1290179 - Automate SHA1 repacks

Status: RESOLVED FIXED

(Release Engineering :: Release Automation: Other, defect)

Description

[Rail Aliiev [:rail]]

https://bugzilla.mozilla.org/show_bug.cgi?id=1284484#c5 gives some idea how to run them.

* Modify the repack script to accept a config variable which allows skipping the following steps:
  self.copyFiles()
  self.addPadding()
  self.internallySignBuild()
  self.repackBuild()

* create repack configs

* schedule the repacks similar to https://github.com/rail/releasetasks/blob/master/releasetasks/templates/firefox/partner_repacks.yml.tmpl#L47-L88

* make sure to exclude them from the main push similar to eme-free in https://github.com/rail/releasetasks/blob/master/releasetasks/templates/firefox/push_to_releases.yml.tmpl#L81 (they will be copied by https://github.com/rail/releasetasks/blob/master/releasetasks/templates/firefox/partner_repacks.yml.tmpl#L129)

* Create aliases for installers (Firefox-{,beta,esr}-sha1?) aliases in bouncer

* add new bouncer product in https://dxr.mozilla.org/mozilla-central/source/testing/mozharness/configs/releases/bouncer_firefox_release.py and friends, make sure to use the alias, so we update it automatically.


Updating the alias may lead to a race condition, especially for betas, when we have repacks still in progress, but we publish to beta. Not sure how to prevent this... I don't really want to block publishing a release on partner repacks. Maybe we can split the update aliases step into 2 and make the second depend on partner repacks.

Comment 1

[Justin Wood (:Callek)]

Rail,

Before we go down this path, is it actually worthwhile to do an SHA1 signed installed based off an installer for a Firefox version that requires SSE2?

Specifically-->

* User goes to firefox.com to download the browser, on windows, in a system that we can't detect if they are new enough for SHA2 (and thus also can't presumably detect SSE2 support)
* User needs to download a Firefox that doesn't require SSE2 (Firefox 48 being the last in that set)
* User can then update to a newer Firefox that is signed without SHA1 via our updater code, AND Firefox+Balrog will also detect if the user can download/run a Firefox with SSE2 as a requirement.

If however, we don't deem lack of SSE2 Support to be a blocker to detect on the website front, we certainly can do this [SHA1] with newer versions on the download page itself.

I'm not sure who makes that call, on the user facing clickthrough paths, but I think it makes sense to get that call before investing in this work.

Comment 2

[Rail Aliiev [:rail]]

Attachment: Bug 1290179 - Automatically update SHA1 bouncer aliases a=release DONTBUILD

Review commit: https://reviewboard.mozilla.org/r/72642/diff/#index_header
See other reviews: https://reviewboard.mozilla.org/r/72642/

Comment 3

[Justin Wood (:Callek)]

Comment on attachment 8782477 [details]
Bug 1290179 - Automatically update SHA1 bouncer aliases  a=release DONTBUILD

https://reviewboard.mozilla.org/r/72642/#review70282

Comment 4

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/mozilla-inbound/rev/2c239b4efab7
Add bouncer products for SHA1 installers r=Callek a=release DONTBUILD

Comment 5

[Rail Aliiev [:rail]]

Comment on attachment 8782477 [details]
Bug 1290179 - Automatically update SHA1 bouncer aliases  a=release DONTBUILD

https://hg.mozilla.org/integration/mozilla-inbound/rev/2c239b4efab7
https://hg.mozilla.org/releases/mozilla-aurora/rev/18102c4b2276
https://hg.mozilla.org/releases/mozilla-beta/rev/98f83e513e45
https://hg.mozilla.org/releases/mozilla-release/rev/74c318753778
https://hg.mozilla.org/releases/mozilla-esr45/rev/0d30594f9725

Comment 6

[Rail Aliiev [:rail]]

I'll keep an eye on this

Comment 7

[[Back August 16] Ryan VanderMeulen]

https://hg.mozilla.org/mozilla-central/rev/2c239b4efab7

Comment 8

[Rail Aliiev [:rail]]

Attachment: PR

Let's try mkaply this time ;)

Comment 9

[Mike Kaply [:mkaply]]

Comment on attachment 8784911 [details] [review]
PR

reviewed and merged.

Comment 10

[Rail Aliiev [:rail]]

Attachment: PR for releasetasks

Comment 11

[Rail Aliiev [:rail]]

Worked fine in Firefox 49.0b8. Still need to enable aliases (https://hg.mozilla.org/integration/mozilla-inbound/rev/2c239b4efab7#l1.16 and friends) after bug 1295275 is live

Comment 12

[Rail Aliiev [:rail]]

Comment on attachment 8782477 [details]
Bug 1290179 - Automatically update SHA1 bouncer aliases  a=release DONTBUILD

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/72642/diff/1-2/

Comment 13

[Jordan Lund (:jlund)]

Comment on attachment 8782477 [details]
Bug 1290179 - Automatically update SHA1 bouncer aliases  a=release DONTBUILD

https://reviewboard.mozilla.org/r/72642/#review73292

Comment 14

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/mozilla-inbound/rev/64b21c1e7b92
Automatically update SHA1 bouncer aliases r=jlund a=release DONTBUILD

Comment 15

[Rail Aliiev [:rail]]

https://hg.mozilla.org/integration/mozilla-inbound/rev/64b21c1e7b92
https://hg.mozilla.org/releases/mozilla-aurora/rev/845d630fe4a9
https://hg.mozilla.org/releases/mozilla-beta/rev/a947fabffddd
https://hg.mozilla.org/releases/mozilla-release/rev/0e808827a3ed

Comment 16

[[Back August 16] Ryan VanderMeulen]

https://hg.mozilla.org/mozilla-central/rev/64b21c1e7b92

Comment 17

[Rail Aliiev [:rail]]

Worked fine in 49.0b9

Comment 18

[Firefox Bug Husbandry Bot]

Removing leave-open keyword from resolved bugs, per :sylvestre.