Frequently when debugging cron jobs, one finds that:

• One needs to use locks to keep more than one copy of frequently running jobs from running at the same time.

• One wishes to load the user environment -- /etc/profile as well as RC files in HOME.

• Generally, one would cd to HOME for application specific users.

• The path should include /usr/local/bin.

• The output of the cron job should be logged to Syslog, which both prevents cron from raising errors about mailers and allows you to see what went wrong if something did.

ohmycron does all this and more, managing locks with a POSIX API, so that the OS takes care of cleaning up locks for failed proceseses.

Prepending ohmycron to the commands in your crontab keeps your crontab clean and transparently adds logging, locking and environment loading.

* * * * *  root  ohmycron sleep 10
* * * * *  root  ohmycron --tag update:ohmycron -- curl -sSfL 'https://raw.githubusercontent.com/instacart/ohmycron/master/ohmycron' -o /usr/local/bin/ohmycron

Setting SHELL=/usr/local/bin/ohmycron (or another path if you have installed ohmycron elsewhere) transparently adds locks, logging and environment setup to all the jobs in a cron file. (Tasks are actually run with Bash.) It is does something the wrapper can't do, too: support multi-statement commands which use the shell operators &&, |, || and so forth.

SHELL=/usr/local/bin/ohmycron
* * * * *  root  sleep 10
* * * * *  root  : update:ohmycron ; curl -sSfL 'https://raw.githubusercontent.com/instacart/ohmycron/master/ohmycron' -o /usr/local/bin/ohmycron

You can explicitly name a cron job with a Bash "no-op comment": : <some words> ;. (In Bash, : is a no-op; the arguments to the no-op command are ignored.)

sudo curl -sSfL 'https://raw.githubusercontent.com/instacart/ohmycron/master/ohmycron' -o /usr/local/bin/ohmycron
sudo chmod a+rx /usr/local/bin/ohmycron