Vadim and Berin,
Thank you both for your assistance. We got past this. The problem was
actually simple. The XMLFileModule was configured with reloadable false
and cacheable false. Since this file is read at the beginning of every
user session it was causing a huge bottleneck. After setting both to
true we actually started seeing pool sizes that were too low and then
after they were corrected we ended up CPU bound. The rates were good
but not spectacular.
However, I still have a feeling that holding the lock while checking the
validity will create some contention in our usage since multiple threads
do want to read the file simultaneously.
If you want to fly out to L.A. I'll happily buy you a beer! :-)
Ralph
Vadim Gritsenko wrote:
Vadim Gritsenko wrote:
"http-8080-Processor17" daemon prio=1 tid=0x2e3d58c8 nid=0x19eb
waiting for monitor entry [2d7f3000..2d7f587c]
at
org.apache.avalon.excalibur.pool.ResourceLimitingPool.get(ResourceLimitingPool.java:262)
- waiting to lock <0x60088180> (a java.lang.Object)
