1152457 - Which add-ons should we sign at first?

Status: RESOLVED FIXED

(addons.mozilla.org Graveyard :: Administration, defect, P1)

Description

[Wil Clouser [:clouserw]]

We talked about all the Mozilla and Reviewer ones.  Can we get a list of URLs or IDs or whatever so we can sign a bunch and make sure things work?  Thanks!

Comment 1

[Kris Maglione [:kmag]]

I've made a list of all add-ons owned by reviewers or users @mozilla.com email addresses. I'll email these people today to give them a chance to opt out, and post IDs of those who haven't replied by Monday.

Comment 2

[Kris Maglione [:kmag]]

Attachment: Add-on IDs for initial run

Here's the final list of add-on IDs. No-one opted out, but I did exclude any user accounts whose email bounced.

Comment 3

[Wil Clouser [:clouserw]]

Attachment: signme.sql

converted to SQL

Comment 4

[Wil Clouser [:clouserw]]

Jlaz:  can you run this on production and give us the output so we have the IDs and not the GUIDs?

I think this would work:      mysql < signme.sql > output.txt

Comment 5

[Justin Lazaro [:jlaz] (use needinfo)]

Attachment: output.txt

Results from sql

Comment 6

[Kris Maglione [:kmag]]

Attachment: List of IDs

It seems I forgot to exclude the hotfix from my list. Attaching a new list in a slightly more convenient format (bare ID numbers, one per line) with that corrected.

Comment 7

[Kris Maglione [:kmag]]

Attachment: auto-sign-ids.txt

Removed Thunderbird Hotfix (which should be automatically skipped in any case)

Comment 8

[Mathieu Agopian [:magopian]]

Attachment: test_versions.txt

I've started checking manually the first few dozens and it seems everything went as expected.

I've made a list of the corresponding versions pages for each addon (that exists on stage), attached to the bug.

There's one specific addon I see has been signed twice (but not for all its versions), which is odd, and I'll be investigating this a bit more using the logs.

Comment 9

[Mathieu Agopian [:magopian]]

The addon that got some of its versions signed twice: https://addons.allizom.org/fr/firefox/addon/about-addons-memory/versions/ (addon id 424070)

Comment 10

[Mathieu Agopian [:magopian]]

I found (and hopefully fixed) the issue for this addon in bug 1156768

Here's a summary of what I saw in the logs following our first attempt at signing all the addons on stage:
- it took roughly two hours to run for 4497 files 
- 3076 files were signed
- 954 files were not signed because not reviewed (unreviewed or beta)
- 6 failed signing
  - 1 bad zipfile: empty XPI on the filesystem for the "google-plus-new-tab" addon
  - 5 failures because of lxml while parsing invalid/garbled install.rdf file for the "noscript" addon
- 594 files were not signed because not present on the filesystem

The numbers don't add up because of bug 1156768: some files where not signed nor logged.

Comment 11

[Kris Maglione [:kmag]]

Hmm... That's a lot of runtime... Can we skip signing files that aren't compatible with with Firefox 39+?