Service Monitoring

[dave-lew99]

Hello,

One of our linux based services occasionally falls over and in these situations i'd like it to auto restart (like windows services do) but as far as i can tell there isn't anything 'built in' to CentOS and the method is via a custom script.

Any suggestions?

Thanks in advance.

 

[Volitous]

I'd pop a shell script in cron to run say every 5 mins (or 1 min).

Obviously your milage may vary (your service may be a whole bunch of processes/jvm's etc) - but if its just a process you could do something like the below.

e.g

#!/usr/bin/ksh
MYVAR=`ps -ef | grep "your process string - be specific"`

if [[ $MYVAR = 0 ]]; then
/opt/path/to/your/command -args
fi

Pop this into cron to run on an interval:
E.g 5 Mins

5 * * * * /your/script.sh > /dev/null 2>&1

You could improve something like this by echoing the results of your script with a timestamp into a logfile for you to review.

 

[dave-lew99]

Having done a bit of digging, it looks like monit may be the answer

http://mmonit.com/monit/

Seems to do exactly what i want without having to write scripts for each service, plus detects hung services rather than ones which have died.