On hot-plug tests, many times the action fails, and debugging shows that the migration part fails, because the source virt-launcher pod remains in Running state.

This actually looks like a migration issue, but we observe it in hot-plug tests. I would guess that this is related to the fact the starting 4.20, migration can only be done by cluster admin (and this is how we test it in tier-2), while hot-plug is done by namespace admin (i.e. non-privileged user), and the required migration part of it is executed in the background, by granting the required privileges in the background.

CNV v4.20.0.rhel9-146

Not always

1.
Perform interface hot-plug in a VM.
For QE - this can be done by running one of ther tier-2 hot-plug tests.

In some results - the action is not completed because the migration gets stuck

Migration succeeds and NIC hot-plug completed.

We see that when this happens, we have 2 pods in running state - the new target virt-luancher and the old source virt-launcher.
In the virt-controller log, we see a message explicitly saying that the target pod cannot be scheduled because there are multiple pods in running state, e.g.:

2025-09-06T22:16:57.636421642Z {"component":"virt-controller","kind":"","level":"info","msg":"Waiting to schedule target pod for migration because there are already multiple pods running for vmi l2-bridge-test-bridge-nic-hot-plug/hot-plug-test-vm-1757196614-051137","name":"kubevirt-workload-update-gj7mm","namespace":"l2-bridge-test-bridge-nic-hot-plug","pos":"migration.go:1192","timestamp":"2025-09-06T22:16:57.636345Z","uid":"c908e54d-bcf9-43d0-be38-f20c9a7b01e3"}

* I submitted this as a network bug, but there's a good chance that this is a virt issue, so my apologies in advance.