Forum archive 2000-2006

ymarkov - WebWork Webserver crash

ymarkov - WebWork Webserver crash

by Arnold Pizer -
Number of replies: 0
inactiveTopicWebWork Webserver crash topic started 9/17/2002; 2:12:52 AM
last post 9/17/2002; 11:39:54 PM
userymarkov - WebWork Webserver crash  blueArrow
9/17/2002; 2:12:52 AM (reads: 1346, responses: 4)
Hello. Here at Rutgers we run WebWork 1.7 on a dual (700MHz) processor Dell with 1 GB of RAM and plenty of disk space. About 1000 students in Business Calc I are served, spread through the week (roughly 200 a night 5 days a week). We run precreateSets for all sets. I the last three days we had two Webserver (latest Apachie I understand) crashes at times which are not characterized by heavy student use. A kill and restart of the Webserver solved the problem, but not for long. We run WebWork 1.6 on the same machine last year with the same student loads, and we had 3 crashes of the same type in the Fall 2001 and one in the Spring 2002. We thoght this to be an isolated issue but we are not so sure now. The sysadmin here is adamant that the Webserver is set-up well. Have you heard/had any similar problems. Any idea of possible fixes? I would be gratefull for any input, since people are rather tense, and a quick resolution of this issue will help.

Best regards, Yavor Markov

<| Post or View Comments |>

userJohn Jones - Re: WebWork Webserver crash  blueArrow
9/17/2002; 4:14:18 AM (reads: 1634, responses: 0)
The only system crashes I have experienced under linux have turned out to be hardware problems, so that would be one thing to try to check, although it can be hard to track down (aside from disk problems, which you can check for).

If it is crashing more often now, it could give you a greater chance to observe conditions prior to crash. Keep an eye on load (for a brief time over the summer we say WeBWorK processes hanging around using 100% cpu -- then they stopped), disk usage (even if the whole disk is large, having the partition containing /tmp run out of space will bother WeBWorK), and numbers of files in directories (could be large in tmp directories).

Just some thoughts.

John

<| Post or View Comments |>

userArnold K. Pizer - Re: WebWork Webserver crash  blueArrow
9/17/2002; 8:40:26 AM (reads: 1603, responses: 1)
Hi,

This may or may not be relevent to your situation at Rutgers. But if not it may be helpful to others. We use the following parameters in apache's configuration file (httpd.conf). The most important one is that we limit MaxClients to 25 (by default this is often set to 150). With our new server that has more memory we could probably safely up this limit, but 25 has worked very well for us. The main point is that you never want your server to have to swap memory and if your server ever swapps memory, then MaxClients is probably set too high. If your server starts swapping memory, things will probably degrade quickly and your system may get so bogged down that you think it has crashed.

Timeout 300
KeepAlive On
MaxKeepAliveRequests 100
KeepAliveTimeout 10
MinSpareServers 10
MaxSpareServers 25
StartServers 10
MaxClients 25
MaxRequestsPerChild 30

Arnie

<| Post or View Comments |>

userArnold K. Pizer - Re: WebWork Webserver crash  blueArrow
9/17/2002; 11:39:54 PM (reads: 1608, responses: 0)
Hi Yavor,

Another possibility is that this could be caused by a linux kernel bug. Here is a bug fix advisory from RedHat.

Bug Fix Advisory

Details:

A few bugs, including one in ext3 that could cause a kernel panic on SMP systems, are fixed in this kernel errata.

In the 2.4.18-3 kernel, a timing change exposed a previously-unseen race condition in ext3 which could occasionally cause the kernel to panic on SMP systems. The 2.4.18-4 kernel corrects this problem.

<| Post or View Comments |>

userymarkov - Re: WebWork Webserver crash  blueArrow
9/20/2002; 3:48:57 AM (reads: 1845, responses: 0)
Definitely, the settings are a place we need to look at more carefully.

Thank you.

<| Post or View Comments |>