The usability of alerts.wikimedia.org has long been a noted pain point for many SREs. The interface can be confusing, and there are many avenues for improvement.
This ticket is an overarching parent task for usability improvements we can make to the alerts interface.
Some basics that don't warrant subtasks:
- Label ordering. Karma doesn't guarantee ordering, but we can use labels.order to ensure that we consistently order them (severity, team, instance being good first params)
- Make links action buttons - makes runbooks and dashboards immediately accessible (already done)
- Phab link support (This already exists in WMCS!)
- Is there anything to be gained from changing our ack defaults? Longer silences? 15 minutes is not really useful for anyone, an hour is probably better.
- Simplify the silence form - we currently have a lot of high-cardinality matchers in our silence form, and we can use silenceForm to drop some of these if they're not useful
- Maybe a bit more contentious, but use labels.color.unique to colour by team? WMCS also already does this
- Trim labels - there's a lot of clutter in each card, can we trim some info from labels without impairing our ability to silence usefully by label?
- Are there improvements we could make to alert grouping? There are a bunch of config options for it.