-
Bug
-
Resolution: Unresolved
-
Normal
-
None
-
4.18.z
-
Quality / Stability / Reliability
-
False
-
-
None
-
None
-
None
-
None
-
None
-
None
-
None
-
None
-
None
-
None
-
None
-
None
-
None
-
None
Description of problem:
In case of agent-based OpenShift cluster in a disconnected environment, using `oc adm node-image create` approach to generate discovery ISO for node addition, the oc client always tries to contact quay first instead of using the mirror/internal registry as reference because of the nature that oc uses ICSPOnErrorStrategy. For environments with strict firewall restrictions where the outside connectivity is completely forbidden, due to this policy, even though the node-joiner pod runs and takes all the reference from ICSP/IDMS of the cluster, a connection failure to quay.io just hangs the process instead of falling back to mirror registry.
Version-Release number of selected component (if applicable):
4.18.23
How reproducible:
OpenShift agent-based cluster in a disconnected environment
Actual results:
The node-joiner process hangs and ISO couldn't be generated
Expected results:
The oc client should be able to create the ISO with all the references from the ICSP/IDMS
Additional info:
- is blocked by
-
OCPSTRAT-2757 readOnlyRootFIleSystem:true for the ephemeral container creating using oc admin node-image command
-
- New
-