There's a link to the runbook but it's time-consuming to find the relevant parts. The current format (e.g. T419140) is just a big list of largely useless / hard to interpret data fields. Something like "Run this command to check the logs, then that command to delete the alert" would be useful (the first half mostly exists already, but it's not obvious one needs to delete the alert if everything is OK and the job will recover on its own, and the exact commands need to be copied from the runbook and adapted, which is a bit cumbersome).
Description
Description
Related Objects
Related Objects
Event Timeline
Comment Actions
On an aside, it would be nice if the @phaultfinder user's profile description contained instructions on how to file tasks about it.
Comment Actions
(Tagging for your information as I believe mw-cron jobs are in the domain of ServiceOps; feel free to untag if incorrect :))
Comment Actions
(ServiceOps bug triage here.)
Thanks @Tgr. We have some work planned on improving periodic job alerts (T416576) and may take this on either in parallel or afterward. In particular, when the alerts fire less frequently and each is more actionable, we can make the documentation more targeted too. (I'm marking it Low accordingly, not because we think this isn't important.)
The actual text of the phab task (list of data fields) is mostly not configurable: anything we write will appear in the "description" line. If we wanted to change the whole task body, we'd be forking the upstream phalerts.py and the feasibility of that would be a question for SRE Observability. In the meantime we can either improve that description, the linked playbook, or both.