防止计划的 cron 作业在失败时重复

Question

I have a cron job that runs every minute that involves logging into a database. Our organization mandates we change our passwords every 6 months - and the password change process can be pretty laggy - sometimes taking up to 10 minutes. However, during the latency period between when the password is changed via our tool (and goes into effect) and the change can be made in hard coded instances on our Linux box, this job will incur several failed login attempts to the database. Once it gets past a certain number, the account is locked - so all our subsequent jobs fail until we can get in touch with the DBA to unlock the account.

All the questionable practices demonstrated here notwithstanding (hard coded passwords, etc.) - I'm looking for a means for the cron job to essentially withdraw itself from the schedule if/when the login fails. Yes, doing so manually prior to the password change is possible - but not everyone going through this process is terribly linux/cron/vi literate - hence my search for some form of silver bullet that might avoid this if possible.

Any suggestions much appreciated.

Answer 1

To prevent a cron job from failing more than once, you need a semaphore file. This solution is similar to the comment by tvm . Create the semaphore file outside of the cron job like so:

touch is_ok.txt

It needs to be recreated after every failure to enable the cron job to run. Then wrap your command like so:

if [[ -e is_ok.txt ]] && command ; then touch is_ok.txt ; else rm -f is_ok.txt ; fi

Example (using grep foo bar as your cron job command ):

# Enable `grep foo bar` to run:

touch is_ok.txt

# Create the input for `grep foo bar` to run successfully:

echo foo > bar

# `grep foo bar` runs successfully repeatedly:

if [[ -e is_ok.txt ]] && grep foo bar ; then touch is_ok.txt ; else rm -f is_ok.txt ; fi

# Make `grep foo bar` fail:

rm bar

# `grep foo bar` runs once, then never runs until you do `touch is_ok.txt`:

if [[ -e is_ok.txt ]] && grep foo bar ; then touch is_ok.txt ; else rm -f is_ok.txt ; fi

Answer 2

Depending on your box, OS, shell, etc ... this might and might not work.

For bash for instance, you might want to change the command run by cron to something like this :

*/1 * * * * Command || exit

Which basically means : Every minute, run this command OR exit. so, if the command fails, it will exit. but that will still be logged as a failed attempt to login.