How do I fix a cron job that stopped due to clock drift?

[Andrew.J]

I just spent two hours trying to figure out why my app's scheduled task wasn't firing, and it turns out the system clock had drifted by a few seconds, throwing the entire cron job out of sync. Has anyone else run into this kind of subtle timing issue where a seemingly minor system clock drift caused a major failure in your scheduled processes?

[EleanorZB]

That exact thing happened to me too. A few seconds of clock drift a couple days in a row pushed the nightly job just out of its window and nothing ran. We started logging the time against a reference and could see the drift in the timestamps.

[ElizabethM]

I spent days chasing code paths only to realize the problem was the worker getting blocked on GC for a few seconds; the cron part wasn't actually failing, just the backlog made it look like it wasn't firing.

[JohnVR]

We added a drift check that compares system time to a trusted NTP source every hour and sends an alert if it goes beyond a threshold. It helped us notice drift sooner, though it didn't fix the core timing once the drift happened.

[KevinMP]

I tried loosening the tolerance and letting a supervisor trigger if the queue was empty, but that just hid the symptom.

[BrianXG]

Do you have a drift monitor in place, or do you rely on NTP alone?

[MiaCT]

From a reliability angle, cron is brittle here; you end up chasing a non deterministic clock rather than the code. It feels like a reminder that we should use monotonic clocks for measurements and separate the time source for scheduling.