This sounds like the bug I fixed in Tomcat 4.1.x where infrequently I saw a POST request put Tomcat into an infinite loop. We run Apache and Tomcat on different servers. I had noticed the increased CPU usage on the Tomcat server but hadn't noticed the increased CPU usage on the Apache server. I went back and reviewed our system load history and verified that when this bug is triggered both Apache and Tomcat cause increased CPU usage.
This bug was fixed in the Tomcat 4.1.x branch and I ported it back to the Tomcat 4.0.x branch. This bug fix should be in the Tomcat 4.0.5 release. Regards, Glenn Chris Read wrote: > Greetings... > > Have you also noticed a spike in CPU usage on your Apache server running > mod_jk, or your network throughput? > > The reason I ask is we've got a similar problem here. We've got a slightly > different setup (Apache 1.3.x, mod_jk, Tomcat 4.0.3, Sun JDK 1.3.1 - all on > Solaris), but the same symptoms. > > Basically what we're seeing is sometimes mod_jk and the AJP13 connecter get > stuck in an infinite loop spewing data at each other as fast as possible > (and always the same data: Tomcat asking for a BODY_CHUNK of 8186 bytes, and > mod_jk replying with an empty packet). > > The problem will only affect one connection at a time from the pool, but > will slowly step up one connection at a time. Restarting either Tomcat or > Apache solves the problem, but I've not been able to find any info on this > either. I've been trying for a few days to replicate the problem on demand, > but no luck yet.... > > Any of this sound familiar? > > Chris -- To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]> For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>
