[Deploying Sakai] Server slowdown

[mizematr at notes.udayton.edu]

It is also possible that your permgen space is set too low, 128m seems 
borderline to me.  If this fills up, Java could start constantly garbage 
collecting which would slow down the Application without necessarily 
pegging the processor.  We had this happen to us once or twice.  I would 
install jvmstat, http://java.sun.com/performance/jvmstat/, and use the 
visualgc tool that comes with it to see how close to full the permgen 
space is.  If it's close to full, I'd try upping the max size.  We 
currently have "-XX:PermSize=512m -XX:MaxPermSize=512m" in our JVM 
settings.
---------------------------------------------------------------------------
Matt Mize, Systems Administrator
 - Pay no attention to the man in the back office....
Matt.Mize at notes.udayton.edu
(937) 229-1024

UDit Department, University of Dayton
300 College Park, Dayton, OH, 45469-1302

production-bounces at collab.sakaiproject.org wrote on 04/28/2010 04:50:16 
AM:

> On 27 April 2010 22:31, Grossman,John E <john.grossman at mdanderson.org> 
wrote:
> > One of our Sakai production servers slowed to a near standstill with 
about
> > 90 users logged in and taking final exams. We are running Sakai 2.6.0 
on a
> > 4-processor 64-bit Windows machine with 8 GB of RAM with 5 GB 
allocated to
> > the JVM. The chain of errors started with those in the attached zip 
file.
> > Eventually no one could log in and it would take 10-15 minutes for a 
user
> > interface action such as a button click to take effect. Processor
> > utilization on the server was 0-2% during this episode and there was 
about
> > 200 MB of available memory according to the Memory tool in the admin
> > workspace. Would appreciate any suggestions.
> 
> I think the errors in your attached file could well be related to this
> issue in JIRA:
> 
> http://jira.sakaiproject.org/browse/SAK-6553
> 
> I don't have any good ideas as to why the server became slow, if there
> the CPU load is low and there isn't much disk activity locally I'd
> look for exhaustion somewhere:
> 
> - Database connection pool.
> - Tomcat worker threads.
> 
> Do you have the DB on a separate machine?
> Do you have more than one tomcat worker node?
> Is so what's doing your frontend load balancing?
> 
> -- 
>   Matthew Buckett
>   VLE Developer, LTG, Oxford University Computing Services
> _______________________________________________
> production mailing list
> production at collab.sakaiproject.org
> http://collab.sakaiproject.org/mailman/listinfo/production
> 
> TO UNSUBSCRIBE: send email to production-unsubscribe at collab.
> sakaiproject.org with a subject of "unsubscribe"
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://collab.sakaiproject.org/pipermail/production/attachments/20100428/023b16c2/attachment.html