649779 - Don't force *.chk in MARs

Status: RESOLVED FIXED

(Release Engineering :: General, defect, P5)

Description

[Rail Aliiev [:rail]]

Bug 313956 fixes bug 404340, so probably it would be better to not force *.chk files in update.manifest.

Comment 1

[Rail Aliiev [:rail]]

Attachment: patcher-configs

Comment 2

[Rail Aliiev [:rail]]

Attachment: update-packaging

Comment 3

[Rail Aliiev [:rail]]

Attachment: tools

Comment 4

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

Do the zip builds also use the same .chk files? I'm not terribly concerned if they don't but there will likely be app update bug reports for failing partials if they aren't the same.

Comment 5

[Rail Aliiev [:rail]]

(In reply to comment #4)
> Do the zip builds also use the same .chk files? 

AFAIK, zip files are used only for tests. We don't use them for MAR generation.

Comment 6

[Rail Aliiev [:rail]]

FTR, see also bug 404340 comment 44 and below.

Comment 7

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

Nightly users also download zip builds and update from them. So, when a nightly user downloads a zip build and the next day gets a partial update if the chk files don't match the partial will fail.

This change basically moves the problem over to people using zips. IMO that is better than where we were and don't really think it is a case that needs to be fixed. iirc there is also a bug for generating the zips from the installer as well which would also fix it.

Comment 8

[[e-mail] (:bhearsum)]

(In reply to comment #7)
> Nightly users also download zip builds and update from them. So, when a nightly
> user downloads a zip build and the next day gets a partial update if the chk
> files don't match the partial will fail.
> 
> This change basically moves the problem over to people using zips. IMO that is
> better than where we were and don't really think it is a case that needs to be
> fixed. iirc there is also a bug for generating the zips from the installer as
> well which would also fix it.

ZIP users can fall back to completes in the meantime.

Comment 9

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

Agreed... just popped in to verify my suspicion that partials for zip builds will fail so if I get bug reports I can move them to the correct component vs. trying to figure out whether or not something with app update broke. Thanks!

Comment 10

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

Comment on attachment 525784 [details] [diff] [review]
patcher-configs

I don't think we can do this if bug 313956 has only landed on m-c.

Comment 11

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

Comment on attachment 525785 [details] [diff] [review]
update-packaging

This seems fine to land anywhere bug 313956 has landed.

Comment 12

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

Comment on attachment 525786 [details] [diff] [review]
tools

This seems fine for new configs on new branches (where bug 313956 has landed).

Comment 13

[Rail Aliiev [:rail]]

Back to pool, maybe next Q...

Comment 14

[[e-mail] (:bhearsum)]

We stopped doing this a long time ago, probably around the same time bug 404340 ang bug 489961 were fixed.