Restart apache on crash
 

I have a copy of apache2 running under a user on RHE to serve up images for a website. The site has been experiancing heavy load lately and the apache2 occationally crashes, but if no one is around to notice it, it might stay out for a while or over night depending on when it crashes.

Is there a way to monitor apache and have it restart if it crashes? Sort of what cpanel does for its version of apache?

Thank for anything you guys can suggest.

You could add a script to cron.daemon to run every (15?) minutes (the time interval is up to you)
which runs 'ps aux | grep httpd' - if it can't find httpd then it 'apachectl start' :)

Good luck

Steve

do u have any ideas why it crashes? u may take a look at the server logs to find out. apache does not tend to crash without reason as far as i know.

Yes it crashed because it gets maxed out on its connections for a extended period of time. We are running at around 12 million page loads a day and 1200 Gigs of bandwidth a month on that server and the apache2 is hosting the images.

The only question I have is can i get the grep to check the user name as well? Cause there is two copies of apache a 1.3 that comes with cpanel and a 2 to host layouts and images since the cpanel couldn't handle it all. So I just want to check the one running as a specific user not nobody.

Thanks for the help.

sure,

ps aux | grep httpd | grep -v nobody | grep -v grep

grep -v is an inverse grep,

the first grep -v ignores httpd running by nobody
the second grep -v ignores grep

or you could grep for the specific user

ps aux | grep -httpd | grep <user>

Steve

So just to make sure I did this right I put

if ps aux | grep -httpd | grep <user>
then
everything ok exit script
else
restart the server
fi

So that should be all i need right?

Essentially yes, in pseudo:

Maybe you can post your final script for others to use :)

Steve

Here is my final script, I used a different version of the ps command just cause it required one less grep command so I figured less commands is better and I did the if backwards cause I could never get the = "" to work for some odd reason, I tried like 20 ways but hey this works. The only thing is when you run it manually you get a long list from the ps but hey this runs in a cron so I don't care I never see it. I also got it to send me an email when it happens so that I can sort of check up on the server, if its crashing to much its not that good. The restartMsg.txt just contains a sort little restart message.


#!/bin/bash
#script checks and restarts apache if required

if ps U layout | grep httpd
then
echo "Layouts server up and running"
exit 0
else
#restart server
echo "Layouts server down, restarting server"
cat restartMsg.txt | mail -s "Layouts failed on host.domain.com" root
/home/layout/bin/apachectl restart
echo "Layouts server restarted"
fi


Hope someone else might find the script useful, I know its small and easy to write but you never know.
Thanks Steve for the help :cool:

No problem, just glad you got it all sorted :)

Shame apache crashes though, still, heavy loads cause these sort of things to happen :(

Steve

Edit:

I know it's not a problem since you're running it through cron, but for anyone else, try this instead:

if ps U layout | grep httpd > /dev/null

It'll suppress the output from ps :)