On 28/07/2011 09:26, Robert Elliot wrote: > ----- Original Message ----- >> From: "Pid" <p...@pidster.com> >> To: "Tomcat Users List" <users@tomcat.apache.org> >> Sent: Thursday, 28 July, 2011 9:07:13 AM >> Subject: Re: Wrong Response returned for a Request >> >>>>> On 27/07/2011 17:23, Robert Elliot wrote: >>>>>> We are using Tomcat 6.0.29, and occasionally experiencing >>>>>> concurrency issues >>>>>> which look very much like the wrong response is being returned >>>>>> for a request. >>>>>> While we cannot reproduce it reliably, it seems to happen when >>>>>> Keep-Alive is turned on. >> >> In most cases this symptom turns out to be a result of the session, >> request or response object being retained (perhaps as an instance >> field >> in a Servlet) where they should not be. >> >> Can you double check your code to see if this is possible? >> > > Unfortunately it's a third party application we're running.
Ah. > However as a gut feeling I'd expect a straightforward bug like that to be > easier to reproduce; we've spent a day chucking between 100 and 1200 > concurrent clients at it and we only see the problem very occasionally. Try reducing the number of connector threads, try 10 and work down from there. > Oddly once it has happened it keeps happening for a period - and then stops. Maybe not. Speculation: once the JVM is hot, JIT optimisation might reduce the window size further. > I think we need to think harder about how to diagnose it - at the moment we > can't pin it to a specific layer, it could be the application (most likely), > Tomcat, mod_proxy or Apache HTTPD (we're running 2.2.3). I only asked here as > the bug sounded so like what we were seeing. Fair enough. p --------------------------------------------------------------------- To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org For additional commands, e-mail: users-h...@tomcat.apache.org
