We found a bug with the schema migration scheduler where a job with two tasks was in status Enqueued where the tasks were in status DryRunSuccess and RunSuccess. The scheduler would get stuck because it was expecting tasks in status Enqueue and found none. The scheduler should be more resilient to this, and update the job status if a tasks is found in an unexpected state.
We found a bug with the schema migration scheduler where a job with two tasks was in status
Enqueuedwhere the tasks were in statusDryRunSuccessandRunSuccess. The scheduler would get stuck because it was expecting tasks in statusEnqueueand found none. The scheduler should be more resilient to this, and update the job status if a tasks is found in an unexpected state.This is likely due to a crash.