1728330 - Stop publishing the Snap to the Snap store

Status: RESOLVED FIXED

(Release Engineering :: Release Automation: Snap, task)

Description

[Mike Kaply [:mkaply]]

With the transition of the Snap, we need to stop the final step of publishing to the store.

We should keep all the other work.

Per discussions, this was going to happen for 92.

Unfortunately I forgot to file this bug earlier.

Comment 1

[Aki Sasaki (not active)]

Attachment: [pushsnapscript] Disable pushing to the store

Comment 2

[Aki Sasaki (not active)]

This is a full "don't push to the store at all", so we will need to

1. disable the ability of the pushsnap pool to push to the store pre-92.0 push Tuesday - that's the PR above, to roll out once I get the tests green
2. delete the release-{,secondary-}snap-push kinds and any transforms etc specific to those kinds
3. uplift patch from 2 to release trains (do we do esr?)
4. delete pushsnapscript from scriptworker-scripts, delete pools (k8s-autoscale, taskcluster-cloudops, k8s-sops)

Comment 3

[Aki Sasaki (not active)]

Attachment: Bug 1728330 - remove pushsnap tasks. r=#releng

Comment 4

[Aki Sasaki (not active)]

Attachment: [scriptworker-scripts] delete pushsnapscript

Comment 5

[Aki Sasaki (not active)]

Attachment: [k8s-autoscale] delete pushsnapscript

Comment 6

[Aki Sasaki (not active)]

1 is done. 2 is the gecko patch above, which we'll uplift for step 3.
I have 2 of the four delete pushsnapscript patches attached. 3/4 is here; 4/4 is removing the secrets in the k8s sops repo.

Comment 7

[Aki Sasaki (not active)]

(2) is scheduled to Lando.
I'll leave steps 3 and 4 for next week, after the long weekend. Because (1) landed already, the pushsnap task for 92 should be noop.

Comment 8

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/0f0309f1f8b6
remove pushsnap tasks. r=releng-reviewers,bhearsum DONTBUILD

Comment 9

[Aki Sasaki (not active)]

Ben is suggesting we leave steps 3 and 4 for a cycle or two, until we know the distro builds are stable.
That sounds like it could be a good idea, especially if backing out the removal is non-trivial.

Comment 10

[Marian-Vasile Laza]

https://hg.mozilla.org/mozilla-central/rev/0f0309f1f8b6

Comment 11

[Aki Sasaki (not active)]

Attachment: reenable pushsnap for esr78

Comment 12

[Aki Sasaki (not active)]

https://hg.mozilla.org/releases/mozilla-release/rev/bfc96f4a3e02dcfad906c46f03ab8aa95253ce52
https://hg.mozilla.org/releases/mozilla-esr91/rev/45bc8d1f4315b5380728ee2265c5abe036e851d5

We got mozilla-beta for free on merge-day yesterday. So unless we do a 92 dot build off a relbranch, we should be clear of pushsnap tasks on all release trains except esr78. I'll reenable the push in the pushsnap pool, then rerun the esr78 pushsnap task.

Comment 13

[Aki Sasaki (not active)]

Current status:

• I was not able to use the retrigger (decision) action, because of scopes. Perhaps if I went through the TC UI instead of Treeherder? But:
• I was able to craft a relpro :snowman: graph with just the push snap task on esr78. We should get this task for free for the last esr78 release next month.
• Once esr78 is EOLed, we can proceed removing the pushsnap pool entirely, but until then let's leave this bug open.

Comment 14

[Aki Sasaki (not active)]

Attachment: Bug 1728330 - remove pushsnap scopes. r=#releng

Comment 15

[Aki Sasaki (not active)]

Attachment: [scriptworker] remove pushsnap support

Comment 16

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/ci/ci-configuration/rev/f992905fe25d
remove pushsnap scopes. r=releng-reviewers,jcristau

Comment 17

[Aki Sasaki (not active)]

I also removed the secrets from our k8s sops repo.