Defunct workers can go undetected
In Users are not receiving upcoming token expirati... (#432518 - closed), PersonalAccessTokens::ExpiringWorker
has not been running for a long-time. Surprisingly, its constant failures went unnoticed. During this time, our error budget was around 100%. This worker runs once a day, so it's failure has close to 0 effect on the error budget.
Proposal: figure out ways to get insights into workers. At minimum, a team should get notified when a worker constantly fails on every run.