Automating some basic admin stuff

[bluemike]

I am afraid I am not skilled enough to write a Linux script from scratch, so I need to know if what I have done will cause problems for me.

Basically I want to ensure that, in the event of a key service stoppage, they don't stay stopped for days (like over the weekend when no one is here).

For example, here's what I've done for the SMTPD service. I an using the 'task scheduler' in Fedora to create a job that will run every 15 minutes. It is set to run '/opt/scalix/bin/omon smtpd'. As far as I know, this will have no ill effects if the SMTPD is already up.

During the regular workday, the server would be monitored via the SAC like normal. But I've been burned a few times when the SMTPD or Service Router stops on Friday night, and isn't started again until Monday morning.

The ultimate goal is, of course, to determine why these services are stopping and fix that problem. But, in the meantime, will my solution work?

[bluemike]

So can I assume that the forums silence equals their collective approval?

[STXRich]

You may have to do a couple of tweaks, but the sample maintanence script in the administrative directory of the install Tar-ball is pretty extensive.

There is a frequent section that checks for aborted services, daemons, & ports. Which if found will attempt to restart them, and notify you via email.

It also has some sample script for the nightly backups, which needs some tweaking, but sets you in the right direction.

I would check that out.

The script is ommaint.

[pete]

I have found that the sample maintenance script does not restart smtpd when it fails.
FWIW this has only happened a couple of times, so I have not really looked into it that much...

[bluemike]

The task scheduler option just seemed too easy. I keep waiting for some support person to tell me that it won't work over the long term.

Right now, it's only attempting a restart of smtpd, but if no one has any reason why I shouldn't. i think I may set it up to try a couple of differnt things (i.e. resubbing any error queue messages every few hours)

[pete]

I can't see anything wrong with doing that - in fact I may try it myself!

P