Thanks for the ideas. I cranked my debugging up to 99.
There are a couple of things that I see, but don't know if they are
serious:
1. SEVERE: The scratchDir you specified:
/development/jakarta-tomcat-5.0.28/work/Catalina/localhost/admin is
unusable.
-- I never set this, I am assuming it is the default
2. WARNING: Duplicate name in Manifest: Class-Path
-- I think this is due to some JAR in struts
3. In the apache log I get warnings about compiling mod_jk with EAPI on
I am mapping through mod_jk, so the error I get is the Apache Internal
Server Error. The Tomcat process is dead, so there is no tomcat error
page.
I am using struts, I will am pretty sure I am returning everything to
the pool. I have a top level servlet filter that logs all exceptions,
including Throwable all to no avail.
It seems to crash at random, but mostly b/c I am not always aware of
when it crashes. Today, after rebooting, it didn't even start up, but
then started fine when I called startup.sh
-Grant
On May 22, 2005, at 9:51 PM, Steve Kirk wrote:
What is your actual logging config?
Hazy memory, but don't you want debug=99 rather than debug=1 to get
more
detail?
If you really can't get logging to work, you could insert
System.out.println("blah") statements at key points around where you
think
the crash might be caused, in lieu of your log statements. Not pretty
but
it can get you results.
Some Qs to narrow it down a bit:
Is the crash triggered by a single type of request, or maybe a burst of
traffic, or will it crash even with no requests?
Is the crashing request direct to TC or via apache httpd?
Is a TC html error page generated, what does it say?
Do you start any of your own threads?
Does your code always return pooled objects to the pool?
Is there some app that people use to make sure the server stays up by
checking it every so often and restarting if needed?
I've never come across that. TC is pretty reliable, shouldn't need it
IMHE.
-----Original Message-----
From: Grant Ingersoll [mailto:[EMAIL PROTECTED]
Sent: Monday 23 May 2005 02:39
To: [email protected]
Subject: Mysterious failures
Hi,
Have been a long time user of TC, but first time poster.
I am running 5.0.28 on OS X with PostgreSQL and Struts and connecting
to Apache using mod_jk. I have a webapp deployed that is pretty
mature. However, I am having some mysterious crashes of Tomcat that I
haven't been able to get my head around. The whole process is dying
w/o so much as a peep. Not one single log/exception is being written
anywhere (as far as I know) that gives even the most remote clue as to
why and I am at a loss for how to get at the problem.
Here is what I have done to date:
1. Put a catch (Throwable) with a log message at my top level part of
the servlet that logs the exception and then throws it out.
2. I have turned on DEBUG level logging for every piece of
the application
3. I have set debug="1" everywhere I could in server.xml
Anyone have any suggestions on what else to do for debugging this?
Part of me feels that it might be JDBC related as I am not using the
Tomcat JNDI lookup methods for getting connections (but am managing a
pool myself). Should I be using the JNDI lookup methods for getting
connections? My only other guess was that it seems to happen after it
has been running for a little while and I thought it might have
something to do with the session timeout stuff, but I can't see why
that would cause the process to exit.
Is there some app that people use to make sure the server stays up by
checking it every so often and restarting if needed?
Thanks for any advice,
Grant
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
