1430834 - fennec: push-apk should also depend on beetmover-checksums

Status: RESOLVED FIXED

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

Description

[Kim Moir [:kmoir] ET]

When we trigger the publish graph before the promote graph is finished, push-apk-breakpoint triggers after the builds finish, rather than waiting for the promote graph to finish.

This is optional, but is probably what we want. Not a blocker.

Comment 1

[Kim Moir [:kmoir] ET]

Attachment: bug1430834.patch

Comment 2

[Kim Moir [:kmoir] ET]

generated json for maple-ship-fennec.json

"push-apk/opt": {
        "attributes": {
            "always_target": false,
            "build_platform": "android-nightly",
            "kind": "push-apk",
            "nightly": true, 
            "run_on_projects": [
                "mozilla-central",
                "mozilla-beta",
                "mozilla-release",
                "maple"
            ],    
            "shipping_phase": "ship",
            "shipping_product": "fennec"
        },    
        "dependencies": {
            "beetmover-checksums": "beetmover-checksums-de-android-api-16-nightly/opt",
            "build-signing-api-16": "build-signing-android-api-16-nightly/opt",
            "build-signing-x86": "build-signing-android-x86-nightly/opt",
            "google-play-strings": "google-play-strings",
            "push-apk-breakpoint": "android-push-apk-breakpoint/opt"
        },    
        "kind": "push-apk",
        "label": "push-apk/opt",
        "optimization": null,

Comment 3

[Mihai Tabara [:mtabara]⌚️GMT]

Comment on attachment 8942961 [details] [diff] [review]
bug1430834.patch

Might be worth using `post-beetmover-checksums-dummy` rather than `beetmover-checksums` as in a full-locale graph, we'd have more than the 100 max limit dependencies. 

In this particular case, it's optional as we don't have single-locales turned on so we never more than a handful of dependencies there.

Comment 4

[Mihai Tabara [:mtabara]⌚️GMT]

Completion to my previous comment: in Fennec, we have single-locales deactivated AFAIK so it doesn't necessarily make sense. But if we ever turn on those back again, we will hit this.

Comment 5

[Kim Moir [:kmoir] ET]

Comment on attachment 8942961 [details] [diff] [review]
bug1430834.patch

I tried adding post-beetmover-checksums-dummy but it didn't work, so perhaps this patch is good for now.

Comment 6

[Mihai Tabara [:mtabara]⌚️GMT]

Comment on attachment 8942961 [details] [diff] [review]
bug1430834.patch

Review of attachment 8942961 [details] [diff] [review]:
-----------------------------------------------------------------

Sgtm! 
Out of curiosity, why didn't it work? Was there errors at generation time in-graph or elsewhere?

Comment 7

[Aki Sasaki (not active)]

Landed https://hg.mozilla.org/projects/maple/rev/cd6b97fce1ec on maple to avoid adding target.apk from beetmover-checksums to push-apk upstreamArtifacts.

Comment 8

[Aki Sasaki (not active)]

Attachment: fix-pushapk.diff

Comment 9

[Johan Lorenzo [:jlorenzo]]

Comment on attachment 8946359 [details] [diff] [review]
fix-pushapk.diff

Review of attachment 8946359 [details] [diff] [review]:
-----------------------------------------------------------------

Makes sense to me

Comment 10

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/mozilla-inbound/rev/f4f1f4df5e76
fennec: push-apk should also depend on beetmover-checksums r=mtabara
https://hg.mozilla.org/integration/mozilla-inbound/rev/0d40d859cff0
don't add beetmover-checksums to push-apk upstreamArtifacts. r=jlorenzo

Comment 11

[Sebastian Hengst [:aryx] (needinfo me if it's about an intermittent or backout)]

https://hg.mozilla.org/mozilla-central/rev/f4f1f4df5e76
https://hg.mozilla.org/mozilla-central/rev/0d40d859cff0