tomcat-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Christopher Schultz <ch...@christopherschultz.net>
Subject Re: FW: Tomcat 6.0.20 Causes Kernel Crash on Linux
Date Wed, 18 Nov 2009 19:26:56 GMT
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Timir,

On 11/18/2009 7:09 AM, Timir Hazarika (thazarik) wrote:
> Hm. I thought all required information was in the last email.

Hardly:

Does it panic every time? Does it panic in the same place every time?
Does it panic only under load? While idle? When you request
/the_whole_internet.tar.gz from the web server?
Does it fail when your webapp is never deployed?
What connector(s) are you using?
Are you using APR?
Do you have any JNI code running?

> [I simply replaced] 6.0.18 [with] 6.0.20, [started] tomcat and ka-boom[.]

You could have a look at this:
http://tomcat.apache.org/tomcat-6.0-doc/changelog.html

> - Java 1.5.upd10

Sun's JVM, presumably? I've got 1.5.0_19 on my Linux box. You might want
to upgrade to the latest supported patch level on your version.

> Oh, and the maxheap, minheap, permsize  are all set to 512M.

That's a pretty big permsize. How much memory does this machine have?
I'm guessing you don't have 4GiB on an i586.

> http://markmail.org/message/mrpgvn4mqvyrq64a reports a memory leak, 
> though I don't have enough debugs to say it matches mine.

Given that the stack trace doesn't seem to include any GC code or
anything having to do with memory, why would you suspect that you have
the same problem as the one being discussed in this thread? They are
talking about sessions apparently never being cleaned-up, not kernel panics.

Your original stack trace does not seem complete. Either that, or there
are lots of problems everywhere (all those invalid parameter values...).

> - Linux kernel, custom built 2.6.23, i586

What does "custom built" mean? Does that mean that you have hacked that
kernel yourself, or does that mean that you merely compiled it yourself
from the canonical kernel sources?

No user code ever written ought to be able to crash a kernel. That's
called a security problem: the best case is that you're open to a DOS
attack which is pretty dramatic: your kernel halts and you need to power
cycle the box :(

Is your architecture really i586? Yuk.

Note that the Linux kernel has moved-on to bigger and better version
numbers: the latest stable version listed on kernel.org shows as
2.6.27.39 for the 2.6 series. Is it possible that there is a bug in your
kernel that has been fixed in the intervening versions?

Unfortunately for you, there's nothing that anyone on this list can
really help you with (unless there's a kernel hacker out there who feels
like donating their time to help, here) because there is no way Tomcat
can affect the operation of the JVM or the OS kernel in this way,
therefore, no Tomcat configuration can prevent it from happening other
than to avoid triggering this particular problem.

I agree with Peter that if you really have changed nothing aside
upgrading from Tomcat 6.0.18 to 6.0.20, then you probably have either a
kernel bug (and you win a beer) or you have bad hardware, and you win a
trip to the hardware store. Just because hardware has been running for
years doesn't mean it's guaranteed to keep running forever.

But, if Tomcat is running okay at 6.0.18 and not 6.0.20, then I guess
the temporary solution is clear: stick with 6.0.18, though I have a hard
time believing that Tomcat is the source of the problem.

- -chris
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (MingW32)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAksESoAACgkQ9CaO5/Lv0PDhjgCePwBeWeVgAt0stXkwEftHVH4q
yHcAn0R0S16BAbgYSTdaAsJcNKDLY36n
=Jk4v
-----END PGP SIGNATURE-----

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org


Mime
View raw message