Since Plesk 6, apache restarts fail with corrupted httpd.include. For example:

I assume this is happening when clients add/delete domains. Anyone experience and fix this?

what you have on line 4807 ?

its apear they miss a ">"
in it..

Is apache is restarting when you retry it ?



You mean they just goes down then you have to restart it manualy ?

Apache fails with such errors and a restart fixes it. I think this maybe happening when apache gets restarted in the middle of my_apci_rst.

Try this then let me know.. do a backup of httpd before changing it ;


editi "stop" function in /etc/init.d/httpd script, add line like:


ps -ax | grep httpd | grep -v restart | grep -v stop | awk '{print "kill -9 " $1}' | sh > /dev/null 2>&1


so, it should be like:

....
stop() {
echo -n $"Stopping $prog: "
killproc $httpd
RETVAL=$?
ps -ax | grep httpd | grep -v restart | grep -v stop | awk '{print "kill -9 " $1}' | sh > /dev/null 2>&1
echo
[ $RETVAL = 0 ] && rm -f /var/lock/subsys/httpd /var/run/httpd.pid
}

Sorry for not getting back to you sooner. I don't think is addressing the issue. For some reason my_apci_rst is not completing the build of httpd.include. We have theories but can't be sure what's causing it. If my_apci_rst is triggered and starts to rebuild httpd.include and someone deletes a domain or maybe just updates hosting at that point, maybe my_apci_rst chokes when it gets to a section that isn't there or is in the process of being wiped.

But this is happening everyday. At least 3 occurences between 10 servers. The frequency kind of knocks that theory in the teeth as the timing has to be so precise. Hard to know how to fix when we can't even find the root cause.

i dont have other thing to help...

I have seen problem of httpd going down because not properly shuttdown , then each time for ex: a domain is deleted apache will goes down ...

This have fixed it for me.. But your look like not similar thing may be..

What you do when this append just reastart apache and it work ?
If so have you try my fix ?

Well, we don't know why, but apache is getting restarted when my_apci_rst's writing of httpd.include is incomplete.

We've modified the restart script, and after a restart it will check after
10 seconds if apache has really started and will start it again if it hasn't
and then will check again in 30 seconds. We've done this on the 2 worst afficted servers and give it 2 days or so. Then we'll add it to the apache rpm and add to the apt-get wheel.