Apache mod_jk2 fizzles out

[James Courtney <Jc...@inphonic.com>]

We recently upgraded our production systems to Tomcat 4.1.24 and Apache 2.0.45 from 3.2.2 and 1.3.19 respectively and have noticed a VERY pleasing increase in performance and over an 80% reduction in system load.  We also decided to deploy with mod_jk2 and the whole system behaved itself quite well during QA and some load testing (admittedly not overly thorough).  We pushed to production and things ran fine for a couple of days but now, every day or two, our Apache partially loses communication with one of the two Tomcats it's load balancing accross (not always the same one).  The result is that requests block on Apache and the load goes up as users click away trying to get their page(s) to load.  Restarting the Tomcat seems to rectify the problem until the next time...

Any thoughts on this problem and the overall maturity of mod_jk2 vs. mod_jk would be very helpful.  Many thanks to all!


*******************
** Apache Errors **
*******************
inunison.com:8081 145 Connection timed out [Wed Jun 25 15:08:10 2003] [error] ajp13.connect() failed ajp13:portal4.somedomain.com:8081 [Wed Jun 25 15:08:10 2003] [error] ajp13.service() failed to connect endpoint errno=145 Connection timed out [Wed Jun 25 15:08:10 2003] [error] ajp13.service() Error forwarding ajp13:portal4.somedomain.com:8081 1 1 [Wed Jun 25 15:08:10 2003] [error] lb.service() worker failed 120000 for ajp13:portal4.somedomain.com:8081 [Wed Jun 25 15:08:13 2003] [error] channelSocket.open() connect failed portal4.somedomain.com:8081 145 Connection timed out [Wed Jun 25 15:08:13 2003] [error] ajp13.connect() failed ajp13:portal4.somedomain.com:8081 [Wed Jun 25 15:08:13 2003] [error] ajp13.service() failed to connect endpoint errno=145 Connection timed out [Wed Jun 25 15:08:13 2003] [error] ajp13.service() Error forwarding ajp13:portal4.somedomain.com:8081 1 1 [Wed Jun 25 15:08:13 2003] [error] lb.service() worker failed 120000 for ajp13:portal4.somedomain.com:8081 [Wed Jun 25 15:08:22 2003] [error] channelSocket.open() connect failed portal4.somedomain.com:8081 145 Connection timed out [Wed Jun 25 15:08:22 2003] [error] ajp13.connect() failed ajp13:portal4.somedomain.com:8081 [Wed Jun 25 15:08:22 2003] [error] ajp13.service() failed to connect endpoint errno=145 Connection timed out [Wed Jun 25 15:08:22 2003] [error] ajp13.service() Error forwarding ajp13:portal4.somedomain.com:8081 1 1 [Wed Jun 25 15:08:22 2003] [error] lb.service() worker failed 120000 for ajp13:portal4.somedomain.com:8081

****************************
** WORKERS2 CONFIGURATION **
****************************
# only at beginning. In production uncomment it out
[logger.apache2]
level=ERROR

[shm]
file=/tmp/workers2shmDONOTDELETE.file
size=1048576

# portal3 socket
[channel.socket:portal3.somedomain.com:8081]
host=portal3.somedomain.com
port=8081
tomcatId=portal3
group=lb:balanced
graceful=1

# portal3 worker
[ajp13:portal3.somedomain.com:8081]
channel=channel.socket:portal3.somedomain.com:8081
lb_factor=12

# portal4 socket
[channel.socket:portal4.somedomain.com:8081]
host=portal4.somedomain.com
port=8081
tomcatId=portal4
group=lb:balanced

# portal4 worker
[ajp13:portal4.somedomain.com:8081]
channel=channel.socket:portal4.somedomain.com:8081
lb_factor=12


# Load balanced worker
[lb:balanced]
worker=ajp13:portal3.somedomain.com:8081
worker=ajp13:portal4.somedomain.com:8081
# sticky is on by default but setting it adds a harmless error to the log
#stickySession=1


# Define a status worker
[status:status]


# URI mapping
[uri:/*]
group=lb:balanced


# Status URI mapping (should not be publicly accessible!!!)
[uri:/jkstatus]
group=status:status
 


---------------------------------------------------------------------
To unsubscribe, e-mail: tomcat-user-unsubscribe@jakarta.apache.org
For additional commands, e-mail: tomcat-user-help@jakarta.apache.org