I can't say I know FOR SURE that the threads were dead, but after a
considerable length of time they never accumulated any more CPU usage
time (as seen via top or ps) whereas some of the others did. I know for
sure I had far more threads than the max_threads setting (12 for me)
because I'd see in excess of 80 threads for tomcat on the system after
several days uptime. I might try upping the max_threads to see if that
helps. We were trying to limit concurrent usage of the box (via tomcat)
because other apps use the same box and we didn't want to use too many
resources. That said, we'd rather have a stable tomcat and overload the
box a bit on occasion :) than an unstable/really slow tomcat. -Devon
----------------
Devon C. Ziegler
[EMAIL PROTECTED]
2504 Preston Lake Dr.
Tucker, GA 30084
770-908-9462
-----Original Message-----
From: Thom Park [mailto:[EMAIL PROTECTED]]
Sent: Monday, May 21, 2001 7:25 PM
To: [EMAIL PROTECTED]
Subject: Re: 3.2.1 Dies
I wasn't able to check for dead threads - My symptom was that the tomcat
process
would 'hang' and no longer process connections.
I never got it to crash per se, but it sure stopped on me real good...
Increasing the pool size resolved the problem, specifically the
following
parameters (neatly copied from the tomcat users guide)
<Parameter
name="max_threads"
value="30"/>
<Parameter
name="max_spare_threads"
value="20"/>
<Parameter
name="min_spare_threads"
value="5" />
The values were arbitrary - I stuck mine up to max_threads = 200,
max_spare_threads=20, min_spare_threads=20
and lo - the hang problem went away.
It seems to me that when tomcat (3.2.1) get's pressed for thread
resources, it
doesn't do a very good job of recovering from this.
I would have thought that having a min_spare_threads set to some
non-zero value,
that there would still be some spare threads for clean up
work to be performed but, as far as I can tell, when you max out your
threads,
your stuck pending a reboot.
-Thom
Hunter Hillegas wrote:
> How can I check to see if I have a lot of "dead threads"?
>
> Hunter
>
> > From: "Devon Ziegler" <[EMAIL PROTECTED]>
> > Reply-To: [EMAIL PROTECTED]
> > Date: Mon, 21 May 2001 16:43:04 -0400
> > To: <[EMAIL PROTECTED]>
> > Subject: RE: 3.2.1 Dies
> >
> > Interesting.....used up all free threads.....increase thread
pool.....
> > shouldn't users just have to wait for a free thread if they hit the
pool
> > limit? Increasing it should let more requests be handled
simultaneously,
> > but how would it help longevity? I'm not saying it won't. Just
skeptical.
> >
> > That said, I seem to "lose" threads with tomcat 3.2.1 on Linux
2.2.14 using
> > Sun's 1.3 JDK. What I mean is that "dead" threads pile up WELL in
excess of
> > the thread pool limit. These threads don't seem to handle requests
any
> > more. Tomcat slows down quite noticeably too. I'm not sure I have
ever
> > left it up long enough to know if it might just die after enough
threads
> > "pile" up(well, it definitely would when it hit the OS' thread
limit). Are
> > you seeing far more threads than you would expect (well in excess of
the
> > pool size)? Maybe we are suffering from the same malady.
--
http://www.borland.com/newsgroups
http://www.borland.com/devsupport/disclaim.html
