-
Bug
-
Resolution: Done
-
Normal
-
Pipelines 1.6, Pipelines 1.7, Pipelines 1.8.0
-
None
-
None
-
2
-
False
-
None
-
False
-
Pipelines Sprint 231
Expected behavior
When votes-ui or votes-api is build second time, new image is used for deployment, UI shows new container ID
Actual behavior
Pods are never recreated with a new image, logs show "deployment.apps/pipelines-vote-ui configured" but in fact it only change image reference from image-registry.openshift-image-registry.svc:5000/mynamespace/pipelines-vote-ui to the same value
Steps to reproduce
Follow https://docs.openshift.com/container-platform/4.10/cicd/pipelines/creating-applications-with-cicd-pipelines.html or run scripts from https://github.com/openshift/pipelines-tutorial
./demo.sh setup ./demo.sh run APP_URL=$(oc get route pipelines-vote-ui -n $NAMESPACE --template='http://{{.spec.host}}') lynx $APP_URL --dump # notice container ID route=$(oc get route -l eventlistener=vote-app -n $NAMESPACE -o name) url=$(oc get $route --template='http://{{.spec.host}}' -n $NAMESPACE) curl -X POST \ ${url} \ -H 'Content-Type: application/json' \ -H 'X-GitHub-Event: push' \ -H 'X-Hub-Signature: sha1=7d52b3ec735083bcaf14892bcf70ab403424b56e' \ -d '{"head_commit": { "id": "master"},"repository":{"url": "https://github.com/openshift/pipelines-vote-ui.git", "name": "pipelines-vote-ui"}}' ./demo.sh logs lynx $APP_URL --dump # notice container ID
Possible fix
Simplest fix would be to propagate IMAGE_DIGEST from buildah task to the update-deployment task
Alternative approach would be to use deploymentconfig instead of deployment and do "oc rollout" instead of manually applying patch.
