.admin-listener.http.request-timeout-seconds=900
We managed to replicate the problem by simply changing #3 to 10 seconds, and
trying to open the update page from the admin console. It then produces the
same error but on the admin thread pool. This error repeats constantly. ----
When our servers (4) are running, and one of the problems (1-3) occur, the
first thing we notice is an exception matching the problem, then the
"GRIZZLY0023: Interrupting idle Thread: http-thread-pool-8080(37).|#]" about
15 minutes later. We also notice that thread 37 will keep logging to the log
file (it could be any thread number). If the problem (1-3) persists for any
number of seconds, this will effect all our threads (and they all keep
logging the same message). Every time this happens to a thread, it becomes
non-responsive and will not output any data. The client machines calling us
often report no data coming back when they call our web-service. If any
threads made it through the problem, they may still respond, but very often,
if all threads are locked, our server will not respond to any requests and
eventually the log stops... ---- I see now that maybe you have another
problem too, but i suspect the thread death may be effecting you if you are
seeing this particular error. Cheers, Blue
--
[Message sent by forum member 'bluehadlow']
View Post: http://forums.java.net/node/900505