Firefox/Channels/Postmortem/59
Notes for 59 post mortem
10:30am PDT Tuesday April 3 Vidyo channel: Release Coordination IRC: release-drivers
Meta post mortem: let's announce this meeting on the calendar invite, and more times on more channels
Contents
Some stats on 59 (lizzard)
- Around 3900 bugs fixed in 59. (Caution, this will take a while to load: https://mzl.la/2FxAmAc)
- Around 240 uplifts to beta
- Uplifts during beta:https://mzl.la/2GqQgda
- RC week uplifts: https://mzl.la/2FAabcn
- The overall number of uplifts is low, and feels like a successful cultural change
- Comparison: Firefox 56: 352 Firefox 46: 430
- The low uplift # also may be from having a longer than usual time in Nightly (10 weeks, because of the holidays)
- More collected info on 59: https://wiki.mozilla.org/Releases/Firefox_59
What's New Page
Can we have a confirmed list of WNP locales before the go to build of RC? (bhearsum)
Needs coordination between relman, releng, Eric Renaud, l10n team (flod?) and QA.
Action : Liz will follow up with Eric . Regular meeting in beta cycle near the end of the cycle but not RC week, add to milestones list
Sec Issues for ESR/release out of sync
(Al's feedback on ESR) (fixes that landed for ESR52.6.0 that we had fixed in 58 or earlier but not in the corresponding ESR)
- If we do this, which I believe is rare, we need to let the sec team know, so they can write advisories.
- Dan: we shouldn't even do this, don't land it on release if it isn't going on ESR.
- sec triage issue and for relman - check the flags. do we know when esr is affected?
Uplifts late in the beta/RC week cycle
In general, the later these issues land, the more important it is to coordinate closely with QA.
- Autoconfig/search hijacking issues. https://bugzilla.mozilla.org/show_bug.cgi?id=1435862
- Other late search changes (Yahoo)
- Search suggestion issue Friday before the release.
- https://bugzilla.mozilla.org/show_bug.cgi?id=1444548
Is anything new and secret happening for 60? Really, really , honestly no :)
ESR52 italian locale issue
- https://bugzilla.mozilla.org/show_bug.cgi?id=1445278
- Follow-up bugs filed to try to make the errors get caught during CI and/or during l10n packaging
We aren't quite testing what we ship because of l10n repacks. Long term questions here for our process. Tom, mkaply, ryan are discussing it
we may be waiting for l20n n
Actions: This is mostly resolved, from having more tests, except bug 1445628
Dot release drivers
59.0.1 Desktop + Fennec
Security issue - pwn2own related chemspill. https://www.mozilla.org/en-US/security/advisories/mfsa2018-08/
- Fairly smooth process, we were well prepared
- Suggestions for anything we can do better here?
- We should have respun the ESR52 build to avoid the tagging and source tarball issue, even though it was going to be a no-op for the build we ship
59.0.2 Desktop
Prepped by Liz, handed off to RyanVM and Ritu. I think the handoff went well.
Details: https://docs.google.com/spreadsheets/d/12HGUaMrt7-BAZblheJUTDaLngMhQPR8vJcMq0G3eRXc/edit#gid=0
- a11y crash. Bug 1424505 - Crash in I_RpcNegotiateTransferSyntax volume is high, we knew this, we just didn't have a solid fix in time for 59.
- Rendering issue. https://bugzilla.mozilla.org/show_bug.cgi?id=1435472 reported widely on release. We knew about it in Feb. but didn't set the priority high. It was unclear that it was widespread at that point.
- https://bugzilla.mozilla.org/show_bug.cgi?id=1420553 already got filed by a user a few days after the changes landed on nightly with a full regression range
- liz will follow up with devs/triage owners
59.0.2 Fennec
- Leanplum crash. https://bugzilla.mozilla.org/show_bug.cgi?id=1444549 Startup crash. Found in the RC week.
- We need more follow up here: Do we have enough resources to support fennec dev and troubleshooting. We have more contractors now.
Ongoing issue with mystery android crashes! Marcia suggests follow up Possible meeting?
