Releases/Firefox 3.5rc1/BuildNotes

Build Engineers

nthomas/bhearsum
release tracking bug

Signed-off Revision(s)

Build1:

Tags

On releases/mozilla-1.9.1:

Build #	Tag	Changeset
1

Build data

Type	Build ID	Build machine
[Windows installer/zip]
[Mac compressed]
[Linux compressed]

Notes

• clean up directories on slaves from 3.5b99 (everything but in particular the repack dirs)
• patch buildbot-configs/mozilla2/release_config.py, and get review. Add the driver specified changeset and land the change
• checkin latest copy of l10n-changesets from the l10n tracking bug
• reconfig production-master
• do sendchange

buildbot sendchange --username=bhearsum --master=localhost:9010 --branch=releases/mozilla-1.9.1 -m "Firefox 3.5rc1 build1" goforit

Build 1

Tag

Hung on the first 'hg out' because the slave it was running on didn't have the host key for hg.mozilla.org accepted. Went around and accepted the host key on all slaves (not just Linux), cleaned out the tag directory on moz2-linux-slave19, and restarted. Tagging worked fine afterwards.

Source

No problems

Build/Repack

No problems with Linux build or l10n. No problems with Mac build or l10n. No problems with Win32 build or l10n.

TODO prioritize release builders above others or use different slave set for l10n. mac l10n took awhile to start because the nightly builds got triggered first.

Removed .../build1/en-US.xpi, which is not handled properly until bug 485860 is fixed.'

Sign

instructions

• Failed out first time due to disk space. Cleared some space and restarted it
• No problems on second run
• Everything after this manually triggered because of bug 479041

L10nVerify

No problems. Lots of path differences in comments in firefox-l10n.js on linux and mac, and gl updating some bookmarks strings.

Generate updates and push betatest snippets

moz2-linux-slave19 hadn't had stage-old.m.o's signature accepted, so partial uploads could not be uploaded. Fixed there and on slave18 to make whole pool OK, fixed linux refplatform doc.

The failure is actualy helpful, as we need to deal with the same issue as bug 496917 - that we can't use appv=3.5 RC 1, and in fact have to use just appv=3.5 in order for bug 485624 to work as planned.

On moz2-linux-slave19, applied this patch and regenerated snippets with the following:

wget -Oconfigfix.diff https://bugzilla.mozilla.org/attachment.cgi?id=383234
cd /builds/slave/updates/build/temp/firefox/3.5b99-3.5rc1
mkdir borked
mv aus2* borked
cd /builds/slave/updates/build/patcher-configs
patch -p1 < ~/configfix.diff
cd ..
perl patcher2.pl --create-patchinfo --partial-patchlist-file=patchlist.cfg --app=firefox --config=patcher-configs/moz191-branch-patcher2.cfg

With stage-old.m.o's sig accepted, we now manually upload the partial MARs and snippets:

rsync -av -e 'ssh -oIdentityFile=~/.ssh/ffxbld_dsa' --exclude=*complete.mar update ffxbld@stage-old.mozilla.org:/pub/mozilla.org/firefox/nightly/3.5rc1-candidates/build1/
rsync -av aus2.test/ cltbld@aus2-staging.mozilla.org:/opt/aus2/snippets/staging/20090615-Firefox-3.5rc1-test
rsync -av aus2/ cltbld@aus2-staging.mozilla.org:/opt/aus2/snippets/staging/20090615-Firefox-3.5rc1

Now, we push the test channel snippets:

# cltbld@aus2-staging
cd /opt/aus2/snippets/staging
~/bin/backupsnip 200906015-Firefox-3.5rc1-test
~/bin/pushsnip 200906015-Firefox-3.5rc1-test

Update verify

• In the first half of the logs for each platform we had differences channel-prefs.js. This is expected because this is the first release we've shipped on the 'release' channel.
• All platforms threw 'download_mars' errors on 3.5b4 mn, tr because they were shipped in 3.5b4 but not in 3.5rc1. These lines are ignorable.
• Win32 threw the usual messages about .chk files.
• No real problems here.

Update Bouncer

Done

RESPIN WERE DECLARED HERE bug 498379

Build 1

Sign

instructions

Update Bouncer

Done.