Hi, We have a dedicated web server running on Redhat linux 8.0. The installed applications are Apache 2.0.47,Tomcat 4.1.27,Mysql-4.0.13,with mod_jk2 connectors. After a period of time (approximately 20 or 25 days), THE ENTIRE SERVER CRASHES. Tomcat and apache fail to serve and we are getting 500 internal server errors. We are also unable to login to our server through SSH or Telnet at that time. Just incase if we are able to login we do not get # prompt. After hard reboot of the server every thing works fine. We are not getting any clues regarding what went wrong in any of the error logs. We had checked the server 2 hours before it crashed, the memory utilisation was low, all services were running fine and there was no load on the server also during that period. This is the second time the server has crashed like this in a time gap of 45 days. Why the server is crashing like this periodically? How to log messages so that we will know what went wrong when the server crashes? How to configure the server to send alert messages on critical conditions like low memory, running service failures, etc? Regards., Raja

Have you at least checked to see your /tmp is not filling up??? That would be my first guess but you need to look at your /var/log/messages to see what kind of error messages you get before it crashes... taurus

Bah, I thought you people running servers atleast had the knowledge to check your log files! There are 10 kinds of people, those who count in binary and those who don't.

To diagnose a problem, you need all your logs and non-default configuration settings. There are several types of logs you should have running: 1) Firewall 2) Server (Apache) 3) System (syslog) Your firewall logs are usually easy to check. Make sure nothing strange or excessive happens (<2 hours) before the server dies. Your server logs are the most help here. Apache has a module configured for logging (error and access), which is probably going somewhere (find error_log, maybe in /var/log/apache). You can also pipe these logs to other processes and files. Don't forget you can add conditionals to your logs (to automatically edit out verbose information) and rotate your logs to keep them smaller and easier to manage. Some errors are also sent to syslogd. System-wide issues will be reported here. A simple way to get an alert email is with an outside daemon monitoring the service. You could write a script to attempt to connect to Apache every N minutes, fail three times, and then email (this can be configured many ways, and even run on another host). Check online repositories for some scripts. You can also check for various monitoring programs (freshmeat.net, sourceforge.net, etc). There are plenty. It is a good idea to move your entire logging to another, small server. 500MHz< with a new hard drive would work fine.

Hi ., thanks for your information. we check with the file /var/log/messages ,and apache and tomcat error log. we could not find any errors on that file also. Raja

Errors do not always produce error messages. Check for patches to the software you are using (All the software you listed have new builds available - MySQL, Apache, and Tomcat) and look at the cronjobs to make sure nothing is running regularly that you don't know about. Look at the hardware. Check the integrity of your RAM (via memtest86), temperature of the machine, and power supply.