The sequence that has happened historically with management systems and scheduling goes something along these lines:
- “Scheduling tools are too complicated - can you just build in a ‘repeating task’ that runs at a specific time, until I decide to cancel it?”
- “I don’t want to specify it in Zulu time, I want to specify it in my time zone”
- “What do you mean, the server is running in Madagascar?”
- “Oh - it needs to only run on Fridays”
- “Actually, it needs to not-run if the previous run failed”
- “It needs to run/not-run when the following things are currently-true about the system”
- “Why did DST mess up my runs?”
- “Don’t run on holidays”
- “Why did Easter mess up my runs?”
- “Why can’t thiis just implement cron’s interface?”
Eventually, every time a “simple” scheduling function is built into a management tool, it turns into cron. Eventually. After a lot of RFEs, and bugs.
“cron” and friends have decades of “how to schedule” experience, so it’s better to take advantage of them, than take advantage of a low-budget version built into the tool