1. Proposed title of this feature request
—> This RFE is to enhance the internal imageRegistry in RHOCP4

2. What is the nature and description of the request?
—> Failing imageStreams should be reconciled quicker than in 15minute intervals and that frequency should only be used on failed streams.

3. Why does the customer need this? (List the business requirements here)
—> To have the ImagesStream being automatically reconciled if there is an error - like many other objects in OpenShift.
Pods for example get reconciled if there is an error during pulling the requested image. We would like to get an equivalent processing we dealing with imagestreams.

4. List any affected packages or components.
OpenShift InternalImageRegistry

5. Notes
The RFE is to make the InternalImageRegistry more kubernetes self-healing by having an additional queue watching failed error streams more frequently in large-scale deployments.

The issue leading to the RFE was coming from deploying 16k deployments with ~2k imageStreams which hammered the backend to return 504 errors.
Even though the quick mitigation was to increase the backend resources, it was still necessary to manually fix the imageStreams that received the error and a self-healing mechanism would be highly appreciated.

Please consider the error handling queue/retry also for streams with `imageStream.spec.tags.importPolicy.scheduled: false` as mentioned the amount of `re-occuring` requests to the backend in the default iteration(15m, scheduled=true) is not in focus of using the internal Imageregistry.