Test-infra: Add branch filtering in deck/gubernator/testgrid

Created on 4 Jan 2018 · 17Comments · Source: kubernetes/test-infra

@mfojtik wanted to know whether a job fails across a release branch but couldn't figure it out. We should probably have an extra filter for branches in deck.

cc: @cjwagner @stevekuznetsov @rmmh @BenTheElder

/area prow

areprow areprodeck aretestgrid kinfeature lifecyclrotten

Source

kargakis

👍1

All 17 comments

Deck isn't really the right place to look for this for the kubernetes deployment at least, considering how often sinker runs, the best places are testgrid and gubernator. Deck is really best at recently run jobs.

More filtering options could be handy though.

BenTheElder on 7 Jan 2018

Gubernator was the first place that @mfojtik was looking at so I am fine with it too.

kargakis on 7 Jan 2018

I've made this a catch-all issue and need others to weigh in about what's the best place to do it.

cc: @stevekuznetsov @rmmh @michelle192837

kargakis on 7 Jan 2018

👍1

Yeah, i'm fine with whatever tool I can use. It is very useful to see the runs of the same job per-branch, so I can see if the job is broken just for me or for other PR's into that branch.

mfojtik on 8 Jan 2018

We're recording the Prow job configuration but I don't think k8s jobs are, so Gubernator would need more data to make that decision. One of the natural places for this is where the data is declared, and that's from Deck. We could probably add best-effort filtering in Gubernator as well.

stevekuznetsov on 8 Jan 2018

We're recording the Prow job configuration but I don't think k8s jobs are, so Gubernator would need more data to make that decision. One of the natural places for this is where the data is declared, and that's from Deck. We could probably add best-effort filtering in Gubernator as well.

With ditching Jenkins I'd actually love to record these somewhere Gubernator could read them, it would be great to look back on a job and be able to completely recreate it.

BenTheElder on 8 Jan 2018

👍1

Could this be a periodic BigQuery job? That's the best source of all the test results.

rmmh on 9 Jan 2018

Would be nice to not need to bring in more (heavyweight) machinery to expose something in the views we currently use to look at job status and history (outside of TestGrid)

stevekuznetsov on 9 Jan 2018

Older request: https://github.com/kubernetes/test-infra/issues/3449

kargakis on 12 Jan 2018

👍1

/cc @qhuynh96
who is currently looking at replacing some of the existing dropdowns with fuzzy-search / dropdown upgrades

BenTheElder on 12 Jan 2018

Issues go stale after 90d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.

If this issue is safe to close now please do so with /close.

Send feedback to sig-testing, kubernetes/test-infra and/or fejta.
/lifecycle stale

fejta-bot on 12 Apr 2018

Issues go stale after 90d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.

If this issue is safe to close now please do so with /close.

Send feedback to sig-testing, kubernetes/test-infra and/or fejta.
/lifecycle stale

fejta-bot on 24 Jul 2018

/lifecycle frozen

kargakis on 5 Aug 2018

I think it would be pretty useful to show branch for PR like we do for postsubmit. Filtering per branch is also very useful when wanting to know if a job is failing on multiple PRs.

sebastienvas on 15 Mar 2019

/remove-lifecycle frozen
I'm taking no action on this in ~2 years as a sign that it's no longer that important
/remove-area gubernator
we're not doing new development for gubernator

/area prow/deck
I think this could make deck more useful.

Alternatively, maybe we could consider renaming all branch-specific presubmits something different. Having jobs named the same makes it difficult to effectively triage in multiple places (testgrid, spyglass job history, triage.k8s.io)

spiffxp on 11 Sep 2020

Issues go stale after 90d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.

If this issue is safe to close now please do so with /close.

Send feedback to sig-testing, kubernetes/test-infra and/or fejta.
/lifecycle stale

fejta-bot on 10 Dec 2020

Stale issues rot after 30d of inactivity.
Mark the issue as fresh with /remove-lifecycle rotten.
Rotten issues close after an additional 30d of inactivity.

If this issue is safe to close now please do so with /close.

Send feedback to sig-testing, kubernetes/test-infra and/or fejta.
/lifecycle rotten

fejta-bot on 9 Jan 2021

Was this page helpful?

0 / 5 - 0 ratings