0

My home server runs a couple of shell scripts regularly for maintenance tasks - mostly backup, but also other stuff. I would like to be alerted in case anything fails but also keep a log of when it works.

Currently my setup looks like this:

  • Cron calls one shell script which calls other scripts (just so the one won't get too complex). I decided to use one script with many tasks instead of individual cron items as I don't know how long each will take and I don't want them to interfere with one another.
  • My cron setup contains a MAILTO line. I never get any errors.
  • I don't have any logging. I just check from time to time whether the backup actually exists.

I know, I could implement into each script the functionality to log to a file (or syslog). Is there a way to define this from a central point so that I do not have to code this into every script individually?

Not sure how to achieve a better monitoring. I think a log analyzer system would be too much for this. Someone suggested running the scripts through Jenkins instead of shell/cron, but that seems to be even more effort.

What is a simple and good option?

0

I have implemented the following:

  • Enabled output to stdout for various steps or added custom output, e.g.:
    • echo "Starting backup..."
    • rsync whatever && echo "Backup successful" || echo "Backup failed"
  • Checking the return codes of each step of the script, either exiting the sub-script immediately or continuing, returning an error code at the end of the script
  • wrote a wrapper for my maintenance script which redirects all the outputs to a log file and if there are any errors within the maintenance script, I get a mail.

Example of maintenance script (does not exit if any individual step breaks, but returns an error in the end):

#!/bin/bash

RETURNCODE=0

echo "Execution started $(date)"

/root/do_something.sh || RETURNCODE=1

# (...)

exit $RETURNCODE

Example of wrapper script that calls the other script, this one is now in my crontab:

#!/bin/bash

# exit on any error (there should not be any in this script)
set -e

LOGFILE="/var/log/my.log"

# redirect STDOUT and STDERR to logfile...
if /root/maintenance.sh > $LOGFILE 2>&1; then
    # the colon ":" means: do nothing
    :
else
    # on error, send me an email
    mail -s "maintenance script failed" email@address.com < "$LOGFILE"
fi

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.