Open mwes opened 5 years ago
This would be useful to resolve going forward - there's been a lot of good work with the adapter recently, but we want to make sure those code changes make it into the base image that everything else in the program uses: reactors, applications, etc. Given that the Jenkins build is working now, it's likely a minor lift to get this to trigger the next desired step of rebuilding the base images. I believe http://jenkins.sd2e.org/job/Admin/job/base-images-build/ is likely close to what we need. I'll bring this up with Matt Vaughn as well.
@zachzundel Would you be willing to take this over since it should be an easy clone of what you're doing for rolling deployment of SynBioHub?
I could take this on. It is not quite the same as SBH deployment, but it is easier. I can also help you set up a protected master branch so that all changes have to go through PRs, which you can require to be reviewed or pass CI before they can merge.
I've configured the synbiohub_adapter Jenkins to trigger base-images-build when it succeeds. @mwes / @mwvaughn is this the right path?
👍 to "set up a protected master branch so that all changes have to go through PRs, which you can require to be reviewed or pass CI before they can merge." We've had some build breakages here, and don't want to build images that have SBHA code that will fail when run. Suggest that be done separately from what we're discussing here, however. Thanks for jumping on this @zachzundel.
@mwvaughn should we just change the base images job to run make languages
(etc.) instead of make-languages-build
and other variants ? That seems like it would do it.
@mwvaughn @zachzundel I'd anticipated that need with #58, though I hadn't assigned it to anybody yet. It's definitely time to professionalize our development of this tool. If #58 gets resolved as a byproduct of this, so much the better.
I think the configuration change I made to synbiohub_adapter
Jenkins should be enough to close this issue. Moving discussion of protected branches to #58 .
Actually, now that I think about it, we probably need to gate the trigger for the other pipeline to only changes to the master branch of synbiohub_adapter
.
Thanks - FYI that job looks like it builds the images, but I don't believe it currently pushes them. Once we can confirm the images are being built and pushed, we can close this.
What is the status of this issue? There have been no updates since Feb 1.
@tcmitchell let me check with Matt on the image building; this should likely only require a minor change to the Jenkins job. While related, it looks like this issue got looped into some CI / build restriction issues. I would separate these out as their own issue / discussion. e.g. if Jenkins is configured to build on any push to master, the restrictions around those pushes are not directly relevant to this issue and image building (whether it's from a passing PR (preferred) or a direct push to master (definitely not preferred) the images should get rebuilt
Repository management items have been moved to #81
Great, thanks @tcmitchell
@mwes @3ach @jakebeal What is the status of this issue? I see Jenkins base-images-build appears to get rebuilt when synbiohub_adapter gets built. Does that match the goal of this task, and thus this task is complete?
The SD2E python base-images natively integrate the SBHA. In particular
sd2e/reactors:python3-edge
When the SBHA adapter is updated on its master branch (ideally through a passing pull request run through CI (
http://jenkins.sd2e.org/job/synbiohub_adapter/
) we should trigger a rebuild and push of the base-images that make use of the SBHA.This way the latest passing version of the SBHA is automatically integrated into the reactor base images. Otherwise we have to remember to manually do this, and the integrated SBHA will lag behind.
The build and push command from base-images should be:
make reactors CHANNEL=edge LANGUAGE=python3