640361 - For some 4b12 locales complete updates are offered for 4RC1 updates

Status: RESOLVED INCOMPLETE

(Release Engineering :: General, defect)

Description

[Henrik Skupin [:whimboo][⌚️UTC+2]]

I did a full locale test-run for updates from the Firefox 4 b12 release to Firefox 4 RC 1 on releasetest channel, and as what I can see from our results we offer complete updates for some locales. Check the given URL and search for 'complete' to find all instances.

Not sure if that's wanted but I would assume we always offer partial updates.

Comment 1

[[e-mail] (:bhearsum)]

A mozmill result, from pastebin:
Status: 	PASS
Build (pre): 	Mozilla/5.0 (Windows NT 5.1; rv:2.0b12) Gecko/20100101 Firefox/4.0b12 (pt-BR, 20110222210221)
Build (post): 	Mozilla/5.0 (Windows NT 5.1; rv:2.0) Gecko/20100101 Firefox/4.0 (pt-BR, 20110303194838)
Channel: 	releasetest
Type: 	minor, complete, direct
Download: 	http://mirror.informatik.uni-mannheim.de/pub/mirrors/mozilla.org/firefox/releases/4.0rc1/update/win32/pt-BR/firefox-4.0rc1.complete.mar (16089654 Byte, 125052 ms)
Disabled Add-ons:

Comment 2

[[e-mail] (:bhearsum)]

The AUS result from https://aus2.mozilla.org/update/3/Firefox/4.0b12/20110222210221/WINNT_x86-msvc/pt-BR/releasetest/Darwin%2010.5.0/default/default/update.xml points at a partial, specifically:
http://download.mozilla.org/?product=firefox-4.0rc1-partial-4.0b12&os=win&lang=pt-BR

And indeed, that ends up getting me a partial MAR.

Hard to debug without full HTTP headers. Can you reproduce this still, whimboo?

Comment 3

[u279076]

http://mozmill-archive.brasstacks.mozilla.com/#/update/report/f179c276497eff82ec93f78b1eddc29b

I just tried to verify the 4.0b12-da build and I get a partial update when testing manually. It applies cleanly without any issues.

Comment 4

[[e-mail] (:bhearsum)]

Can you pastebin that report? Us poor build-vpn users can't access brasstacks.

Comment 5

[u279076]

(In reply to comment #4)
> Can you pastebin that report? Us poor build-vpn users can't access brasstacks.

Raw report data:
http://pastebin.mozilla.org/1141303

Comment 6

[[e-mail] (:bhearsum)]

So far, we've been unable to reproduce this by hand nor have we seen the issue during the mozmill checks of the "beta" channel. If we don't see any issues on that channel during the full check of it, this bug's priority is much lower.

Regardless, we'll retest the releasetest channel when we can, and see if mozmill can repro there.

Comment 7

[Henrik Skupin [:whimboo][⌚️UTC+2]]

It can't be reproduced anymore. Last testrun for all locales was clean. Could that only happen in the first hours after opening this channel?

http://mozmill-archive.brasstacks.mozilla.com/#/update/detail?branch=4.0&channel=releasetest&from=2011-03-10&to=2011-03-10&target=4.0

Comment 8

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

Would we see the same effect if a partial was offered, and the download failed, or didn't match the hash/size expected ?

Comment 9

[Henrik Skupin [:whimboo][⌚️UTC+2]]

Well, in all those cases I run updates from b12->RC1 which should offer partial and completes together. Given that everything works now, it looks like inconsistencies in the snippet. I can't say anything about the client updater. We only use the patch info of the activeUpdate.selectedPatch.

Comment 10

[[e-mail] (:bhearsum)]

I don't know how else to debug this at this point, so I'm closing as INCOMPLETE. I've filed bug 640524 about getting even more debugging info out of Mozmill for failed update tests.