Firefox 58 release post mortem
Attendees: Selena, DVeditz, TomGrab, Liz, Marcia, RyanVM, Julien, Ritu, Arynx, Philipp.
- [marcia] Service worker crashes were high volume during 58 and the fixes landed very late in the cycle. At its peak we were hitting over 1000 crashes per day
- https://bugzilla.mozilla.org/show_bug.cgi?id=1416879 was filed on 11-13 well before the break and All Hands
- (Ritu) Github repo syncs remain a challenge
- Discussions underway on how to mitigate this with Myk Melez, Julien etc.
- activity stream, screenshots, anything with external libraries, Test Pilot,
- (Ritu) 58 has worsening of OOM crashes (even noticeable without retain display lists)
- [selena] can we get the rates to compare?
- (Ritu) Add Selena to the email thread
- we may be switching to one content process for low memory devices/platforms to mitigate this (selena is following up)
- fennec slow startup (safe browsing related?)
- apparently an issue in 57, then people woke up towards the end of 58 rc
- we ended up fixing it in 58.0.1
- How is the triage situation?
- is anyone watching (who?) the fennec-tracking+ flag? (Yes: chris karlof, mike comella, snorp, susheel's team are triaging fennec issues regularly)
- Selena mentioned that some Fennec issues are landing into Core::General.
- (Ritu) Chemspill process for 58.0.1
- we are working to better document the chemspill process, from wennie, ritu, and dan
- (liz) we went pretty fast - less than 24 hours between landing on m-r and releasing.
- Daily stand-ups really helped (thanks Wennie)! ++ really helpful!
- Secure slack channel helped as well, and the planning document + checklist.
- QA coverage on the weekend is still TBD
- Thanks to Gerry for smooth, unexpected, last minute hand off of the release.