Environment status is set to inactive

[gergokee]

Describe the bug
If you have two jobs in same workflow, which uses same environment, even though the second job is running after the first one, the first jobs is rewriting the second job, causing the environment to be shown as "inactive".

To Reproduce
Steps to reproduce the behavior:
1.) create a workflow with two jobs with same environment and url which are running after eachother.
2.) run the workflow and see that the first job deactivates the second one, even though it should be the other way around.

Expected behavior
The job which was ran later should have the final word, and thus the environment would show up as active with the url.

[yaananth]

We always ensure there is ONE active deployment per URL.

Meaning a new deployment with URL github.com will make all previous deployments with same URL "inactive". However, if we had a deployment with URL github-analytics.com, we will leave it alone.

We are working on clarifying these in docs.

Thanks for the issue!

[daveykane]

there is an option for deployments auto_inactive which can disable this behaviour. Can it be added to a jobs Environment properties along with the url?

[daveykane]

@yaananth can we revisit this one please? currently the behaviour is not as you describe.. There seems to be ONE active deployment per Environment Name rather than Environment URL which would be better?

[michaelst]

I am seeing the same thing as @daveykane

[CicerosPatience]

Yeah same here, we use a matrix to deploy to multiple k8s clusters in the same environment and it always reads Inactive a few minutes after the deploy job finishes successfully.

[error418]

This issue may be relate to #2120

[zburgermeiszter]

@yaananth

We always ensure there is ONE active deployment per URL.

Based on this, I ran some experiments.
If multiple apps are deployed against the same environment, as long as the URL is unique for each app in a (mono)repo both deployments will show with a green tick.

A minor issue is only one of them will have the Active pill next to it, and only one of them is shown in the latest deployments box, but at least none of them have a grey tick.

[pegiadise]

Worked around it by adding a final job dependent on all the jobs and keeping the environment property only on that.

  deployment-status:
    needs: [deploy-job-1, deploy-job-2]
    runs-on: ubuntu-latest
    environment: production
    steps:
      - name: Deployment Completed
        run: echo "Production deployment completed successfully"