[It] should run a deployment to completion and then scale to zero [Conformance]

[bparees]

deploymentconfigs
/data/src/github.com/openshift/origin/test/extended/deployments/deployments.go:213
  with test deployments
  /data/src/github.com/openshift/origin/test/extended/deployments/deployments.go:189
    should run a deployment to completion and then scale to zero [Conformance] [It]
    /data/src/github.com/openshift/origin/test/extended/deployments/deployments.go:188

    Expected
        <string>: --> pre: Running hook pod ...
        test pre hook executed
        --> pre: Success
        --> Scaling up deployment-test-2 from 0 to 1, scaling down deployment-test-1 from 0 to 0 (keep 1 pods available, don't exceed 2 pods)
            Scaling deployment-test-2 up to 1
    to contain substring
        <string>: --> Success

    /data/src/github.com/openshift/origin/test/extended/deployments/deployments.go:183
------------------------------
[k8s.io] Pods 
  should *not* be restarted with a docker exec "cat /tmp/health" liveness probe [Conformance]
  /data/src/github.com/openshift/origin/Godeps/_workspace/src/k8s.io/kubernetes/test/e2e/pods.go:713

as seen in:
https://ci.openshift.redhat.com/jenkins/job/test_pull_requests_origin_conformance/1704

[smarterclayton]

The pod for the deployment comes up, but looks like it never becomes ready. It is terminated after 200s when the deployment fails.

[mfojtik]

I saw this flake under the rebase branch.

[stevekuznetsov]

Changed title for searchability

[bparees]

https://ci.openshift.redhat.com/jenkins/job/test_pull_requests_origin_conformance/3373/consoleFull

[bparees]

@smarterclayton we're at a month on this... are we going to fix this test or disable it?

[smarterclayton]

Fix it, because it's indicating we're broken.

@Kargakis the only way to get this is if we changed the behavior of test deployments.

• Failure [18.357 seconds]
deploymentconfigs
/data/src/github.com/openshift/origin/test/extended/deployments/deployments.go:515
  with test deployments
  /data/src/github.com/openshift/origin/test/extended/deployments/deployments.go:254
    should run a deployment to completion and then scale to zero [Conformance] [It]
    /data/src/github.com/openshift/origin/test/extended/deployments/deployments.go:253

    Expected error:
        <*errors.errorString | 0xc821423b30>: {
            s: "deployment is complete but doesn't have expected spec replicas: 2 0",
        }
        deployment is complete but doesn't have expected spec replicas: 2 0
    not to have occurred

    /data/src/github.com/openshift/origin/test/extended/deployments/deployments.go:213

So at some point since june 7 we broke test deployments.

[smarterclayton]

I can't see where it changed, but the easy fix is to move the check for the invariant (scale should be zero) into the end of the deployments controller loop, so we guarantee that before we ever set complete or failed scale is zero, we'd be ok.

[smarterclayton]

Pulling up a fix.

[smarterclayton]

It's also possible that we are triggering a failure condition (due to the last rebase slowing down certain operations) that we didn't previously. Either way, this guarantees we never mark an RC complete or failed without setting scale to zero.