UNIX for Dummies Questions & Answers

hi guys

what is the common way to restart a webserver that got hung upon high load?

i have a loop shell script running, checking for a small textfile on the webserver, in case it doesnt get that file within 5 seonds it kills the server and restarts it.

i know its very crude, but i needed a solution and have not know better.


how does a professional achieve this goal?

I think that's ok. Maybe place that script onto some more remote clients (maybe in different networks/subnets) so you can also check the connectivity over the network and if more than <n> clients wait more than 5 secs for the answer, you can restart it.
Not sure what this script looks like, maybe use wget to get some web content from your webserver. If your script works, don't worry, I think

i have this in /opt/running.sh

Code:

running=no
running=`curl --connect-timeout 3 http://localhost/running.txt`
if [ $running == "no" ]
then
killall -9 lighttpd
/etc/init.d/lighttpd start
fi
sleep 60
/opt/running.sh

and in /var/www/running.txt there simply the word

Code:

running

i invoke the script at boot time with sudo -b /opt/running.sh

but what i long wondered

why the heck do webservers that are distributed for production use like APACHE not bring such a function with them selves??

Hm, when assigning the $running the second time, you will always get something different than "no", no matter if curl was successful or not, I think. So it will always fail and restart. Maybe you can check this by the process' time of your httpd.

Better drop the 1st line and have something like checking the exit status of curl:

Code:

curl --connect-timeout 3 http://localhost/running.txt > /dev/null 2>&1
if [ $? != 0 ]; then
...

or drop the 1st line and do

Code:

running=`curl --connect-timeout 3 http://localhost/running.txt`
if [ $running != "running" ]; then
...