schedule Remove drift from job scheduling

This PR is to remove drift from job scheduling, by basing the next run time off the last scheduled run time, instead of datetime.datetime.now().

One perhaps nice possibility this PR opens up is, since the next scheduled time would be based on the previous scheduled time, one could store and restore the .last_run between invocations of a schedule script to let it pick up from where it left off.

Feb 01 '19 01:02 brasie

Coverage remained the same at 100.0% when pulling ab44f6def686177efb94faff97bfa890078630ac on brasie:nodrift into 5aaed06d5867cfe783eeea0651e2e7a57241e5d5 on dbader:master.

Feb 01 '19 01:02 coveralls

Would very much like this merged.

Dec 07 '19 06:12 lalinsky

I rebased this against current master in case it is still of interest... I think it would help with the sort of issue brought up in places like #144, #163, #174, #207, #307, #355

Dec 14 '20 23:12 brasie

Is there any update on this?

Apr 12 '21 14:04 marcelhohn

Do we have plan to merge this? This makes more sense especially if the task takes long to run (and the running time may vary)

Aug 21 '21 03:08 tianzhuqiao