Loadbalance with mod_jk performance

[J Hartley]

Hi,
We have mod_jk_1.2.23 on apache 2 load balancing a cluster of 2 Tomcat 5.5.23 instances.

We are load testing with JMeter. After around 9000 requests to our webapp the load balancer slows down significantly and eventually grinds to a complete halt. No more requests will be routed to the tomcat instances. We know mod_jk is the issue, because when it gets into this state we are still able to access pages served by apache without going through the LB. The tomcats are also responisve when accessed directly.

Has anyone experienced this? Or have any idea how we can resolve this?
Thanks
Joe

[Ramesh Ch]

could you please provide mod_jk.conf used in this ?

[bob stiles]

We have the same issue!

We are using grinder.

This is our setup:
4 apaches (2.0.53)
8 jboss servers (4.0.5, tomcat 5.5)
mod_jk (using ajp13, version 1.2.22)

We have tried everything. Need help.

[Ramesh Satya Venkata]

can you paste your contents of jk.conf ?

here you can find sample jk.conf below

LoadModule jk_module modules/mod_jk.so
JkWorkersFile /etc/httpd/conf/workers.properties
JkLogFile /var/log/httpd/mod_jk.log
JkLogLevel info
JkLogStampFormat "[%a %b %d %H:%M:%S %Y] "
JkOptions +ForwardKeySize +ForwardURICompat -ForwardDirectories
JkRequestLogFormat "%w %V %T"
JkMount /*.jsp worker1
JkMount /*.htm worker1