Description of problem:

RBAC backend edit operations inconsistent for high availability deployment. You can update, delete roles, permissions only if client is doing that for the same pod. If you will try to create update or edit roles, which was created from another replica pod it will fail with uncler for customer error.

Prerequisites (if any, like setup, operators/versions):

Steps to Reproduce

The steps to reproduce the issue are not entirely clear, but here is a general outline:

1. Deploy RHDH with 2 replicas and enable the router annotation option:

haproxy.router.openshift.io/balance: roundrobin

2. Create a test role in the RBAC UI.

3. Attempt to edit the test role multiple times using a different browser or incognito mode.
This ensures the client connection is not cached. Cached connections on the client side may always point to the first pod, which is why a separate browser session is required.
At least one of the new client connection sessions will use the second pod for the edit operation, and this operation will fail.

Actual results: Some edit, delete operations fail

Expected results: You should be able to edit or delete any previously created role from any replica pod.

Reproducibility (Always/Intermittent/Only Once):

Build Details:

Additional info (Such as Logs, Screenshots, etc):