After deploying the latest release (which contained the hotfixes released recently) we noticed that certain, but not all, schedule jobs had lost their settings as well as the history of previous runs. Since the hotfix release contained several fixes for schedule jobs we were quite quick to suspect that something had been overlooked. This lead to the following situation.
Select isn’t broken
After some digging we quite sheepishly realized that we during the sprint had changed namespace of some of our scheduled jobs. Since this is used as a key of sorts for the job it quite naturally “loses” it’s settings and history because it’s considered a new job. This was a great reminder of the first rule of programming: It’s always your fault.