mod_jk error under load

[David Ward <da...@dotech.com>]

In running an automated load test against our app, we get the following 
errors in or jk.log file:

[jk_connect.c (143)]: jk_open_socket, connect() failed errno = 146
[jk_ajp12_worker.c (152)]: In jk_endpoint_t::service, Error sd = -1
[jk_connect.c (143)]: jk_open_socket, connect() failed errno = 146
[jk_ajp12_worker.c (152)]: In jk_endpoint_t::service, Error sd = -1
[jk_connect.c (143)]: jk_open_socket, connect() failed errno = 146
[jk_ajp12_worker.c (152)]: In jk_endpoint_t::service, Error sd = -1
[jk_connect.c (143)]: jk_open_socket, connect() failed errno = 146
[jk_ajp12_worker.c (152)]: In jk_endpoint_t::service, Error sd = -1
[jk_connect.c (143)]: jk_open_socket, connect() failed errno = 146
[jk_ajp12_worker.c (152)]: In jk_endpoint_t::service, Error sd = -1
[jk_connect.c (143)]: jk_open_socket, connect() failed errno = 146

Anyone know what these errors are? I can't find anything conclusive 
about them searching various lists/forums.

We are running Apache 1.3.12 on a Solaris 7 Ultra 10 using mod_jk/ajp12 
to talk to JBoss-2.4.7_Tomcat-3.2.3 on a Solaris 7 E250, JDK 1.3.1_02.

The symptoms we see are that the number of httpd processes maxes out on 
the web server box, though the CPU is almost completely idle. The weird 
thing is that the app server is mostly idle too, and doing a thread dump 
on the java (jboss+tomcat in same vm) process shows that there are lots 
of threads waiting for work to do. Once we stop the load test, things 
are still messed up until I restart apache, then we can access the app 
again. Note that I didn't have to touch the app server at all. Accessing 
URLs that aren't configured to go through mod_jk have no problem, until 
the max http children process gets reached, of course...

I haven't gotten a response from the Tomcat forum at JGuru.

Thanks all,
David


--
To unsubscribe, e-mail:   <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>