james-server-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Jason Webb" ...@inovem.com>
Subject RE: High CPU Load Followup
Date Wed, 27 Aug 2003 12:05:19 GMT
I've used HP Jmeter to track down a leak in James before.
Its not bad and can give some very useful insight into your code.
I found the Sun one not so nice interface wise and not as rich in the
analysis of the information as the HP.

-- Jason

> -----Original Message-----
> From: Steve Brewin [mailto:sbrewin@synsys.com] 
> Sent: 27 August 2003 12:53
> To: 'James Developers List'; sbrewin@synsys.com
> Subject: RE: High CPU Load Followup
> 
> 
> Kenny,
> 
> Assuming you don't want to splash out on a commercial 
> performance analysis tool, you could also try this...
> 
> http://www.hp.com/products1/unix/java/hpjmeter/infolibrary/jmeter/
> 
> or this...
> 
> http://developer.java.sun.com/developer/technicalArticles/GUI/
> perfanal/
> 
> Both are free. I haven't tried either myself, but they should 
> shed some light on where the CPU time is being spent within 
> the JVM without extensive analysis of thread dumps.
> 
> If you do try one, it would be interesting to know if you 
> found them useful.
> 
> -- Steve
> 
> > -----Original Message-----
> > From: Steve Brewin [mailto:sbrewin@synsys.com]
> > Sent: 27 August 2003 10:29
> > To: 'James Developers List'
> > Subject: RE: High CPU Load Followup
> >
> >
> > Kenny,
> >
> > As described in 
> > http://developer.java.sun.com/developer/technicalArticles/Prog
> > ramming/Stackt
> > race/...
> >
> > "For busy programs: The best course of action you can take for busy 
> > programs is to generate frequent stack traces. This will narrow down
> > the code path
> > that is causing the errors, and you can then start your
> > investigation from
> > there."
> >
> > For "frequent", read 10 at 10 second intervals as a starting point. 
> > You could write a script that generates these "frequent" thread
> > dumps for you.
> > With "frequent" thread dumps you should be able to determine
> > if James is
> > always looking up MX records, as shown in the stacktrace you
> > attached, or
> > typically doing something else.
> >
> > Just to confirm, when you say "James is running above 90% 
> CPU", what 
> > do you mean? Is the process that JVM on which James is running using
> > 90% CPU or is
> > the total CPU usage at 90%? If the latter, what other processes are
> > increasing their CPU consumption?
> >
> > -- Steve
> >
> > > -----Original Message-----
> > > From: Kenny Smith [mailto:james-list@journalscape.com]
> > > Sent: 27 August 2003 00:33
> > > To: James Developers List
> > > Subject: High CPU Load Followup
> > >
> > >
> > > Hi all,
> > >
> > > Thanks to a pointer from Steve Brewin I was able to get 
> some thread 
> > > dumps while James is running above 90% CPU. I took 4 
> thread dumps: 3 
> > > within a minute and 1 an hour later. It looks like there are 4 
> > > threads that are runnable. 3 that are waiting for socket 
> connections
> > > (I assume
> > > that's SMTP, POP, and the Admin Manager), the last 1 is
> > doing DNS for
> > > remote delivery. Here's the stack:
> >
> >
> > 
> ---------------------------------------------------------------------
> > To unsubscribe, e-mail: server-dev-unsubscribe@james.apache.org
> > For additional commands, e-mail: server-dev-help@james.apache.org
> >
> 
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: server-dev-unsubscribe@james.apache.org
> For additional commands, e-mail: server-dev-help@james.apache.org
> 
> 


---------------------------------------------------------------------
To unsubscribe, e-mail: server-dev-unsubscribe@james.apache.org
For additional commands, e-mail: server-dev-help@james.apache.org


Mime
View raw message