Description of problem: Version-Release number of selected component (if applicable): oc cluster up pulling origin images How reproducible: Steps to Reproduce: 1.oc cluster up -- Checking OpenShift client ... OK -- Checking Docker client ... OK -- Checking Docker version ... OK -- Checking for existing OpenShift container ... OK -- Checking for openshift/origin:v3.5.0.16 image ... Pulling image openshift/origin:v3.5.0.16 FAIL Error: error pulling Docker image openshift/origin:v3.5.0.16 Caused By: Error: manifest unknown: manifest unknown 2. 3. Actual results: Expected results: pulling the images from registry.access.redhat.com Additional info:
Steve, sending this one your way since you're addressing this. In the 3.4 branch of OSE, the registry for OSE is correct [1]. But I imagine it is reverted in 3.5 because we're starting with origin again? [2] [1] https://github.com/openshift/ose/blob/enterprise-3.4/pkg/bootstrap/docker/up.go#L134 [2] https://github.com/openshift/ose/blob/master/pkg/bootstrap/docker/up.go#L228
Once we pick up the ldflag change [1] in the OSE rebase we will need to provide the correct envars [2] during the build and we should be good here. Not sure why we see a regression. I was not under the impression that we had begun the rebases or rewritten any git history in OSE master yet. [1] https://github.com/openshift/origin/pull/12758/commits/5237fda6a1f1fc0ca770ac04071e07ab30b8de2a [2] OS_IMAGE_PREFIX=openshift3/ose OS_BUILD_LDFLAGS_DEFAULT_IMAGE_STREAMS=rhel7
Sorry, I forgot to reply to this. The newer ose rpms (atomic-openshift) are built with these options. It should be working now I believe on at least atomic-openshift-3.5.0.31 or newer.
Verified with: openshift v3.5.0.32-1+4f84c83 kubernetes v1.5.2+43a9be4 etcd 3.1.0
Since this bug never reached customers, I am closing it.