perl-modperl mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Adam Prime x443" <apr...@brunico.com>
Subject RE: Apache2.2 - openssl 0.9.8d - modperl 2.03 unpredictable timeouts during requests
Date Tue, 06 Mar 2007 16:24:57 GMT
This may be totally unrelated, but I had a problem that I ultimately
solved by setting up a PerlChildInit handler to open DB connections for
each child explicitly.  For some reason it seemed like only one
connection was being opened to connect to the MSSQL DB and that single
connection was somehow being shared between all the clients.  The
symptoms sound similar, it resulted in random timeouts on pages that
used the DB.  in server-status it just showed that connections were
hanging open indefinatetly in the write stage.  

Not sure if this will help you at all or not but I thought i'd mention
it.

Adam

-----Original Message-----
From: Jeremy Wall [mailto:jeremy@marzhillstudios.com] 
Sent: Friday, March 02, 2007 4:05 PM
To: modperl@perl.apache.org
Subject: Apache2.2 - openssl 0.9.8d - modperl 2.03 unpredictable
timeouts during requests

Hello,
	This is my first posting to the mod_perl mail list. I'm looking
forward to 
conversing with the folks here.

	I'm having a problem that has me stumped. I'm running Apache 2.2
with 
mod_perl 2.0.3 on Windows 2000 server. Specifially the latest XAMPP 
distribution because it was the easiest way to get working OpenSSL on 
Windows. It's a dell server with quad 2 gig processors 2 gigs of memory
and a 
raid 5 scsi drive array. The application in question is a custom perl 
application using a custom templating system. I'm also running
PostgreSQL 8.1 
for the database on the same machine. I'm not having any performance
issues 
with the database that I can tell so I don't think it has anything to do
with 
that.

	I'm using a mod_perl response handler for the pages in question.
Periodically 
and with no pattern I can detect a page request will simply timeout. No
error 
appears in the error logs and no information that gives me a clue can be

found even after turning on debug level logging. I've tried google
searches, 
#Apache and #perl on Freenode. No one so far has been able to give me
any 
insight. I'm looking for any help someone can give me that might point
to 
either a solution or where I can look to find one. The only clues I can
find 
that might be related are these:

 1. There are some unclean shutdowns reported in the debug level logging
 2. I do have a few FIN_WAIT_2 connections states lingering when using
netstat
 3. It seems subjectively (but not confirmed) to happen less often when 
connecting without SSL
 4. It does not have any visible correlation to cpu useage, Memory
useage, or 
hard drive useage.

I don't even know if these are helpful but I'm including the information

anyway. The same requests that time out one time will succeed the very
next 
time you make it. You can make the request several times in a row and
only 
fail once or you might fail multiple times. It happens from both Mozilla
and 
IE and Konqueror on Windows or Linux OS'es for the clients. If needed I
can 
provide logs, configuration files, and even *some* sourcecode but I
didn't 
want to flood the list with attachments unless it was going to be
helpful. I 
hope someone out there can give me some insight into this.

Jeremy Wall
http://jeremy.marzhillstudios.com/
jeremy@marzhillstudios.com

Mime
View raw message