Normal Description of problem:
When the image registry is flaky, commands like {{oc image mirror --keep-manifest-list --continue-on-error --filename=....}} fail. It looks like this could benefit from retries. The errors are happening in the Plan phase.
Version-Release number of selected component (if applicable):
How reproducible:
Steps to Reproduce:
1. wait for quay to be flaky
2. run the command with a 200 image mappings in the file list
3. command fails, without much retrying
Actual results:
Expected results:
Additional info:
There are many pinpoints even in planning phase that try to access registry. This https://github.com/openshift/oc/blob/278081ea521e465e77b09d8c91fc6ef7c0654247/pkg/cli/image/mirror/mirror.go#L490 retrieves something or this https://github.com/openshift/oc/blob/278081ea521e465e77b09d8c91fc6ef7c0654247/pkg/cli/image/mirror/mirror.go#L527 retrieves manifests and so on. So that adding a retry mechanism in each execution would not solve the issue permanently (if there is a one that is so clear can be fixed by retries?).
Adding a retry mechanism in planning phase overall would bring about some side effects. I think, it is better to consider this as a new feature rather than a bug. In order to be prioritized or evaluated by PM and managerial side, I'd recommend opening an RFE.
If you don't mind, I'll close this as not a bug.