Use k8s-triage-robot in lieu of fejta-bot for kubernetes issue/pr commenter jobs

[nikhita]

Background: @fejta-bot automatically marks issues as stale/rotten and closes them if they don't have any activity for certain number of days. It is an unprivileged user and is not a member of any of the Kubernetes GitHub orgs.

While enabling fejta-bot automation across all Kubernetes GitHub orgs, it was suggested to rename it to something "official": https://groups.google.com/d/topic/kubernetes-dev/9ps8uNXJ10k/discussion

This issue is to track discussing renaming fejta-bot - what to rename it as or if we should use another account entirely - and making sure that this is rolled out across orgs.

/sig contributor-experience
/sig testing
/cc @cblecker @fejta @spiffxp

[fejta]

serious-fejta-bot with a grumpy cat partially converted into a robot for it's profile pic

[nikhita]

/assign

Assigning it to myself to try to move this along in contribex

[stevekuznetsov]

Any reason not to use one of the existing robots?

[spiffxp]

We need prow to react to the comments posted by these jobs, so it can't be @k8s-ci-robot. We could maybe use @k8s-github-robot, but I would prefer using a bot that has no privileges.

[spiffxp]

You'll want to rename more than the account:

• https://testgrid.k8s.io/sig-testing-fejtabot
• https://github.com/kubernetes/test-infra/blob/master/config/jobs/kubernetes/test-infra/fejta-bot-periodics.yaml

Things @fejta-bot does:

• /lifecycle stale | /lifecycle rotten | /close
• /retest
• /check-cla
• comment that API review may be needed
• create github issues

Naming is hard, WDYT of either of these?

• @k8s-triage-robot
• @k8s-commenter-robot

[spiffxp]

/milestone v1.16
/assign

[chris-short]

k8s-cuddle-bot
k8s-repo-sanity-bot
k8s-triage-bot

[justaugustus]

+1 to k8s-triage-[ro]bot!
Can probably dovetail with some of the stuff mentioned here: kubernetes/community#3456

[dims]

+1 to k8s-triage-bot

[InAnimaTe]

fajita-bot

[nikhita]

+1 to k8s-triage-robot. Since we have -robot suffixes for our other bot accounts, using -robot feels consistent.

To be clear, this would be creating a new account instead of renaming the original one, right? I'm assuming @fejta would like to keep @fejta-bot around for testing, etc... @fejta can you confirm? :)

[spiffxp]

/area github-management

To be clear, this would be creating a new account instead of renaming the original one, right? I'm assuming @fejta would like to keep @fejta-bot around for testing, etc... @fejta can you confirm? :)

I spoke offline with fejta last week and am confirming by proxy, as he's out this week

[spiffxp]

I've squatted on http://github.com/k8s-triage-robot, what needs to happen now is:

• an avatar
• making a token
• getting test-infra oncall to add the new secret to the cluster
• swapping the jobs to use the new token
• renaming the jobs / dashboards referenced above

note that fejta-bot is not a member of the org, doesn't have any privileges, doesn't need to have 2fa enabled; if we want that to change, I'd rather consider that out of scope here, and ask that those so motivated create a followup issue to

[nikhita]

swapping the jobs to use the new token
renaming the jobs / dashboards referenced above

Created #13683 to get the ball rolling

an avatar

Do we know who created the avatar for @k8s-github-robot? It's pretty dope!

[spiffxp]

Do we know who created the avatar for @k8s-github-robot? It's pretty dope!

👋 😄

[spiffxp]

/priority important-soon

[spiffxp]

/reopen
/assign
I'm willing to tackle this before v1.22 is released, or before v1.23 starts

[k8s-ci-robot]

@spiffxp: Reopened this issue.

In response to this:

/reopen
/assign
I'm willing to tackle this before v1.22 is released, or before v1.23 starts

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.

[spiffxp]

/remove-lifecycle rotten

[spiffxp]

Added the k8s-triage-robot token to k8s-infra-prow-build-trusted kubernetes/k8s.io#2389

Next step is to try a sample job to make sure it works, then we can migrate the fejta-bot jobs

[spiffxp]

I originally posted more comments in the wrong issue (#11402) so I'm copy-pasting them here

[spiffxp]

The test job is failing (whoops I must have missed the tab name): https://testgrid.k8s.io/sig-contribex-k8s-triage-robot#api-review-help

It seems like bad credentials? https://prow.k8s.io/view/gs/kubernetes-jenkins/logs/ci-k8s-triage-robot-testcomment/1418328613631037440

So I logged into k8s-triage-robot, and sure enough, no personal access tokens at all.

Created a new one, updated the secret, waiting for the next job run.

Originally posted by @spiffxp in #11402 (comment)

[spiffxp]

Failed run: encoutered 2 failures: [Failed to apply comment to kubernetes/test-infra#11402: the GitHub API request returns a 403 error: {"message":"Must have admin rights to Repository.","documentation_url":"https://docs.github.com/rest/reference/issues#create-an-issue-comment"} Failed to apply comment to kubernetes/org#2465: the GitHub API request returns a 403 error: {"message":"Must have admin rights to Repository.","documentation_url":"https://docs.github.com/rest/reference/issues#create-an-issue-comment"}]

Originally posted by @spiffxp in #11402 (comment)

[spiffxp]

Fixed the token, it needed scope:public_repo

The query also didn't include repo:kubernetes/test-infra, so it ended up commenting here: kubernetes/org#2465 (comment)

Originally posted by @spiffxp in #11402 (comment)

[spiffxp]

Orrrr apparently github's search returned one issue the first time, and two the second. But that's it. Should see no more spam at this point.

Opened #22997 to switch all the fejta-bot jobs over to k8s-triage-robot

Originally posted by @spiffxp in #11402 (comment)

[spiffxp]

https://github.com/issues?q=involves%3Ak8s-triage-robot+archived%3Afalse+sort%3Aupdated-desc+ - so the bot is definitely commenting now

There are still some "failing" jobs on the dashboard

• https://testgrid.k8s.io/sig-contribex-k8s-triage-robot#stale - this is due to attempting to comment on a locked issue, so exclude those by default: robots/commenter: add --include-locked flag #23029
• https://testgrid.k8s.io/sig-contribex-k8s-triage-robot#secping - hasn't run yet, so I've triggered a re-run
• https://testgrid.k8s.io/sig-contribex-k8s-triage-robot#issue-creator - ditto

[spiffxp]

Opened a PR to bump jobs to use the latest commenter:

• config/jobs: bump commenter jobs image #23032

I think because I tried re-running the other two jobs (since I can't re-trigger them), they tried and failed against the wrong cluster again, and their interval might have reset. In other words, may need to wait to sometime after 3pm PT to see if they go back to green

[spiffxp]

/milestone v.122
This has been completed

[k8s-ci-robot]

@spiffxp: The provided milestone is not valid for this repository. Milestones in this repository: [someday, v1.21, v1.22, v1.23, v1.24, v1.25]

Use /milestone clear to clear the milestone.

In response to this:

/milestone v.122
This has been completed

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.

[spiffxp]

/milestone v1.22
/close

[k8s-ci-robot]

@spiffxp: Closing this issue.

In response to this:

/milestone v1.22
/close

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.