CI comment ends up in wrong repo

[addaleax]

I kicked off CI for nodejs/quic#157 (https://ci.nodejs.org/job/node-test-pull-request/25909/) but the bot commented the CI link on nodejs/node#157.

[phillipj]

Looking at the related route that handles request from ci.nodejs.org regarding started jobs, it looks to me as if cURL command executed by ci.nodejs.org specifies node as the repo. Probably been hard coded since that job has historically only been used by that repo, right?

I currently don't have enough ci.nodejs.org privileges to see the bash script executed upon job start, but it should be pretty obvious at first glance if I remember correctly. The URL used in that script should be using $TARGET_REPO_NAME job parameter instead of a hard coded string of node.

Does that make sense? Maybe you've got privileges to give it a shot?

[addaleax]

I don’t think I have any special CI privileges.

[phillipj]

Hopefully someone in @nodejs/build with enough CI privileges can have a look?

[richardlau]

I don't have privileges either, but what's probably going on is that node-test-pull-request isn't passing on the TARGET_REPO_NAME parameter to the post-build-status-update job (which has a REPO parameter that defaults to node).

There are three more instances in the Post-build Actions section. I did a quick sample and it doesn't look like any of the jobs started by node-test-pull-request (e.g. the linter) that call post-build-status-update pass on a REPO parameter so they all default to node too.

[Trott]

@nodejs/jenkins-admins

[richardlau]

The bot is being triggered in the CI by the following pipeline: https://github.com/nodejs/build/blob/master/jenkins/pipelines/post-build-status-update.jenkinsfile

It looks like my earlier hypothesis (https://github.com/nodejs/github-bot/issues/245#issuecomment-539711522) is correct -- we need to set the REPO parameter when we trigger https://ci.nodejs.org/job/post-build-status-update/ otherwise it defaults to node. That's quite a lot of places that need updating.

I'll transfer this issue over to build to look at after the security releases go out.

[richardlau]

I'll transfer this issue over to build to look at after the security releases go out.

Except I can't transfer (probably because I don't have the appropriate permissions for this repository).

[phillipj]

Ahh thanks for referencing that .jenkinsfile 👍 I haven't noticed those in the build repo before..

If you'd rather not update lots of Jenkins related stuff as you mention, let me know if you think this is better solved in github-bot somehow and want some help with that.

[richardlau]

I don't have privileges either,

I didn't at the time, I do now 😀. This does require updating all the node-test-commit-* + the linter job, preferably all at the same time. I'll see if I can carve out some time later next week.

[mmarchini]

Wouldn't updating the node-test-pull-request be enough at least to fix the comment? This one seems more urgent (I'm reluctant of testing auto start CI via labels because it will spam a bunch of old issues), updating the others could be done at a separate time.

[richardlau]

Wouldn't updating the node-test-pull-request be enough at least to fix the comment? This one seems more urgent (I'm reluctant of testing auto start CI via labels because it will spam a bunch of old issues), updating the others could be done at a separate time.

Updated just node-test-pull-request via https://ci.nodejs.org/job/node-test-pull-request/jobConfigHistory/showDiffFiles?timestamp1=2020-06-01_13-53-19&timestamp2=2020-06-27_15-37-02