RE: apache + 2 tomcat cluster memory problem

2004-03-27 Thread Filip Hanik \(lists\)
there was an earlier email about a mem leak in the connector for mod_jk,
please go through the last weeks emails and let us know what you find out.

Filip

-Original Message-
From: Asim Alp [mailto:[EMAIL PROTECTED]
Sent: Saturday, March 27, 2004 7:55 PM
To: Tomcat Users List
Subject: apache + 2 tomcat cluster memory problem


Hello everyone,

I have 2 Tomcats (both 5.0.19) running on two separate machines.  One
machine also has Apache 2.0.49 installed on it for load balancing.
Apache communicates with tomcat1 thru port 11005 and tomcat2 thru
12005.  tomcat1 and Apache are on the same machine.  Apache is set to
forward /* to the load balancer (and then to Tomcats).  We have over a
hundred virtual hosts on our tomcats, so we don't want to worry about
configuring Apache virtual hosts to serve the images and all the other
files (that's why we forward /*, not just /*.jsp and /servlet/*).  Both
tomcats also have their http connector ports listening to port 8080 on
their individual machines.  We're using mod_jk_1.2.5_2.0.47.dll for
Apache to Tomcat communication...  Both our tomcats are running with
the -Xms128m -Xmx256m options.

With this configuration everything works perfect for about an hour,
then our tomcats run out of memory...

When we turn off clustering, and run a single Tomcat on port 80 (of
course by shutting down Apache Httpd), we don't have any memory
problems at all (for now, we're planning to use clustering to avoid
down time, one server is actually enough to handle all the traffic).

Does anyone have any idea why we'd run out of memory when our tomcats
are talking to Apache through mod_jk?

Thanks,

Asim


-
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
---
Incoming mail is certified Virus Free.
Checked by AVG anti-virus system (http://www.grisoft.com).
Version: 6.0.614 / Virus Database: 393 - Release Date: 3/5/2004

---
Outgoing mail is certified Virus Free.
Checked by AVG anti-virus system (http://www.grisoft.com).
Version: 6.0.614 / Virus Database: 393 - Release Date: 3/5/2004


-
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]



Re: apache + 2 tomcat cluster memory problem

2004-03-28 Thread Asim Alp
Thanks Filip.  I only found the thread of mails starting March 17 with  
the title "Horrible memory leak in tomcat 5.0.19"  If this is indeed  
the mails you advised me to look at, here is what I found out:

Our setup differs in a way that we are using mod_jk 1.2.5 instead of  
2.0.2.  So mod_jk 1.2.5 + Tomcat 5.0.19 has a memory leak problem as  
well.  This suggests that maybe the problem is with Tomcat 5.0.19 (?)   
Appearantly someone else (Robert Kruger) had exactly the same problem  
and it was solved when they downgraded to Tomcat 5.0.18.  We cannot do  
that either because of a bug fix regarding session states in 5.0.18.  I  
don't mind using mod_jk 2.0.2 as long as it solves our problem; however  
there already seems to be a memory leak with the mod_jk 2.0.2 + tomcat  
5.0.19 combination as well.

In Robert's email, he predicts that maybe the problem is related to JMX  
registration issue described in  
http://www.mail-archive.com/[EMAIL PROTECTED]/ 
msg53035.html.  I'll reply to his email to see if it was indeed the  
problem.

What's the next step?

Thanks,

Asim

On Mar 28, 2004, at 12:39 AM, Filip Hanik ((lists)) wrote:

there was an earlier email about a mem leak in the connector for  
mod_jk,
please go through the last weeks emails and let us know what you find  
out.

Filip

-Original Message-
From: Asim Alp [mailto:[EMAIL PROTECTED]
Sent: Saturday, March 27, 2004 7:55 PM
To: Tomcat Users List
Subject: apache + 2 tomcat cluster memory problem
Hello everyone,

I have 2 Tomcats (both 5.0.19) running on two separate machines.  One
machine also has Apache 2.0.49 installed on it for load balancing.
Apache communicates with tomcat1 thru port 11005 and tomcat2 thru
12005.  tomcat1 and Apache are on the same machine.  Apache is set to
forward /* to the load balancer (and then to Tomcats).  We have over a
hundred virtual hosts on our tomcats, so we don't want to worry about
configuring Apache virtual hosts to serve the images and all the other
files (that's why we forward /*, not just /*.jsp and /servlet/*).  Both
tomcats also have their http connector ports listening to port 8080 on
their individual machines.  We're using mod_jk_1.2.5_2.0.47.dll for
Apache to Tomcat communication...  Both our tomcats are running with
the -Xms128m -Xmx256m options.
With this configuration everything works perfect for about an hour,
then our tomcats run out of memory...
When we turn off clustering, and run a single Tomcat on port 80 (of
course by shutting down Apache Httpd), we don't have any memory
problems at all (for now, we're planning to use clustering to avoid
down time, one server is actually enough to handle all the traffic).
Does anyone have any idea why we'd run out of memory when our tomcats
are talking to Apache through mod_jk?
Thanks,

Asim

-
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
---
Incoming mail is certified Virus Free.
Checked by AVG anti-virus system (http://www.grisoft.com).
Version: 6.0.614 / Virus Database: 393 - Release Date: 3/5/2004
---
Outgoing mail is certified Virus Free.
Checked by AVG anti-virus system (http://www.grisoft.com).
Version: 6.0.614 / Virus Database: 393 - Release Date: 3/5/2004
-
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]