You can create another scheduled job (eg every minute), which would be writing to some constant or register (new) just the last execution. Through a global handler expectations (AttachIdleHandler) (eg every 2-5 minutes) to check the difference between the current time and the last entry in the constant (or register) if the difference is greater than it should be, then the schedule is not working.
The second one is too complicated to use by the end user and is not available on Linux. The first one is also quite complicated but probably doable. I might try it if there is no better solution.
cd /opt/1C/i386/
ras cluster localhost:1540 &
rac cluster list
rac infobase summary list --cluster=uuid
rac infobase info --infobase=uuid --cluster=uuid
Thank you for your suggestions, but they all where too complicated in our case, so we opted just to show a warning so that a user would not forget to check it manually.