On 03/01/2010 11:09, assan alhamoud wrote: > Happy new year to all , > > I am using tomcat and JBoss.
You'll need to explain that further. Why would you use Tomcat and JBoss for the same app. I would expect one or the other. > Tomcat/5.0.26 That is very, very old and has many known bugs and a number of security vulnerabilities. I'd strongly suggest moving to 5.5.28. > mod_jk/1.2.26 Also worth upgrading to the latest version To get back to your original question, it looks like your app could benefit from some tuning. The hanging is likely to be caused by long running requests using up all the connections between httpd and Tomcat and/or poor mod_jk / AJP connector configuration. Since you haven't shown us this configuration it is hard to provide specific advice but generally I would suggest: - use JkOptions +DisableReuse (or sort out your timeouts but my experience has been disable reuse is simpler to configure and has similar performance) - turn on access logging for Tomcat - check any timeouts are consistent with your longest running requests - ensure the following are compatible - httpd processes/threads - Tomcat AJP connector threads - mod_jk connection pooling Mark > > Regards, > > 2009/12/29 Mark Thomas <ma...@apache.org> > >> On 29/12/2009 16:17, assan alhamoud wrote: >>> Hello Mark , >>> >>> we are using apache Server version: Apache/2.2.11 (Unix) >>> we use mod_jk.so module >>> jboss-3.2.5 >>> Tomcat5 >>> and jrockit-jdk1.4.2_19 >>> >>> at your disposal if you need more information >> >> Are you using Tomcat or JBoss? >> Tomcat version? >> mod_jk version? >> >> Mark >> >>> >>> Regards, >>> Ghassan >>> >>> 2009/12/29 Mark Thomas <ma...@apache.org> >>> >>>> On 29/12/2009 15:21, assan alhamoud wrote: >>>>> Hello All >>>>> >>>>> I have some questions and I hope to find answers about Load balancing >> . >>>>> >>>>> 1- The first question , our DBA says that it not normal and no logic >> to >>>>> have several instances of tomcat server on the same machine with the >>>> same >>>>> configuration , is this correct ? >>>> No. Your DBA is wrong. >>>> >>>>> 2- Each tomcat instance here can serve about 125 user , and I have more >>>> than >>>>> this number on the server I got alot of problems and usually the >>>> instance >>>>> hang , is there anyway to get better performance from tomcat. >>>> That isn't a Tomcat problem, that is an application problem. The >>>> solution is to fix/tune your application. >>>> >>>>> I am using now 3G memory for each instance and 120 for max_pool_size >>>> and >>>>> 750 maxThread >>>> It isn't clear what the 125 users refers to. Is it concurrent requests, >>>> is it concurrent sessions, is it concurrent users (assuming 1 concurrent >>>> user == more than 1 concurrent request). From these numbers the >>>> application seems very resource intensive. It does appear that it would >>>> benefit from some tuning. >>>> >>>>> 3- The most critical issue I had , when one instance hangs , apache >>>> hangs >>>>> and then all the application hangs because apache keep trying to >> connect >>>> to >>>>> the hanged instance , Is there any way to force apache to bypass the >>>> hanged >>>>> instance. >>>> Yes. But since you haven't told us the Tomcat version, httpd version nor >>>> how you are connecting httpd to Tomcat we can't tell you how to fix it. >>>> >>>>> 4- When instance of tomcat hangs , i forced to restart this instance ( >>>>> tomcat server ) the problem is GC can not free memory anymore >>>>> Is there anyway to solve the problem instead of restarting the >>>> server. >>>> Yes. Work out if you have a memory leak or need more memory than is >>>> available. If it is a memory leak, fix it. If you need more memory >>>> either allocate it or tune your app so it uses less. >>>> >>>>> 5- at last sometimes I found that the same action repeated many times >>>> which >>>>> means that apache sends the request more than one time to tomcat server >>>>> ,could I have control over this behavior. >>>> Maybe, maybe not. If this behaviour is initiated by the user agent you >>>> need to manage this in your app. It could also be caused by the >>>> fail-over mechanism, in which case I refer you to my answer to point 3 >>>> above. >>>> >>>> Mark >>>> >>>> >>>> >>>> --------------------------------------------------------------------- >>>> To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org >>>> For additional commands, e-mail: users-h...@tomcat.apache.org >>>> >>>> >>> >>> >> >> >> >> >> --------------------------------------------------------------------- >> To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org >> For additional commands, e-mail: users-h...@tomcat.apache.org >> >> > > --------------------------------------------------------------------- To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org For additional commands, e-mail: users-h...@tomcat.apache.org
