Description
| Status | Subtype | Assigned | Task | ||
|---|---|---|---|---|---|
| Resolved | greg | T89049 Make log responsibilities changes | |||
| Resolved | greg | T89292 Create project/tag to collect fatal/exception log related bugs |
Event Timeline
I told you I wasn't sure on naming! ;)
Honest proposals:
- #WMF-errors
- #WMF-log-errors
- #WMF-fix-yo-stuff
So the intent is that this would be something similar to the browser-test-bug tag? I like Wikimedia-production-error.
Honest question, why put in WMF? Whose errors would they be if that wasn't specified? We have (in new things) tried to stray from putting Wikimedia and WMF in things unless it adds some clarity. Honestly just not sure here.
Production-MW-Exception and then Staging-MW-Exception or something in that vein seems in line.
I did it under the assumption I would get pushback without it :)
Production-MW-Exception and then Staging-MW-Exception or something in that vein seems in line.
NB: It won't be *just* MediaWiki/extensions exceptions, potentially.
If you are thinking of grouping all exceptions from prod (even infrastructure?) under one big umbrella I would be concerned that it would come with a pretty solid bystander effect. The target audience being sufficiently broad as it make it impractical to watch the tag and feel like it's relevant to 'you'.
Maybe a 'production-exception' and 'staging-exception' tag orientation (which will have to be coupled with operations or parsoid or mw-core or something) would be a reasonable way to go. Right now we already have a standard for using (yellow tag-projects) as context overlays like "I wanted to see all operations tickets that are related to the network specifically" https://phabricator.wikimedia.org/maniphest/query/JdqZ.A36XBbA/#R
This would allow any exception raised in production to get a task and an automatic project. That some releng person would rotate doing a clinic duty type triage on while passing it to the right folks and still keeping tags.
IDK :)
Maybe it could be just MW/extension exceptions/fatals and leave the services to their own devices?
prod vs staging-exception could be useful, but probably overkill right now?
the staging suggestion was mostly illustrative yeah :)
In theory we'll never see any prod-exceptions once we are using the heck out of staging-exception...right?...right???
With a good project description you could probably get away with the more generic I suppose. Kinda not sure honestly.
I'm going to go with @Legoktm's suggestion of Wikimedia-production-error unless anyone vetoes me :)
To address the concern of it being too big/unwieldy: hopefully this project will start out quite busy but then calm way down after we get things under control.