629256 - Disable release updates for locales with broken strings

Status: RESOLVED FIXED

(Release Engineering :: General, defect)

Description

[Nick Thomas [:nthomas] (UTC+12)]

Originally filed as bug 628829 for a single mac 4.0 locale, the scope has grown to three branches and 4 locales but still mac only. There's two parts on our side
1) removing any snippets that current exist
2) setting up exceptions in the patcher configs to prevent generating new updates

The list is
*) For 4.0
  * bn-IN, for b8,b9 -> b10
  * kn, ml, te for all 4.0
*) For 3.6: ml, te for all versions on branch  (comment #38)
*) For 3.5: kn, ml, te for all version on branch (comment #44)
All for mac only.

Currently we have bn-IN for 4.0b8/b9 chmod to 0 (bug 628829 comment #16) but that causes issues when we try to run the nightly backup.

Comment 1

[Nick Thomas [:nthomas] (UTC+12)]

Disabling live snippets for end-users (test channels left alone):

movedir() {
 echo $1
 base=~/bug629256-backup/`dirname $1`
 mkdir -p $base
 mv $1 $base/
}

cd /opt/aus2/incoming/3/Firefox

# bn-IN, for b8,b9 -> b10
chmod -v 775 4.0b{8,9}/Darwin*/*/bn-IN/beta/
for d in 4.0b{8,9}/Darwin*/*/bn-IN/beta/; do
  movedir $d
done

# kn, ml, te for all 4.0
for d in 4.0b*/Darwin*/*/{kn,ml,te}/beta/; do
  movedir $d
done

# For 3.6: ml, te for all versions on branch
for d in 3.6*/Darwin*/*/{ml,te}/{beta,release}/; do
  movedir $d
done

# For 3.5: kn, ml, te for all version on branch
for d in 3.5*/Darwin*/*/{kn,ml,te}/{beta,release}/; do
  movedir $d
done

Done in MPT and PHX.

Comment 2

[Nick Thomas [:nthomas] (UTC+12)]

Need to also fix the release snippets for the upcoming releases (MPT only):

cd /opt/aus2/snippets/staging

rsync -a 20110121-Firefox-3.5.17/ 20110126-Firefox-3.5.17-bug629256fix/
touch 20110126-Firefox-3.5.17-bug629256fix/
rm -rf 20110126-Firefox-3.5.17-bug629256fix/Firefox/3.5*/Darwin*/*/{kn,ml,te}

rsync -a 20110124-Firefox-3.6.14/ 20110126-Firefox-3.6.14-bug629256fix/
touch 20110126-Firefox-3.6.14-bug629256fix/
rm -rf 20110126-Firefox-3.6.14-bug629256fix/Firefox/3.6*/Darwin*/*/{ml,te}

rsync -a 20110124-Firefox-3.5.17-3.6.14-MU/ 20110126-Firefox-3.5.17-3.6.14-MU-bug629256fix/
touch 20110126-Firefox-3.5.17-3.6.14-MU-bug629256fix/
rm -rf 20110126-Firefox-3.5.17-3.6.14-MU-bug629256fix/Firefox/3.5*/Darwin*/*/{kn,ml,te}

Comment 3

[Nick Thomas [:nthomas] (UTC+12)]

(In reply to comment #0)
> The list is
> *) For 4.0
>   * bn-IN, for b8,b9 -> b10
>   * kn, ml, te for all 4.0

Turns out kn, ml, and te only shipped from b8 onwards so it's the same condition.

Comment 4

[Nick Thomas [:nthomas] (UTC+12)]

Attachment: [partly checked in] Patcher configs

This will prevent us generating updates for the builds with the problematic updaters. When the next set of branch releases come along the patcher config bumper won't add the exceptions, and we'll get updates to the set following that.

Comment 5

[Nick Thomas [:nthomas] (UTC+12)]

Attachment: [partly checked in] Update verify configs

Similar logic here.

Comment 6

[Rail Aliiev [:rail]]

Comment on attachment 507383 [details] [diff] [review]
[partly checked in] Patcher configs

linux-x86_64 is not listed in exceptions in moz20-branch-patcher2.cfg
r+ with if you add it.

Comment 7

[Robert Strong (they/them - no direct email)]

Nick, after Firefox 3.6.15 and 3.5.18 are released the updates should be re-enabled. The update will fail and they will end up with an unusable build / will have to reinstall but there are very few users for these locales and even fewer on Mac.

Comment 8

[Nick Thomas [:nthomas] (UTC+12)]

Alright. If we're don't respin 3.5.17 and 3.6.14 then I won't land the two attachments here, and we'll generate updates to 3.5.18 (die already!) and 3.6.15 as usual and get what you ask for.

Comment 9

[Nick Thomas [:nthomas] (UTC+12)]

Comment on attachment 507385 [details] [diff] [review]
[partly checked in] Update verify configs

The 1.9.2 branch part of this has landed for 3.6.14 build2.

http://hg.mozilla.org/build/tools/rev/fc5c4f02e03b

Comment 10

[Nick Thomas [:nthomas] (UTC+12)]

Comment on attachment 507383 [details] [diff] [review]
[partly checked in] Patcher configs

The 1.9.2 branch part of this landed for 3.6.14 build2.

Checking in moz192-branch-patcher2.cfg;
new revision: 1.43; previous revision: 1.42

Comment 11

[Armen [:armenzg]]

(In reply to comment #0)
> *) For 3.6: ml, te for all versions on branch  (comment #38)

Just to confirm, update verify only failed for Mac for 3.6.14 build2.
https://wiki.mozilla.org/Releases/Firefox_3.6.14/BuildNotes#Update_Verify_2

Comment 12

[Armen [:armenzg]]

Nick do we have to do anything else for 3.6.14 build2?
After your landing it doesn't seem so but I left your comment from build1 to make sure:
https://wiki.mozilla.org/Releases/Firefox_3.6.14/BuildNotes#Push_release_channel_snippets_2

Comment 13

[Nick Thomas [:nthomas] (UTC+12)]

(In reply to comment #11)
Looks like the update verify bumper added the locales back in :-( It's expected that no updates are found for those.

(In reply to comment #12)
I've removed the comment as it doesn't apply to build2, and we really don't want to push build1 snippets. Thanks for checking.

Comment 14

[Nick Thomas [:nthomas] (UTC+12)]

(In reply to comment #7)
> Nick, after Firefox 3.6.15 and 3.5.18 are released the updates should be
> re-enabled. The update will fail and they will end up with an unusable build /
> will have to reinstall but there are very few users for these locales and even
> fewer on Mac.

Rob, why 3.6.15 and 3.5.18 ? Don't we have the fixed updater in 3.6.14 and 3.5.17 ?

Comment 15

[Robert Strong (they/them - no direct email)]

Bug 628829 was approved / landed for 3.6.15 and 3.5.18... might have been added to the 3.6.14 / 3.5.17 release without my knowledge though so I'll check when I have some time.

Comment 16

[Nick Thomas [:nthomas] (UTC+12)]

Sorry, I should have checked that bug before asking.

Comment 17

[Nick Thomas [:nthomas] (UTC+12)]

Attachment: Patcher config for 3.6.15

This will prevent busted 3.6.14 -> 3.6.15 updates for the two sick locales. Everything else 3.6.x -> 3.6.15 is already disabled for them.

Comment 18

[Nick Thomas [:nthomas] (UTC+12)]

Comment on attachment 516682 [details] [diff] [review]
Patcher config for 3.6.15

Checking in moz192-branch-patcher2.cfg;
/cvsroot/mozilla/tools/patcher-configs/moz192-branch-patcher2.cfg,v  <--  moz192-branch-patcher2.cfg
new revision: 1.46; previous revision: 1.45
done

Comment 19

[Robert Strong (they/them - no direct email)]

Are the releases still blocked? I ask because of bug 646058

Comment 20

[[e-mail] (:bhearsum)]

I don't see any release or beta channel snippets for kn, ml, or te 3.5.18 builds:
https://aus2.mozilla.org/update/1/Firefox/3.5.18/20110319140258/Darwin_x86-gcc3-u-ppc-i386/kn/beta/update.xml?force=1
https://aus2.mozilla.org/update/1/Firefox/3.5.18/20110319140258/Darwin_x86-gcc3-u-ppc-i386/kn/release/update.xml?force=1
https://aus2.mozilla.org/update/1/Firefox/3.5.18/20110319140258/Darwin_x86-gcc3-u-ppc-i386/ml/release/update.xml?force=1
https://aus2.mozilla.org/update/1/Firefox/3.5.18/20110319140258/Darwin_x86-gcc3-u-ppc-i386/ml/beta/update.xml?force=1
https://aus2.mozilla.org/update/1/Firefox/3.5.18/20110319140258/Darwin_x86-gcc3-u-ppc-i386/te/beta/update.xml?force=1
https://aus2.mozilla.org/update/1/Firefox/3.5.18/20110319140258/Darwin_x86-gcc3-u-ppc-i386/te/release/update.xml?force=1


Based on that, I don't think bug 646058 is a dupe.

Comment 21

[Robert Strong (they/them - no direct email)]

Well, the user in bug 646058 is running Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10.5; kn; rv:1.9.1.18) Gecko/20110319 Firefox/3.5.18 and tried to update to Firefox 4.0. Since bug 628829 hasn't made it into any release vehicles any updates offered will cause a crash when Firefox tries to apply it.

Comment 22

[[e-mail] (:bhearsum)]

(In reply to comment #21)
> Well, the user in bug 646058 is running Mozilla/5.0 (Macintosh; U; Intel Mac OS
> X 10.5; kn; rv:1.9.1.18) Gecko/20110319 Firefox/3.5.18 and tried to update to
> Firefox 4.0. Since bug 628829 hasn't made it into any release vehicles any
> updates offered will cause a crash when Firefox tries to apply it.

Right, but what I'm saying is that I don't see *how* they could've gotten that update offer, since there are no snippets available for that locale/version. Perhaps this was something they tried a few days ago, or they have some sort of frankenfox? If we could get the full AUS URL they're using, that would be helpful.

Comment 23

[Robert Strong (they/them - no direct email)]

For clarity, bug 646058 is marked as a dupe of bug 628829 and not this bug. I don't see how they could have been offered an update either.

Why would trying a few days ago cause this to happen? Were updates available to these locales on 3.6.x / 3.5.x? If so, that could have been the cause but I thought updates had been disabled. I very much doubt that it is a frankenfox since they are extremely rare on Windows (under .3 % on Windows and strong anecdotal evidence points to it being even rarer on Mac) and iirc much of the same info used to generate the url is used to generate the UA string.

I'll gice the user instructions to get the url.

Comment 24

[[e-mail] (:bhearsum)]

(In reply to comment #23)
> For clarity, bug 646058 is marked as a dupe of bug 628829 and not this bug. I
> don't see how they could have been offered an update either.
> 
> Why would trying a few days ago cause this to happen? Were updates available to
> these locales on 3.6.x / 3.5.x? If so, that could have been the cause but I
> thought updates had been disabled. I very much doubt that it is a frankenfox
> since they are extremely rare on Windows (under .3 % on Windows and strong
> anecdotal evidence points to it being even rarer on Mac) and iirc much of the
> same info used to generate the url is used to generate the UA string.

It's possible that updates for these locales were available briefly when they first pushed, I don't recall now. In bug 646058, Anthony talks about reproing on releasetest, so I strongly suspect the user is on that or betatest.

Comment 25

[Chris Cooper [:coop] (he/him)]

I'm confused about what exactly [partly checked in] means in reference to the two attachments above. Which parts remain to be checked-in?

I ask because whimboo is reporting update failures on the test channels for 3.5.19 and 3.6.17. Do the above fixes only apply to the beta and release channels? How can we dig ourselves out of this hole on the test channels as well?

Comment 26

[Robert Strong (they/them - no direct email)]

Sounds like bug 646058 again. The patcher configs didn't prevent releasetest from getting the update previously and I doubt that has changed. btw: the fix for the locales hasn't as of yet made it into a release vehicle.

Comment 27

[Armen [:armenzg]]

Attachment: snippets currently live on aus

Let me say out loud what I think it is going on.
rstrong could you verify what I am saying?

> *) For 3.6: ml, te for all versions on branch  (comment #38)
> *) For 3.5: kn, ml, te for all version on branch (comment #44)

There is a bug for the aforementioned locales and they are stranded (bug 628829). A fix got landed that finally made it to 1.9.1 [1] and 1.9.2 [2].
This means that anyone downloading 3.6.17 and 3.5.19 will be able to update from now on.
I assume that no users on those 5 locales could update forward and crashed.

A) - With these assumptions, we should remove every live snippet for those locales in 3.5 and 3.6 (/opt/aus2/incoming/3) to the exception of the available MUs to 4.0 from the latest 3.5/3.6 (since the updater is now fixed).

B) - We should remove any traces of those locales on /opt/aus2/

C) - We should land patches on patcher to add the exceptions for sanity's sake.

To accomplish A we should do this:
> cd /opt/aus2/incoming/3
> rm -rf Firefox/3.6.{15,16}/Darwin*/*/{ml,te} # we don't touch the latest 3.6.17
> rm -rf Firefox/3.5.{1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17}/Darwin*/*/{kn,ml,te} # we don't touch the latest 3.5.19
> # see attachment to see that there isn't anything matching for 3.5.18

To accomplish B we should do this:
> cd /opt/aus2/snippets/staging
> rm -rf Firefox-3.6.17-build2-bug629256fix/Firefox/3.6*/Darwin*/*/{ml,te}
> rm -rf Firefox-3.5.19-build1-bug629256fix/Firefox/3.5*/Darwin*/*/{kn,ml,te}
> rm -rf Firefox-3.6.17-build2-4.0.1-build1-MU-bug629256fix/Firefox/3.6*/Darwin*/*/{ml,te}
> rm -rf Firefox-3.5.19-build1-4.0.1-build1-MU-bug629256fix/Firefox/3.5*/Darwin*/*/{kn,ml,te}

If we want to be more conservative you can look at note [3]. The rsyncs take forever for the record.

I will attach a patch for C later.

[1] http://hg.mozilla.org/releases/mozilla-1.9.2/file/FIREFOX_3_6_17_RELEASE/toolkit/mozapps/update/src/updater/readstrings.h#l43
[2] http://hg.mozilla.org/releases/mozilla-1.9.2/file/FIREFOX_3_6_17_RELEASE/toolkit/mozapps/update/src/updater/readstrings.h#l43

[3]
##### BEGIN
cd /opt/aus2/snippets/staging

rsync -a Firefox-3.6.17-build2/      Firefox-3.6.17-build2-bug629256fix/
rsync -a Firefox-3.6.17-build2-test/ Firefox-3.6.17-build2-test-bug629256fix/
rsync -a Firefox-3.6.17-build2-beta/ Firefox-3.6.17-build2-beta-bug629256fix/

rsync -a Firefox-3.6.17-build2-4.0.1-build1-MU/      Firefox-3.6.17-build2-4.0.1-build1-MU-bug629256fix/
rsync -a Firefox-3.6.17-build2-4.0.1-build1-MU-beta/ Firefox-3.6.17-build2-4.0.1-build1-MU-beta-bug629256fix/
rsync -a Firefox-3.6.17-build2-4.0.1-build1-MU-test/ Firefox-3.6.17-build2-4.0.1-build1-MU-test-bug629256fix/

rsync -a Firefox-3.5.19-build1-4.0.1-build1-MU/      Firefox-3.5.19-build1-4.0.1-build1-MU-bug629256fix/
rsync -a Firefox-3.5.19-build1-4.0.1-build1-MU-beta/ Firefox-3.5.19-build1-4.0.1-build1-MU-beta-bug629256fix/
rsync -a Firefox-3.5.19-build1-4.0.1-build1-MU-test/ Firefox-3.5.19-build1-4.0.1-build1-MU-test-bug629256fix/

rsync -a Firefox-3.5.19-build1/      Firefox-3.5.19-build1-bug629256fix/
rsync -a Firefox-3.5.19-build1-beta/ Firefox-3.5.19-build1-beta-bug629256fix/
rsync -a Firefox-3.5.19-build1-test/ Firefox-3.5.19-build1-test-bug629256fix/

touch Firefox-3.6.17-build2-bug629256fix/
touch Firefox-3.5.19-build1-bug629256fix/
touch Firefox-3.6.17-build2-4.0.1-build1-MU-bug629256fix/
touch Firefox-3.5.19-build1-4.0.1-build1-MU-bug629256fix/

touch Firefox-3.6.17-build2-beta-bug629256fix/
touch Firefox-3.5.19-build1-beta-bug629256fix/
touch Firefox-3.6.17-build2-4.0.1-build1-MU-beta-bug629256fix/
touch Firefox-3.5.19-build1-4.0.1-build1-MU-beta-bug629256fix/

touch Firefox-3.6.17-build2-test-bug629256fix/
touch Firefox-3.5.19-build1-test-bug629256fix/
touch Firefox-3.6.17-build2-4.0.1-build1-MU-test-bug629256fix/
touch Firefox-3.5.19-build1-4.0.1-build1-MU-test-bug629256fix/
##### END OF [3]

Comment 28

[Armen [:armenzg]]

(In reply to comment #27)
> [1]
> http://hg.mozilla.org/releases/mozilla-1.9.2/file/FIREFOX_3_6_17_RELEASE/toolkit/mozapps/update/src/updater/readstrings.h#l43

The correct URL:
[1] http://hg.mozilla.org/releases/mozilla-1.9.1/file/FIREFOX_3_5_19_RELEASE/toolkit/mozapps/update/src/updater/readstrings.h

Comment 29

[Robert Strong (they/them - no direct email)]

(In reply to comment #27)
> Created attachment 526374 [details]
> snippets currently live on aus
> 
> Let me say out loud what I think it is going on.
> rstrong could you verify what I am saying?
> 
> > *) For 3.6: ml, te for all versions on branch  (comment #38)
> > *) For 3.5: kn, ml, te for all version on branch (comment #44)
From the summary of bug 628829 bn-IN, kn, ml, and te.

> There is a bug for the aforementioned locales and they are stranded (bug
> 628829).
Correct. The bug goes back to the original creation of the updater and when I last looked it appeared there were under 1200 ADU's for all platforms (this on;y affects Macs) of the affected locales (bug 628829 comment #53).

> A fix got landed that finally made it to 1.9.1 [1] and 1.9.2 [2].
It has not made it into a release vehicle due to chemspills where the release didn't include the fix. The fix should be in 1.9.2.17 and 1.9.1.19 barring another chemspill taking precedence.

> This means that anyone downloading 3.6.17 and 3.5.19 will be able to update
> from now on.
> I assume that no users on those 5 locales could update forward and crashed.
Only Mac users 

I think that covers most of the details.

Comment 30

[Henrik Skupin [:whimboo][⌚️UTC+1]]

(In reply to comment #29)
> > > *) For 3.6: ml, te for all versions on branch  (comment #38)
> > > *) For 3.5: kn, ml, te for all version on branch (comment #44)
> From the summary of bug 628829 bn-IN, kn, ml, and te.

bn-IN was not affected anymore for me when testing the updates 3.6.16->3.6.17 and 3.5.18->3.5.19 on betatest.

http://mozmill-archive.brasstacks.mozilla.com/#/update/report/f552bb95ab494bdcc91d9b001a451ac9
http://mozmill-archive.brasstacks.mozilla.com/#/update/report/f552bb95ab494bdcc91d9b001a4c4b2e

Comment 31

[Armen [:armenzg]]

Attachment: patcher config changes

Comment 32

[Armen [:armenzg]]

Attachment: Update verify configs

Comment 33

[Armen [:armenzg]]

Comment on attachment 526985 [details] [diff] [review]
Update verify configs

This patch has few substitution errors. I will fix that.

Comment 34

[Armen [:armenzg]]

Attachment: Update verify config changes

Comment 35

[Chris Cooper [:coop] (he/him)]

Comment on attachment 526983 [details] [diff] [review]
patcher config changes

>--- moz191-branch-patcher2.cfg	14 Apr 2011 19:00:46 -0000	1.56
>+++ moz191-branch-patcher2.cfg	19 Apr 2011 14:26:37 -0000
>@@ -95,16 +95,18 @@
>                 schema   1
>                 version   3.1b1
>             </3.1b1>
>             <3.1b2>
>                 completemarurl   http://stage-old.mozilla.org/pub/mozilla.org/firefox/nightly/3.1b2-candidates/build2/update/%platform%/%locale%/firefox-3.1b2.complete.mar
>                 <exceptions>
>                     ja   linux-i686, win32
>                     ja-JP-mac   mac
>+                    te   linux-i686, win32
>+                    ml   linux-i686, win32
>                 </exceptions>
>                 extension-version   3.1b2
>                 locales   af ar be bg bn-IN ca cs cy de el en-GB en-US eo es-AR es-ES et eu fi fr fy-NL ga-IE gu-IN he hi-IN hu id is it ja ja-JP-mac ka kn ko lt lv mr nb-NO nl nn-NO pa-IN pl pt-BR pt-PT ro ru si sk sl sq sv-SE te tr uk zh-CN zh-TW
>                 <platforms>
>                     linux-i686   20081201061202
>                     mac   20081201061100
>                     win32   20081201080242
>                 </platforms>

Are you missing a kn exception here?

Everything else looks fine.

Comment 36

[Chris Cooper [:coop] (he/him)]

Comment on attachment 526983 [details] [diff] [review]
patcher config changes

>--- moz191-branch-major-update-patcher2.cfg	15 Apr 2011 17:44:29 -0000	1.31
>+++ moz191-branch-major-update-patcher2.cfg	19 Apr 2011 14:26:37 -0000
>@@ -25,16 +25,19 @@
>             updateType   major
>         </current-update>
>         <release>
>             <3.5.19>
>                 completemarurl   http://stage-old.mozilla.org/pub/mozilla.org/firefox/nightly/3.5.19-candidates/build1/update/%platform%/%locale%/firefox-3.5.19.complete.mar
>                 <exceptions>
>                     ja   linux-i686, win32
>                     ja-JP-mac   mac
>+                    kn   linux-i686, win32
>+                    ml   linux-i686, win32
>+                    te   linux-i686, win32
>                 </exceptions>
>                 extension-version   3.5.19
>                 locales   af ar as be bg bn-BD bn-IN ca cs cy da de el en-GB en-US eo es-AR es-CL es-ES es-MX et eu fa fi fr fy-NL ga-IE gl gu-IN he hi-IN hr hu id is it ja ja-JP-mac ka kk kn ko ku lt lv mk ml mn mr nb-NO nl nn-NO oc or pa-IN pl pt-BR pt-PT rm ro ru si sk sl sq sr sv-SE ta ta-LK te th tr uk vi zh-CN zh-TW
>                 <platforms>
>                     linux-i686   20110414083440
>                     mac   20110414083440
>                     win32   20110414083440
>                 </platforms>

Do we need the exceptions here? The string length change from bug 628829 is in 3.5.19.

Comment 37

[Chris Cooper [:coop] (he/him)]

Comment on attachment 526990 [details] [diff] [review]
Update verify config changes

>diff --git a/release/updates/moz191-firefox-mac-major.cfg b/release/updates/moz191-firefox-mac-major.cfg
>--- a/release/updates/moz191-firefox-mac-major.cfg
>+++ b/release/updates/moz191-firefox-mac-major.cfg
>@@ -1,2 +1,2 @@
> # 3.5.19 macosx
>-release="3.5.19" product="Firefox" platform="Darwin_x86-gcc3-u-ppc-i386" build_id="20110414083440" locales="af ar as be bg bn-BD bn-IN ca cs cy da de el en-GB en-US eo es-AR es-CL es-ES es-MX et eu fa fi fr fy-NL ga-IE gl gu-IN he hi-IN hr hu id is it ja-JP-mac ka kk kn ko ku lt lv mk ml mn mr nb-NO nl nn-NO oc or pa-IN pl pt-BR pt-PT rm ro ru si sk sl sq sr sv-SE ta ta-LK te th tr uk vi zh-CN zh-TW" channel="betatest" from="/firefox/nightly/3.5.19-candidates/build1/mac/%locale%/Firefox 3.5.19.dmg" aus_server="https://aus2.mozilla.org" ftp_server="stage-old.mozilla.org/pub/mozilla.org" to="/firefox/nightly/4.0.1-candidates/build1/mac/%locale%/Firefox 4.0.1.dmg"
>+release="3.5.19" product="Firefox" platform="Darwin_x86-gcc3-u-ppc-i386" build_id="20110414083440" locales="af ar as be bg bn-BD bn-IN ca cs cy da de el en-GB en-US eo es-AR es-CL es-ES es-MX et eu fa fi fr fy-NL ga-IE gl gu-IN he hi-IN hr hu id is it ja-JP-mac ka kk kn ko ku lt lv mk mn mr nb-NO nl nn-NO oc or pa-IN pl pt-BR pt-PT rm ro ru si sk sl sq sr sv-SE ta ta-LK th tr uk vi zh-CN zh-TW" channel="betatest" from="/firefox/nightly/3.5.19-candidates/build1/mac/%locale%/Firefox 3.5.19.dmg" aus_server="https://aus2.mozilla.org" ftp_server="stage-old.mozilla.org/pub/mozilla.org" to="/firefox/nightly/4.0.1-candidates/build1/mac/%locale%/Firefox 4.0.1.dmg"

Again, I don't think we need to remove locales here, because the fix from bug 628829 is in 3.5.19. If we do, you forgot to remove kn anyway. 

>--- a/release/updates/moz192-firefox-mac-major.cfg
>+++ b/release/updates/moz192-firefox-mac-major.cfg
>@@ -1,2 +1,2 @@
> # 3.6.17 macosx
>-release="3.6.17" product="Firefox" platform="Darwin_x86-gcc3-u-ppc-i386" build_id="20110414082317" locales="af ar as ast be bg bn-BD bn-IN ca cs cy da de el en-GB en-US eo es-AR es-CL es-ES es-MX et eu fa fi fr fy-NL ga-IE gd gl gu-IN he hi-IN hr hu id is it ja-JP-mac ka kk kn ko ku lt lv mk ml mr nb-NO nl nn-NO oc or pa-IN pl pt-BR pt-PT rm ro ru si sk sl sq sr sv-SE ta ta-LK te th tr uk vi zh-CN zh-TW" channel="betatest" from="/firefox/nightly/3.6.17-candidates/build2/mac/%locale%/Firefox 3.6.17.dmg" aus_server="https://aus2.mozilla.org" ftp_server="stage-old.mozilla.org/pub/mozilla.org" to="/firefox/nightly/4.0.1-candidates/build1/mac/%locale%/Firefox 4.0.1.dmg"
>+release="3.6.17" product="Firefox" platform="Darwin_x86-gcc3-u-ppc-i386" build_id="20110414082317" locales="af ar as ast be bg bn-BD bn-IN ca cs cy da de el en-GB en-US eo es-AR es-CL es-ES es-MX et eu fa fi fr fy-NL ga-IE gd gl gu-IN he hi-IN hr hu id is it ja-JP-mac ka kk kn ko ku lt lv mk mr nb-NO nl nn-NO oc or pa-IN pl pt-BR pt-PT rm ro ru si sk sl sq sr sv-SE ta ta-LK th tr uk vi zh-CN zh-TW" channel="betatest" from="/firefox/nightly/3.6.17-candidates/build2/mac/%locale%/Firefox 3.6.17.dmg" aus_server="https://aus2.mozilla.org" ftp_server="stage-old.mozilla.org/pub/mozilla.org" to="/firefox/nightly/4.0.1-candidates/build1/mac/%locale%/Firefox 4.0.1.dmg"

Same for 3.6.17

Comment 38

[Armen [:armenzg]]

Attachment: Update verify config changes

More than that! We don't need changes to moz19{1,2}-firefox-mac-major.cfg as we update to 4.0.1 where this if fixed ;)

Comment 39

[Armen [:armenzg]]

Attachment: patcher config changes (take 2)

In these 2 last attachments I have removed any changes to the major update configuration files as they both update to 4.0.1

Comment 40

[Chris Cooper [:coop] (he/him)]

Comment on attachment 527012 [details] [diff] [review]
patcher config changes (take 2)

>Index: moz191-branch-patcher2.cfg
>===================================================================
>RCS file: /cvsroot/mozilla/tools/patcher-configs/moz191-branch-patcher2.cfg,v
>retrieving revision 1.56
>diff -p -U 8 -r1.56 moz191-branch-patcher2.cfg
>--- moz191-branch-patcher2.cfg	14 Apr 2011 19:00:46 -0000	1.56
>+++ moz191-branch-patcher2.cfg	19 Apr 2011 15:49:02 -0000
>@@ -95,16 +95,18 @@
>                 schema   1
>                 version   3.1b1
>             </3.1b1>
>             <3.1b2>
>                 completemarurl   http://stage-old.mozilla.org/pub/mozilla.org/firefox/nightly/3.1b2-candidates/build2/update/%platform%/%locale%/firefox-3.1b2.complete.mar
>                 <exceptions>
>                     ja   linux-i686, win32
>                     ja-JP-mac   mac
>+                    te   linux-i686, win32
>+                    ml   linux-i686, win32
>                 </exceptions>
>                 extension-version   3.1b2
>                 locales   af ar be bg bn-IN ca cs cy de el en-GB en-US eo es-AR es-ES et eu fi fr fy-NL ga-IE gu-IN he hi-IN hu id is it ja ja-JP-mac ka kn ko lt lv mr nb-NO nl nn-NO pa-IN pl pt-BR pt-PT ro ru si sk sl sq sv-SE te tr uk zh-CN zh-TW
>                 <platforms>
>                     linux-i686   20081201061202
>                     mac   20081201061100
>                     win32   20081201080242
>                 </platforms>

Again, kn is in the list of locales for 3.1b2...do we need an exception for it?

Comment 41

[Armen [:armenzg]]

Attachment: patcher config changes (take 3)

Sorry about that.
I have added 'kn' back.

Could you please land this before triggering the new set of builds for 3.5/3.6?

I will land attachment 527008 [details] [diff] [review].

Comment 42

[Armen [:armenzg]]

Comment on attachment 527008 [details] [diff] [review]
Update verify config changes

http://hg.mozilla.org/build/tools/rev/ae2554333345

Comment 43

[Armen [:armenzg]]

(In reply to comment #27)
> Created attachment 526374 [details]
> snippets currently live on aus
From that comment things have changed since we are doing a re-spin of 3.5.19 and 3.6.17.

This means that those last two attachments will makes us generate the right updates on those two branches.

We will still have some snippets live as seen on attachment 526374 [details] but those can be fixed by removing them with:
> > cd /opt/aus2/incoming/3
> > rm -rf Firefox/3.6.{15,16}/Darwin*/*/{ml,te} # we don't touch the latest 3.6.17
> > rm -rf Firefox/3.5.{1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17}/Darwin*/*/{kn,ml,te} # we don't touch the latest 3.5.19
> > # see attachment to see that there isn't anything matching for 3.5.18

Let's wait until after the respins are out to verify this last comment.

Comment 44

[Chris Cooper [:coop] (he/him)]

Comment on attachment 527015 [details] [diff] [review]
patcher config changes (take 3)

Checking in moz191-branch-major-update-patcher2.cfg;
/cvsroot/mozilla/tools/patcher-configs/moz191-branch-major-update-patcher2.cfg,v  <--  moz191-branch-major-update-patcher2.cfg
new revision: 1.32; previous revision: 1.31
done

Comment 45

[Chris Cooper [:coop] (he/him)]

(In reply to comment #44)
> Comment on attachment 527015 [details] [diff] [review]
> patcher config changes (take 3)
> 
> Checking in moz191-branch-major-update-patcher2.cfg;
> /cvsroot/mozilla/tools/patcher-configs/moz191-branch-major-update-patcher2.cfg,v
>  <--  moz191-branch-major-update-patcher2.cfg
> new revision: 1.32; previous revision: 1.31
> done

Erp, wrong check-in. These are the one's for this bug:

Checking in moz191-branch-patcher2.cfg;
/cvsroot/mozilla/tools/patcher-configs/moz191-branch-patcher2.cfg,v  <--  moz191-branch-patcher2.cfg
new revision: 1.57; previous revision: 1.56
done
Checking in moz192-branch-patcher2.cfg;
/cvsroot/mozilla/tools/patcher-configs/moz192-branch-patcher2.cfg,v  <--  moz192-branch-patcher2.cfg
new revision: 1.50; previous revision: 1.49
done

Comment 46

[Armen [:armenzg]]

I tried to help as best as I could.
This is not my area. Passing it back to nthomas.

As mentioned on rail's notes he had to disable the update bumping for this changes not to be flushed out (IIUC):
https://wiki.mozilla.org/Releases/Firefox_3.6.17/BuildNotes#Updates_2

Comment 47

[[e-mail] (:bhearsum)]

kn, ml, and te 3.5.18 Mac builds have active updates on the "beta" channel, pointing them at 3.5.19build1. Is that intentional/expected?

Comment 48

[Armen [:armenzg]]

Not intentional from my part.

Comment 49

[[e-mail] (:bhearsum)]

> (In reply to comment #27)
> > A fix got landed that finally made it to 1.9.1 [1] and 1.9.2 [2].
> It has not made it into a release vehicle due to chemspills where the
> release didn't include the fix. The fix should be in 1.9.2.17 and 1.9.1.19
> barring another chemspill taking precedence.

It does appear that these fixes made it in. Does that mean we should be back to normal with these locales? Eg, treat them like every other one?

Comment 50

[Robert Strong (they/them - no direct email)]

Yes. For the record, the update will still fail for these locales when updating from a version without the fix but downloading / installing the latest version will fix it and as previously noted there are only around 1200 total ADU's for the affected locales and this only affects Mac.

Comment 51

[[e-mail] (:bhearsum)]

(In reply to comment #50)
> Yes. For the record, the update will still fail for these locales when
> updating from a version without the fix but downloading / installing the
> latest version will fix it and as previously noted there are only around
> 1200 total ADU's for the affected locales and this only affects Mac.

Great.

It looks like we need to re-remove ml & te from the Firefox 3.6 minor update verify configs, but that's it? (patcher configs are already updated, and 1.9.1 is dead so we don't need to worry about it). Nick, does that sound right to you? I'm happy to put the finishing touches on this bug.

Comment 52

[Nick Thomas [:nthomas] (UTC+12)]

Attachment: [tools] Get patcher and update verify configs in sync for 1.9.2

If we just want to do 3.6.17 -> 3.6.18 (and any major updates) then the patcher configs look fine already (ie no <exceptions> in the 3.6.17 and 3.6.18), and we'd just want to put ml and te back into moz192-firefox-mac-major.cfg for the 3.6.17 full checks - for consistencies sake.

I agree that 1.9.1 is not worth fixing, and we've never done 4.0 properly as far as I can see, when it was relevant for 4.0b8/b9/b10 and fixed in b11. Users have figured that out and managed to get to 4.0.1 by themselves.

*sharpens stake*

Comment 53

[[e-mail] (:bhearsum)]

Comment on attachment 541595 [details] [diff] [review]
[tools] Get patcher and update verify configs in sync for 1.9.2

Yay!

Comment 54

[Nick Thomas [:nthomas] (UTC+12)]

Comment on attachment 541595 [details] [diff] [review]
[tools] Get patcher and update verify configs in sync for 1.9.2

http://hg.mozilla.org/build/tools/rev/5368302589ef