Releases/Firefox 20.0b1/BuildNotes

• Who: jhopkins
• Bugs Encountered:
  • bug 842987 - release runner didn't mark 20.0b1 as failed
  • bug 842990 - better documentation around dealing with release runner when it fails
  • bug 842992 - l10n changesets error messages in release sanity suck
  • bug 843040 - xulrunner nsinstall command fails symlink create due to file exists
  • bug 791244 - Undo the pref for the temporary hangserver for beta channels until Socorro is awesome
    • relman missed this change for beta
  • bug 784848 - Do Windows l10n repacks on win64
    • win32 repack failed: configure: error: You are targeting i386 but using the 64-bit compiler.
• Tracking: bug 837118

• Ship the l10n milestone ^{DESKTOP MOBILE RELEASE BETA}
• Submit to Ship It ^{DESKTOP MOBILE RELEASE BETA ESR}
• Push to mirrors ^{DESKTOP RELEASE}
• Publish Fennec ^{MOBILE RELEASE BETA}
• Push Fennec to mirrors ^{MOBILE RELEASE BETA}
• Apply Throttling ^{DESKTOP RELEASE}
• Verify throttling ^{DESKTOP RELEASE}
• Run pushsnip ^{DESKTOP RELEASE BETA ESR}
• Update bouncer links ^{DESKTOP BETA RELEASE}
• Update XULRunner wiki pages ^{DESKTOP RELEASE}
• Post-release tasks ^{DESKTOP RELEASE BETA ESR}
• Get the Browser Choice website updated ^{DESKTOP RELEASE}

Build 1

Submit to Ship-It

[6:32am EST] ship-it config submitted & review requested from callek. r+'d and kicked off a couple of minutes later.
[6:35am EST] 'Release runner encountered a runtime error'. Investigating.
[6:48am EST] Resubmitted to ship-it. Reconfig problem on bm13 seems to have been transient (ran it by hand successfully).
[7:30am EST] bhearsum pointed me at the real log file: /var/log/supervisor/release-runner.log. ERROR : cannot find l10n locale zh-CN in repo https://hg.mozilla.org/releases/l10n/mozilla-beta/zh-CN/file/1e65623aaf2e
[7:39am EST] release runner jobs 'stuck' in a grayed out state. No docs on this. bhearsum unstuck the jobs, then unstuck a reconfig.
[7:57am EST] release runner running ok now

Build Results

[09:23am EST] xulrunner mac failed, filed bug 843040. [10:xxam EST] builds failing due to improper partialUpdates setting fed to shipit UI.

Build 2

Submit to Ship-It

[12:06pm EST] Email to r-d: Patches for bug 791244 and bug 843040 are landed on mozilla-beta. Can you please send a go-to-build for Firefox/Fennec 20 beta 1 build #2?
[12:20pm EST] Received go-to-build for 20.0b1 build #2
[01:00pm EST] Shipit desktop release config reviewed and submitted. Holding off on mobile, per lsblakk.
[01:19pm EST] Reconfig unstuck (ec2 build slave), Tagging started.

Build Results

[03:46pm EST] linux/linux64/mac/win32 Firefox desktop builds are all ready.
[06:30pm EST] all desktop repacks on win32 are failing with: "configure: error: You are targeting i386 but using the 64-bit compiler."
[07:35pm EST] aki determines that win32 repack failure are due to bug 784848
[07:42pm EST] aki is backing out the patches in bug 784848 and rolling it out to the buildbot masters.
[08:14pm EST] aki announces reconfigs are done, jhopkins rebuilds failed win32 repacks.
[09:49pm EST] win32 repacks still running. buildapi/running does not show them (need to file a bug, investigate further). Notified lsblakk in #planning that releng will pick up the Fennec build request tomorrow morning, assuming it is ready by then.
[03:05am EST Feb. 21] update_verify_1 failed on MacOS X due to bugXXXX.
[09:59am EST] rail informs jhopkins that bm30 has been spamming us with HTTP/400s re: bouncer entries. Investigation reveals that build#1's wrong partialUpdates entries resulted in incorrect bouncer entries being created after tagging.
[10:39am EST] old bouncer entries removed, bouncer_submitter rebuild submitted and completed.

Awaiting a new "go to build" for Fennec build 2.
[12:15pm EST] lsblakk does not have an updated ETA for Fennec 20.0b1 build 2. She is waiting for mfinkle and blassey.
[03:58pm EST] mfinkle says we are waiting on bug 843361. landed in beta, waiting for a green cycle.