I don't recognize the problem, but AJP12 was deprecated even for 3.3.1a (which, for Pid's benifit, was a single security issue bugfix release from 3.3.1). In any case, it will be a bottleneck for any attempt to scale your application. You should try with the AJP13 connector, which works better on 3.3.1a then either the AJP12 connector or the HTTP/1.0 connector (the HTTP/1.0 connector in particular is known to have issues on Windows). If you can set up a test environment, the other thing to try is 3.3.2 with the CoyoteConnector (either AJP13 or HTTP/1.1). Short-term it is probably less painful than upgrading to TC 5.5.x or TC 6.0.x.
Additional information I'd like to see is the mod_jk version (AJP12, while still shipping with the most recent mod_jk hasn't been supported in a very long time. It wouldn't surprise me at all if it is broken on recent versions), and a thread dump if you can get one. Also turning up your JkLogLevel to debug on mod_jk might help. As Mark noted, 3.3.x is still somewhat officially supported (although I have trouble seeing what would trigger a 3.3.3 release, even if it does have some goodies in it :). If you can get the thread dump and/or the debug mod_jk logs, it may be better if you open a BZ report and attect them there, so they are easier for me to find. "Marko Krejic" <[EMAIL PROTECTED]> wrote in message news:[EMAIL PROTECTED] Hi, We are running Tomcat 3.3.1a on a Windows Server 2003. A apache server is also installed on the machine as a proxy and it is connected with the tomcat through AJP12. Tomcat is running on JDK1.4.2_13. Now to our problem. We have been running this setup for quite a while (a few years). But until recently, the load has increased. It seems like when there are quite many connections, then suddenly it is not possible to read the requestbody anymore. We get the request and we are seeing the request header, but when we want to read the body, it all seems to hang... At this time it seems like the threads are just stacking up and nothing is processed. Other threads, such as timer-threads and file-listener threads, are running normally. It seems like only the request comming through the connectors (the AJP12 and HTTP) are hanging. Does anybody recognise these problems? And does anybody have any suggestions about how to fix this? We have tried to run without Apache and AJP12, but we got the same problem. I could mention that we know that we have mobile clients sending requests to the server and if the server is not responding within a certain time, they will "cut-off" the request. Could it be that tomcat is not "releasing" these requests? Please, any suggestions would be very apprieciated! Thanks in advance! Marko Krejic --------------------------------------------------------------------- To start a new topic, e-mail: users@tomcat.apache.org To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
