Pods in the openshift-markeplace
namespace are in ImagePullBackOff
Virtual Private Cloud
In VPC clusters with worker nodes in a VPC subnet that doesn't have a public gateway, the pods in the openshift-markeplace
namespace show a state of ImagePullBackOff
and cannot reach Running
state.
The images for these containers are hosted on a public image registry but the workers do not have access to the public internet, so the workers cannot to pull the image. The cluster still works, but these pods and the openshift marketplace does
not work. If you run the kubectl get pods
command, you get output such as:
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
certified-operators-48dsc 0/1 ImagePullBackOff 0 4h6m 172.17.70.94 172.24.0.6 <none> <none>
certified-operators-s7qtb 0/1 ImagePullBackOff 0 4h17m 172.17.70.77 172.24.0.6 <none> <none>
community-operators-h7hl9 0/1 ImagePullBackOff 0 4h17m 172.17.70.78 172.24.0.6 <none> <none>
community-operators-hc76f 0/1 ImagePullBackOff 0 4h6m 172.17.70.96 172.24.0.6 <none> <none>
marketplace-operator-66c666bf74-gnltf 1/1 Running 0 4h43m 172.17.126.216 172.24.0.8 <none> <none>
redhat-marketplace-tfwwp 0/1 ImagePullBackOff 0 82s 172.17.126.226 172.24.0.8 <none> <none>
redhat-marketplace-zkvfc 0/1 ImagePullBackOff 0 4h6m 172.17.70.95 172.24.0.6 <none> <none>
redhat-operators-pthhf 0/1 ImagePullBackOff 0 67s 172.17.126.227 172.24.0.8 <none> <none>
redhat-operators-rjvsp 0/1 ImagePullBackOff 0 4h6m 172.17.70.93 172.24.0.6 <none> <none>
You can fix this issue by disabling and mirroring these public images. For more information, see Disabling OperatorHub and mirroring catalog source images to icr.io
.