Releases/Firefox 4.0b4/BuildNotes: Difference between revisions

 
(46 intermediate revisions by 4 users not shown)
Line 3: Line 3:


==Signed-off Revision(s)==
==Signed-off Revision(s)==
* Build1: [https://hg.mozilla.org/mozilla-central/rev/b51b190b9fcc b51b190b9fcc]
* Build1: [http://hg.mozilla.org/mozilla-central/rev/b51b190b9fcc b51b190b9fcc]
* Build2: [https://hg.mozilla.org/mozilla-central/rev/9aa39b619a19 9aa39b619a19]
* Build2: [http://hg.mozilla.org/mozilla-central/rev/9aa39b619a19 9aa39b619a19]
* Build3: [http://hg.mozilla.org/mozilla-central/rev/9d6448b6a677 9d6448b6a677]


==L10N changesets==
==L10N changesets==
Line 47: Line 48:
| 20100805181633
| 20100805181633
| mv-moz2-linux-ix-slave10  
| mv-moz2-linux-ix-slave10  
|
| 52 mins
|-
|-
| [Linux 64 compressed]
| [Linux 64 compressed]
| 20100817135329
| 20100817135329
| moz2-linux64-slave06
| moz2-linux64-slave06
|
| 2 hrs, 7 mins
|-
|-
| [Mac OS X]
| [Mac OS X]
| 20100817134330
| 20100817134330
| bm-xserve11
| bm-xserve11
|
| 3 hrs, 28 mins
|-
|-
| [Mac OS X 64]
| [Mac OS X 64]
| 20100817134655
| 20100817134655
| moz2-darwin10-slave05
| moz2-darwin10-slave05
|
| 1 hrs, 45 mins
|-
|-
| [Win32]
| [Win32]
| 20100817145336
| 20100817145336
| mw32-ix-slave03
| mw32-ix-slave03
|
| 2 hrs, 44 mins
|-
|-
| rowspan="5" | 2
| rowspan="5" | 3
| [Linux compressed]
| [Linux compressed]
| -
| 20100818120848
| -
| mv-moz2-linux-ix-slave10
| -
| 59 mins
|-
|-
| [Linux 64 compressed]
| [Linux 64 compressed]
| -
| 20100818121643
| -
| moz2-linux64-slave06
| -
| 1 hrs, 51 mins
|-
|-
| [Mac OS X]
| [Mac OS X]
| -
| 20100818121614
| -
| bm-xserve11
| -
| 3 hrs, 31 mins
|-
|-
| [Mac OS X 64]
| [Mac OS X 64]
| -
| 20100818121922
| -
| moz2-darwin10-slave05
| -
| 1 hrs, 48 mins
|-
|-
| [Win32]
| [Win32]
| -
| 20100818132640
| -
| mw32-ix-slave03
| -
| 2 hrs, 48 mins
|-
|-
|}
|}


== Bugs that we have hit on this release  ==
* <strike>{{bug|588546}} - Unittests for all but win32 are not being triggered, packageTests step not run either</strike>
NOTE from armenzg: There might be more on the build notes but I have added the one that I have fixed for the next beta.
== Build Notes ==
===Build 1===
===Build 1===
* Marked all builders for all releases on pm for a clobber -- https://build.mozilla.org/clobberer/ has a Release Clobber option if you are '''not''' connected to Build VPN
* Marked all builders for all releases on pm for a clobber -- https://build.mozilla.org/clobberer/ has a Release Clobber option if you are '''not''' connected to Build VPN
Line 184: Line 191:


====Signing====
====Signing====
Using [https://intranet.mozilla.org/Build:CombinedSigning instructions] - I found that the last step of verifying hangs and I had to kill the process. Both times it got stuck trying to unpack a mar (different ones):
en-US.xpi removed from stage {{bug|485860}}:
Error unpacking mar file d:/signing-work/firefox-4.0b4/signed-build2/update/win32/cs/firefox-4.0b4.complete.mar to /cygdrive/c/DOCUME~1/cltsign/LOCALS~1/Temp/1/tmpX_8Vf8
 
#ffxbld@stage
cd /pub/mozilla.org/firefox/nightly/4.0b4-candidates/build3
rm en-US.xpi
 
Using [https://intranet.mozilla.org/Build:CombinedSigning instructions].


====L10N Verify====
====L10N Verify====
* metadiff for win32_l10n had warnings - on all locales - about chrome.jar being different
* metadiff for win32_l10n had warnings - on many locales
source/bin/chrome/${locale}.jar and target/bin/chrome/${locale}.jar differ


====Updates====
====Updates====
Line 198: Line 211:
* clobbered release builders
* clobbered release builders
* no l10n changes - checked with Axel in #release-drivers
* no l10n changes - checked with Axel in #release-drivers
* reconfig pm ('''Note''' updated buildbotcustom before this build)
* reconfig pm ('''Note''': updated buildbotcustom before this build, forgot to remove *.pyc will have to wait and see if that causes any problems)
* notified #developers (changed topic even!) and closed tree for tagging
* notified #developers (changed topic even!) and closed tree for tagging
** Why do we close the tree for tagging? Our automation doesn't cope well with losing a push race
** Why do we close the tree for tagging? Our automation doesn't cope well with losing a push race. {{bug|508896}}
* sendchange:
* sendchange:
  buildbot sendchange --username=lsblakk --master=localhost:9010 --branch=mozilla-central -m "Firefox 4.0b4 build3" doit
  buildbot sendchange --username=lsblakk --master=localhost:9010 --branch=mozilla-central -m "Firefox 4.0b4 build3" doit
Line 215: Line 228:


====Build/Repack====
====Build/Repack====
No problems
====Unittests====
====Unittests====
* Filed {{bug|588546}} because of a regression from {{bug|585098}} where now only the unittestPlatforms specified in the release_config get the make packageTest step run. This results in only sendchanges/tests for win32 -- '''note''' in build2 also got test package for macosx, but see if that still happens now that buildbotcustom was updated for this run
=====Results=====
=====Results=====
* win32 tests run on pm had a timeout on xpcshell consistent with [https://bugzilla.mozilla.org/show_bug.cgi?id=580525 bug 580525]


====Signing====
====Signing====
en-US.xpi removed from stage {{bug|485860}}:
#ffxbld@stage
cd /pub/mozilla.org/firefox/nightly/4.0b4-candidates/build3
rm en-US.xpi
Using [https://intranet.mozilla.org/Build:CombinedSigning instructions].
====L10N Verify====
====L10N Verify====
* no problems Only in... for all platforms was mostly the same except some ja/ja-JP differences (we will be symlinking the ftp later for that locale)
* win32 currently gives warnings on the jar diff step -- this appears to be new since {{bug|552668}} and was not present in previous beta releases
source/bin/chrome/${locale}.jar and target/bin/chrome/${locale}.jar differ
** {{bug|589196}} (new setup layout for win32) fixed it (see [[#Update_Verify]] for the details)
====Updates====
====Updates====
* no problems


====Update Verify====
====Update Verify====
* After some investigation {{bug|589196}} was filed (Windows setup layout was changed in beta 4). After landing the patch the builder became green.
* The output has the following expected diffs:
Only in source/bin/chrome: be.manifest
Only in source/bin/defaults/profile: prefs.js
** $locale.manifest became localized.manifest in b4 ({{bug|579178}}, {{bug|556644}})
** prefs.js was [http://hg.mozilla.org/mozilla-central/rev/b799524f7810#l1.102 removed] in b4
==== Move slaves back to active masters ====
Returned all but 1 slave (kept mv-moz2-linux-ix-slave02) to pm01/03:
*to pm03:
**mv-moz2-linux-ix-slave10
**mw32-ix-slave18
**mw32-ix-slave15
* to pm01:
**mv-moz2-linux-ix-slave03
**mv-moz2-linux-ix-slave04
**mw32-ix-slave02
**mw32-ix-slave03
**bm-xserve06
**bm-xserve11
**bm-xserve12
**bm-xserve16
**moz2-linux64-slave04
**moz2-linux64-slave06
**moz2-darwin10-slave05
**moz2-darwin10-slave06


====Final checks before push====
====Final checks before push====
<pre>
# ffxbld@stage
cd /pub/mozilla.org/firefox/nightly/4.0b4-candidates/build3/
find . ! -user ffxbld # contib/...
find . ! -group firefox
find . -type f ! -perm 644
find . -mindepth 1 -type d ! -perm 755 ! -name 'contrib*' # contib/...
find . -mindepth 1 -type d ! -perm 755 ! -name 'contrib*' # contib/...
^^ Last line is WRONG, should be |find . -maxdepth 1 -type d ! -perm 2775 -name 'contrib*'|
</pre>


====Virus check====
====Virus check====
<pre>
clamdscan -m /mnt/netapp/stage/archive.mozilla.org/pub/firefox/nightly/4.0b4-candidates/build3/
/mnt/netapp/stage/archive.mozilla.org/pub/firefox/nightly/4.0b4-candidates/build3: OK
----------- SCAN SUMMARY -----------
Infected files: 0
Time: 1056.363 sec (17 m 36 s)
</pre>


====Update Bouncer====
====Update Bouncer====
Forced bouncer_submitter builder. Checked using web interface.


====Push to mirrors====
====Push to mirrors====
<pre>
# ffbld@stage
rsync -av --exclude=*tests* --exclude=*crashreporter* --exclude=*.log \
  --exclude=*.txt --exclude=*unsigned* --exclude=*update-backup* \
  /pub/mozilla.org/firefox/nightly/4.0b4-candidates/build3/ \
  /pub/mozilla.org/firefox/releases/4.0b4
</pre>
Add index.html to cover the files before release time:
<pre>
# ffbld@stage
cd /pub/mozilla.org/firefox/releases/4.0b4
curl -sL https://bugzilla.mozilla.org/attachment.cgi?id=456169 | sed 's/Beta 1/Beta 4/g' > index.html
for dir in `find . -mindepth 1 -type d`; do cp -pv index.html $dir/; done
</pre>


====Final verify====
====Final verify====
* went green on first run - sent remaining slave (mv-moz2-linux-ix-slave02) back to pm01
* sent email when mirrors were absorbed for testing at 8:07am PDT
* sent email when mirrors were absorbed for release at 9:17am PDT  (more than 30K uptake)


====Push snippets to beta channel====
====Push snippets to beta channel====
*Upon GO from release-drivers
# cltbld@aus2-staging
cd /opt/aus2/snippets/staging
~/bin/backupsnip 20100817-Firefox-4.0b4
~/bin/pushsnip  20100817-Firefox-4.0b4
*After investigating why the beta updates did not apply cleanly (and throttling updates temporarily it turned out that I had pushed build2 snippets live
# cltbld@aus2-staging
cd /opt/aus2/snippets/staging
~/bin/pushsnip  20100818-Firefox-4.0b4
*Pinged QA to re-test
**QA bug: updates do not apply cleanly {{bug|590235}}
**ServerOps blocker bug to throttle {{bug|590227}}
**ServerOps blocker bug to unthrottle {{bug|590250}}


====Remove index.html files====
====Remove index.html files====
Once we are fully announced (by the driver) we can remove the index.html files:
# ffxbld@stage
cd /pub/mozilla.org/firefox/releases/4.0b4/
find . -name index.html
find . -name index.html | xargs rm -v


====Release====
====Release====
* Emailed the following to metrics@m.c
Firefox,firefox,firefox,4.0b4,4.0
Firefox,firefox,firefox,4.0b5pre,4.0
Confirmed users
4,467

edits