[07:35:24] my 2 cents on the topic are: [07:36:35] - most netbox reports should alert DCOps only, either via their IRC channel or via email [07:36:53] - the failure should link directly the report, not the generic Wikitech page [07:37:28] - if possible the failure should directly report what failed (if by email), safer not via IRC to avoid to leak private info [07:40:04] - they should probably be split between the ones for which failures are generated by operational errors that anyone can fix and the bulk of coherence ones that only DCOps can fix [07:40:14] EOF [08:04:29] should open tasks not irc or email [08:10:49] those are my 2 cent, if there is a shared agreement I'll open a task [13:02:17] +1 volans [15:15:41] is there existing tooling to open tasks from service failures [15:17:35] chaomodus: for RAID alerts there is something [15:32:38] that should be made generic [15:32:56] but then with alermanager comming up, is it worth doing it in Icinga? [15:33:03] I was thinking the same thing for LibreNMS alerts [15:34:00] spicerack has already a phabricator integration, we should add the possibility of creating tasks that is not yet there [15:34:08] one issue that there is with the old implementation in the raid handler [15:34:37] is that it doesn't have the capability to check fo duplicates (based on some conditions) and update the existing one instead of creating a new one [15:36:49] getting it right in all cases probably takes keeping local state, and/or the ability to read from as well as write to phab [15:37:12] yeah and that can be added too, but to start we can just add the write a new task [15:37:23] we've already comment on an existing one [15:54:54] how would an automated system call spicerack [15:57:31] define automated system [22:47:38] in this case arzhel advocates opening tasks as a result of reports going red