I'd try to add a retry to you ProxyPass (retry=60) also try to increase MaxClients.
hope this helps... -- Staf Wagemakers - http://www.wagemakers.be On Dec 10, 2007 9:24 PM, Ângelo Vimeney <[EMAIL PROTECTED]> wrote: > Hi, > > Thanks for your analysis, Karel. I'll try a larger timeout (60 seconds > was used in my unsuccessfully last test). However, recently I put 300 > browsers to access Apache connect to Tomcat using mod_jk without any > problem. Unfortunately, I suspect this could be a bug with > mod_proxy... > > Ângelo > > On Dec 7, 2007 12:19 PM, Karel Kubat <[EMAIL PROTECTED]> wrote: > > -----BEGIN PGP SIGNED MESSAGE----- > > Hash: SHA1 > > > > Hi, > > > > > At the Tomcat's catalina.out I found the following log: > > > ... > > > Dec 4, 2007 4:47:49 PM org.apache.catalina.startup.Catalina start > > > INFO: Server startup in 3549 ms > > > Dec 4, 2007 4:53:51 PM org.apache.jk.core.MsgContext action > > > WARNING: Error sending end packet > > > java.net.SocketException: Broken pipe > > > > This might indicate that your Tomcat server took too long to produce > > its answer. By the time that it started writing its response, the > > network connection was gone. This might be caused because your Apache > > proxy "hung up" already. > > > > Check your Apache proxy for timeouts and choose a longer time. Then > > re-test. If the problem occurs with a higher load than the 50 users > > that you simulated, then you're on to something - in that case, > > Tomcat would be the bottleneck, because the processing times increase > > beyond the proxy's timeout. > > > > Hope this helps, > > -- > > Karel Kubat / M +31 6 2956 4861 (+31 6 AWK 6 HUM 1) > > From the list of "Things You'll Never Hear A Southern Male Say": > > I'll take Shakespeare for 1000, Alex. > > >
