Description of problem: Should show build failure reason for docker build when builder image is wrong. When creating docker build with Wrong Post Commit/Wrong Source, detail failure reason is shown in 'oc get builds' output as expected, this should also applies to docker build when wrong builder image is given Version-Release number of selected component (if applicable): oc v3.5.0.7+390ef18 kubernetes v1.5.2+43a9be4 How reproducible: Always Steps to Reproduce: 1. Create docker build with Wrong Builder Image $ wget https://raw.githubusercontent.com/openshift-qe/v3-testfiles/master/build/ruby22rhel7-template-docker.json $ vi ruby22rhel7-template-docker.json ..... "strategy": { "type": "Docker", "dockerStrategy": { "from": { "kind": "DockerImage", "name": "docker.io/openshift/rubyyyy-20-centos7:latest" }, ........ $ oc new-app -f ruby22rhel7-template-docker.json 2.Wait for #1 build finish, check build status via CLI $ oc get builds Actual results: 2. Only show "Failed" in Status, detail failure reason was not shown $ oc get builds NAME TYPE FROM STATUS STARTED DURATION ruby22-sample-build-1 Docker Git@e79d887 Failed 3 minutes ago 11s Expected results: 2. Should also show failure reason in output like $ oc get builds NAME TYPE FROM STATUS STARTED DURATION ruby22-sample-build-1 Docker Git@e79d887 Failed (PullBuilderImageFailed) 3 minutes ago 11s Additional info:
i hate to criticize, but the description of this issue makes very little sense. Anyway, there's an actual test case that covers this exact problem, see: https://github.com/openshift/origin/blob/master/test/extended/builds/failure_status.go#L92-L107 The problem is a race condition in updating the build object and not a missing implementation. Will update once i have a fix.
Victor what's the status on this bug? Any theories on where/why the race is occurring?
i believe somewhere in the controller, we update the build object and overwrite whatever values we pushed to the API from the build pod. see: https://github.com/openshift/origin/blob/master/pkg/build/controller/controller.go#L134 Usually the fields are empty for actions involving the build process. s2i and docker strategy. This happens very rarely and usually happens more often as a test flake than by running the build yourself. The controller update, should either be updated to use something similar to this https://github.com/openshift/origin/blob/master/pkg/build/builder/sti.go#L121 Or the only updates that are done to the build object should be done only from the controller itself. But, because the build pod is isolated, the controller doesn't know what happens to it unless it receives information from the API that is updated by the build pod.
the build status reason and build status message get cleared by: https://github.com/openshift/origin/blob/master/pkg/build/controller/controller.go#L229-L230 but i can't think my way through a path where we'd have a build enter the failed state, and then still go through nextBuildPhase (which is normally only called for build objects that are in the New state. We wouldn't even have a pod yet if the build were in the New state. And if the build were being updated elsewhere, then the controller's update would fail due to a version conflict, so it wouldn't be able to overwrite the reason/message fields)
Submitted PR https://github.com/openshift/origin/pull/12873
https://github.com/openshift/origin/pull/12873
This has been merged into ocp and is in OCP v3.5.0.32 or newer.
Checked against oc v3.5.0.32-1+4f84c83 with the same step in "Steps to Reproduce", after build finishes, 'oc get builds' returns Build Failure reason $ oc get builds NAME TYPE FROM STATUS STARTED DURATION ruby22-sample-build-1 Docker Git@e79d887 Failed (PullBuilderImageFailed) About a minute ago Move to VERIFIED
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHBA-2017:0884