Page MenuHomeElementl

Message missing repository errors in the scheduler differently in dagit
ClosedPublic

Authored by dgibson on Feb 19 2021, 3:54 PM.

Details

Summary

Two small things to better handle the case in the scheduler when you rename a repo while a schedule is running:

  • Include a recommendation of what to do (go turn off the now-unloadable schedule)
  • Don't recommend restarting the daemon just because it raised an error - we could show something like this when the daemon is *down*, but if it's running and just raised an error we shouldn't recommend a restart.

(When the issue was filed the daemon was *also* actually crashing due an unrelated issue with the liveness check, johann fixed that separately)

One thing I'm not sure about is if we should message this as a 'health check'-type error at all? Since the daemon is running just fine, it's just being given data that's no longer valid so it can't run a schedule. i.e. its more of a user code error than an ops/admin error.

Test Plan

BK, rename a repository witha running daemon and dagit, dialog now includes a thing you should go do

Diff Detail

Repository
R1 dagster
Lint
Lint Not Applicable
Unit
Tests Not Applicable

Event Timeline

python_modules/dagster/dagster/scheduler/scheduler.py
134

Should we call out how to reach the schedule page? Or is that too easy to get out of sync?

Right now it seems a little buried under Status > Schedules. cc @dish

python_modules/dagster/dagster/scheduler/scheduler.py
134

The new UI makes this better since there's just one Schedules page

This revision is now accepted and ready to land.Feb 19 2021, 4:34 PM