On Fri, Mar 9, 2012 at 9:16 PM, Christopher Schultz <
ch...@christopherschultz.net> wrote:

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Doron,
>
> On 3/9/12 12:19 PM, Doron Tsur wrote:
> > Tomcat: 7  latest Java: 1.6.25 OS: Windows Server 2003 R2 SP2
>


> qballer:  sorry. Java 1.6.0_25 tomcat 7.0.23
>


> 1. What specific version of Tomcat 7?
> 2. 1.6.25 is not a version of Java that I know of. Real version number
>   and vendor, please?
>
> > This is a production server that uses high bandwidth. Poking
> > around Poolmon.exe I was able to find that
> >
> > 1. the AfdP tag is consuming a lot of NP (Non Paged) pool.
>
> What is "the AdfP tag"?
>


> qballer: ancillary function driver  it's a Win Socket driver
>


> > 2. This is a tag concerning the windows socket driver (afd.sys).
>
> Googling for "afd non-paged pool exhaustion" comes up with lots of
> hits. Have you read them?
>


> qballer:  Yes I have.
>


> Many seem to suggest that certain connections are left open for a very
> long time. Are you using long-lasting connections with KEEPALIVE
> enabled? If so, it's possible that it could be causing this problem.
>
> > We have many production servers running our tomcat. We have only
> > encountered this issue in two of them, those servers have the same
> > OS/Java/Tomcat installed. One server uses high bandwidth and the
> > other uses much less.
>
> Obviously, the question is: what is different about these two servers?
> If this was a problem with Tomcat itself, the JVM, or even the seevice
> configuration, I would expect it to happen everywhere. Given that it
> doesn't happen everywhere, I would suspect something at a lower-level.
> Do all servers have identical OS images on them, and are they all at
> the same patch-level?
>
> Not the same images, that is why i suspect it's outside of tomcat.

> > I say tomcat because this is the process that consumes all NP pool
> > (about 170MB).
> >
> > 1. Have you ever encountered this issue? 2. Do you know of any
> > application level/Tomcat configuration work around to solve this
> > issue?
>
> None that I know of. Perhaps you could adjust some of the fine-grained
> socket-configuration bits, but I wouldn't start fiddling without a
> very good idea about what you are doing. It's even possible that
> Tomcat does not support configuration of whatever specific socket bits
> you need to fiddle. If that turns out to be the problem, it's fairly
> easy to add configuration options.
>
> > 3. Do you think that there is something wrong in the way Tomcat
> > interfaces with winsock?
>
> Tomcat does not interface at all with winsock unless you are using
> APR. Are you using APR? That would seem like a good thing to tell us.
>
> qballer: Nope not using APR.

> > 4. Do you believe that Tomcat APR might resolve this issue?
>
> If anything, I would say that APR might even make the problem worse.
> On the other hand, it's fairly easy to try it: just drop the .dlls in
> the right place and restart. As long as you haven't changed Tomcat
> from the default configuration, it will use APR for your HTTP connectors.
>
> > 5. I can't seem to download this hotfix to further investigate
> > these files, they might contain a suitable version. Do you know how
> > can I get them?
>
> What hotfix are you talking about?
>
> Sorry ignore the hotfix part, I've checked that direction and it's a dead
end.


> > When Tomcat is killed the Non Paged Pool doesn't return to the OS.
>
> How do you "kill" Tomcat? Does the JVM process shut down entirely? If
> it does, then your problem should lie elsewhere: once the process is
> dead, all memory should be released. If that's not happening, there is
> either a horrible bug in the OS or you are misinterpreting your
> observations.
>
> Yes it does shut down, NPP doesn't get returned to the OS.
That is why i think there is a driver code in the way.


> - -chris
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG/MacGPG2 v2.0.17 (Darwin)
> Comment: GPGTools - http://gpgtools.org
> Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/
>
> iEYEARECAAYFAk9aVxcACgkQ9CaO5/Lv0PCFhwCgmgy9A1mrFg4Me1lHp674i5pf
> jjMAnipMogXefj0ezYhJNs1WSJb13V3F
> =XtAE
> -----END PGP SIGNATURE-----
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org
> For additional commands, e-mail: users-h...@tomcat.apache.org
>
>

Reply via email to