Closed Bug 1248474 Opened 9 years ago Closed 9 years ago

[RelPromo] Run Mac/Windows update verification via BBB

Categories

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

defect
Not set
normal

Tracking

(Not tracked)

RESOLVED FIXED

People

(Reporter: Callek, Assigned: Callek)

References

Details

Attachments

(7 files, 2 obsolete files)

Mac and Windows should also get update verification run via the Task Graph through Buildbot bridge.
Attached patch [custom] update_verify (obsolete) (deleted) — Splinter Review
My largest concern here is the WithProperties for the script repo. I made the call to let it generate builders for linux as well, rather than hardcode which platforms are *not* via TC workers. (I can change that decision if desired). This hasn't been tested beyond making sure checkconfig passes and a dump of the builderlist.
Assignee: nobody → bugspam.Callek
Status: NEW → ASSIGNED
Attachment #8719580 - Flags: review?(rail)
Comment on attachment 8719580 [details] [diff] [review] [custom] update_verify Review of attachment 8719580 [details] [diff] [review]: ----------------------------------------------------------------- Some nits. ::: process/release.py @@ +1842,5 @@ > > + uv_fmt_template = "release-{branch}_{platform}_update_verify_{channel}_{chunk}/{chunks}" > + for platform in branch_config.get("release_platforms"): > + pf = branch_config["platforms"][platform] > + for channel in branch_config.get('release_channels'): Hmmm... So, for an RC it would be beta and release, for a dot release - release... Probably it's ok to keep both builders around, but let's make the branch config variable more verbose to not clash (at least mentally) with similar variable we use for the updates builder, which will be set by releaserunner. Maybe something like possible_release_channels? @@ +1843,5 @@ > + uv_fmt_template = "release-{branch}_{platform}_update_verify_{channel}_{chunk}/{chunks}" > + for platform in branch_config.get("release_platforms"): > + pf = branch_config["platforms"][platform] > + for channel in branch_config.get('release_channels'): > + for n in range(1, 7): Can you move this to branch config? @@ +1849,5 @@ > + branch=branch_name, > + platform=platform, > + channel=channel, > + chunk=n, > + chunks=6, And this 6. @@ +1856,5 @@ > + uv_factory = ScriptFactory( > + scriptRepo=WithProperties("%(script_repo_path)s"), > + interpreter='bash', > + scriptName='scripts/release/updates/chunked-verify.sh', > + extra_args=["UNUSED", "UNUSED", str(6), str(n)], And this 6.
Attachment #8719580 - Flags: review?(rail)
(In reply to Rail Aliiev [:rail] from comment #2) > Comment on attachment 8719580 [details] [diff] [review] > [custom] update_verify > > Review of attachment 8719580 [details] [diff] [review]: > ----------------------------------------------------------------- > > Some nits. > > ::: process/release.py > @@ +1842,5 @@ > > > > + uv_fmt_template = "release-{branch}_{platform}_update_verify_{channel}_{chunk}/{chunks}" > > + for platform in branch_config.get("release_platforms"): > > + pf = branch_config["platforms"][platform] > > + for channel in branch_config.get('release_channels'): > > Hmmm... So, for an RC it would be beta and release, for a dot release - > release... Probably it's ok to keep both builders around, but let's make the > branch config variable more verbose to not clash (at least mentally) with > similar variable we use for the updates builder, which will be set by > releaserunner. Maybe something like possible_release_channels? I'm already using release_channels in the uv stuff for the taskgraph, and we may want/need to revisit that, but I don't think its important for our beta setup, I'm going to submit next patch with that still in place [for now] unless you have a strong objection. > > @@ +1843,5 @@ > > + uv_fmt_template = "release-{branch}_{platform}_update_verify_{channel}_{chunk}/{chunks}" > > + for platform in branch_config.get("release_platforms"): > > + pf = branch_config["platforms"][platform] > > + for channel in branch_config.get('release_channels'): > > + for n in range(1, 7): > > Can you move this to branch config? Will do with the Num-Chunks
Comment on attachment 8720597 [details] [tools] update script to accept "no buildbot mode" from properties json a well. I didn't even look at the update verify configs - we regenerate them every time. There is even bug 1210527!
Attachment #8720597 - Flags: review?(rail) → review+
Attachment #8720792 - Flags: review?(rail)
Attachment #8719580 - Attachment is obsolete: true
Attachment #8720793 - Flags: review?(rail)
Attachment #8720793 - Flags: review?(rail) → review+
Attachment #8720792 - Flags: review?(rail) → review+
Attachment #8721409 - Flags: review?(rail) → review+
Attachment #8721513 - Flags: review?(rail)
Previous patch was my untested version - forgot to hg qref :/
Attachment #8721513 - Attachment is obsolete: true
Attachment #8721513 - Flags: review?(rail)
Attachment #8721514 - Flags: review?(rail)
Comment on attachment 8721514 [details] [diff] [review] [tools] Don't expect us to pass in slavebuilddir lgtm
Attachment #8721514 - Flags: review?(rail) → review+
Attachment #8722075 - Flags: review?(rail) → review+
Attachment #8722131 - Flags: review?(rail)
Comment on attachment 8722075 [details] [diff] [review] [custom] don't use a slash in buildername for buildbot uv jobs https://hg.mozilla.org/build/buildbotcustom/rev/c87b90ed718d
Comment on attachment 8722131 [details] [releasetasks] add the bbb update verifies r+ with a nit in the comments.
Attachment #8722131 - Flags: review?(rail) → review+
Status: ASSIGNED → RESOLVED
Closed: 9 years ago
Resolution: --- → FIXED
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: