Description of problem: Pod keeps in ImagePullBackOff status although image exists in nodes: [wzheng@localhost test]$ oc get pods NAME READY STATUS RESTARTS AGE frontend-moe8i 0/1 ImagePullBackOff 0 3m frontend-yh3tm 0/1 ImagePullBackOff 0 3m frontend-zzut6 0/1 ImagePullBackOff 0 3m redis-master-ey88p 1/1 Running 0 4m redis-slave-6n0s8 0/1 ImagePullBackOff 0 3m redis-slave-xwt3l 0/1 ImagePullBackOff 0 3m Version-Release number of selected component (if applicable): openshift v3.0.2.903-114-g2849767 kubernetes v1.2.0-alpha.1-1107-g4c8e6f4 etcd 2.1.2 How reproducible: always Steps to Reproduce: 0. Change scc Before testing, please add "- system:authenticated" to scc/privileged group and remove it from scc/restricted. $oc edit scc/restricted $oc edit scc/privileged groups: - system:cluster-admins - system:nodes - system:authenticated 1. Create a project 2. Create with guestbook example $oc create -f https://raw.githubusercontent.com/kubernetes/kubernetes/master/examples/guestbook/redis-master-controller.yaml $oc create -f https://raw.githubusercontent.com/kubernetes/kubernetes/master/examples/guestbook/redis-master-service.yaml $oc create -f https://raw.githubusercontent.com/kubernetes/kubernetes/master/examples/guestbook/redis-slave-controller.yaml $oc create -f https://raw.githubusercontent.com/kubernetes/kubernetes/master/examples/guestbook/redis-slave-service.yaml $oc create -f https://raw.githubusercontent.com/kubernetes/kubernetes/master/examples/guestbook/frontend-controller.yaml $oc create -f https://raw.githubusercontent.com/kubernetes/kubernetes/master/examples/guestbook/frontend-service.yaml 3. Check the pods are running $oc get pods Actual results: pod allways in ImagePullBackOff status Expected results: [fedora@ip-172-18-11-225 ~]$ oc get pods NAME READY STATUS RESTARTS AGE frontend-bgs6k 1/1 Running 0 4m frontend-vy4ju 1/1 Running 0 4m frontend-w5236 1/1 Running 0 4m redis-master-fhvoh 1/1 Running 0 4m redis-slave-d6yy0 1/1 Running 0 4m redis-slave-zpjbp 1/1 Running 0 4m Additional info:
It looks like the images it is trying to pull are both gcr.io images. On my test they took longer than three minutes. Can you verify that gcr.io/google_samples/gb-frontend:v3 is actually on the node in 'docker images'?
Thanks, Paul! I just pulled redis, didn't pull gcr.io/google_samples/gb-frontend:v3, pod is running now after images are ready.
This fix is available in OpenShift Enterprise 3.1.