Description of problem:

We found an issue with Hub Recovery. Details are in this Bugzilla link.
It seems that the problem was introduced in a recent pull request. You can find the PR here: GitHub link. Specifically, the issue lies in this code block: GitHub code link.

Let me explain the scenario and the steps that trigger this problem. We are running tests in the ODF-DR test environment. Our setup consists of 4 clusters: 2 active/passive hubs and 2 managed clusters. We have different workloads running in both managed clusters.
During testing, we need to shut down the entire zone, which includes shutting down the active hub and Cluster 1. After that, we recover the hub, and the passive one becomes the active hub. However, the ManifestWorks from the failed hub are not restored. They will be regenerated.

As a result, the line mentioned earlier (line 94) causes the workload to be evicted from Cluster 2.
I need confirmation on whether my assumption is correct or not. If it is, I will reassign the bug to one of the ACM team members. Just wanted to let you know that this bug is marked as urgent.

Version-Release number of selected component (if applicable):

ACM 2.7.4

How reproducible:

Always

Steps to Reproduce:

detailed here