[RFE] can we trigger koji-build (or other packit-service ) ?

[sergiomb2]

Description

hello, last paragraph of https://packit.dev/docs/fedora-releases-guide/non-divergent-dist-git-branches says "You can work around this bug by manually retriggering the Koji build by commenting on the downstream merged pull request with /packit koji-build " .

packit build in-koji doesn't do the bodhi update and packit create-update --dist-git-branch doesn't do the same of packit-service, for example doesn't close the bugs on bugzilla.

Best regards,

Benefit

more uniform automation

Importance

No response

What is the impacted category (job)?

Fedora release automation

Workaround

• [ ] There is an existing workaround that can be used until this feature is implemented.

Participation

• [ ] I am willing to submit a pull request for this issue. (Packit team is happy to help!)

[lachmanfrantisek]

Hi @sergiomb2 ,

I hope you know this already, but Packit CLI is not a client for the Packit Service but a way to run this with your identity locally. (It's on the same level as service.) Do you think we should make this somewhere more explicit?

Regarding the commands:

• You can configure the Packit service to create Bodhi updates for builds triggered by other packages (i.e. you): https://packit.dev/docs/configuration/downstream/bodhi_update#optional-parameters
• Same, you can configure the Pakit service to trigger builds for dist-git changes caused by other users: https://packit.dev/docs/configuration/downstream/koji_build#optional-parameters (allowed_pr_authors and allowed_committers).

So, can these options help in your use case?

• If yes, we can mention it in the documentation part you've mentioned.
• If not, can you please suggest what would help you?

[sergiomb2]

" Fast Forwarding multiple commits does not automatically trigger a Koji build (Yet)

Unfortunately, there is a bug that prevents Packit from triggering the Koji build when more than one commit has been forwarded in a branch.

You can work around this bug by manually retriggering the Koji build by commenting on the downstream merged pull request with /packit koji-build. "

Maybe "with packit command line" is confusing because is one suggestion to solve the RFE.

in resume when packit-service fail to run , I'd like have a way to trigger it again .

[lachmanfrantisek]

@sergiomb2 , if I understand you properly, you want to mention also a way how to overcome this problem via CLI?

We can easily add that. I am just not sure if this isn't too complicated and confusing compared to commenting on a pull-request.

• You need to install packit package locally and set up authentication.
• Once you run a job from CLI, the next steps won't be run by Packit Service (by default):
  • This is to not collide with manual interventions (e.g. from other packagers or even proven packagers).
  • You need to update the config with the two options I've mentioned above to let Packit now that it should run also for actions triggered by you.
• The locally-run jobs are not visible in the dashboard which might be confusing.

So, I can see two options:

1. Don't promote CLI in this case because it's much easier to comment on a dist-git pull request.
2. Mention a way to trigger jobs via CLI, but mention all the steps needed to make it work.

Witch one do you prefer? I slightly prefer the first one because it's easier and does not have any downsides.

[lachmanfrantisek]

From the today's team discussion:

• We would like to create a new FAQ to describe using Packit CLI together with Packit service and we can link it from places like this.

[majamassarini]

From the today's team discussion:

• We would like to create a new FAQ to describe using Packit CLI together with Packit service and we can link it from places like this.

Also related with #1657

[sergiomb2]

The locally-run jobs are not visible in the dashboard which might be confusing.

is more than confusing, doesn't do the same , the owners of builds are different, the user which create update is me and is not packit , I have do it in two steps [1] at least and "packit create-update" doesn't close bugs as packit service do .

On other hand one pull requests just to request packit service , is not a good option for me , history on git and of PRs will not be elegant

[1] packit build in-koji --dist-git-branch f40 packit create-update --dist-git-branch f40

[sergiomb2]

Also related with #1657

"We should give maintainers the ability to overcome a failing automation."

yes, definitively is what I'm asking in this issue , overcome a failing automation without hacking and with the same automation if it isn't much difficult.

we have packit build and packit create-update but isn't the same in my point of view , maybe was meant to be , I don't know