We are running air-gapped, we disabled the default catalogsources as described in https://docs.openshift.com/container-platform/4.12/operators/admin/olm-restricted-networks.html#olm-restricted-networks-operatorhub_olm-restricted-networks after that our air-gapped catalogsource is placed. We do this many times.

But today during a rollout a subscription got stuck, with the following message:

```
- message: 'failed to populate resolver cache from source redhat-marketplace/openshift-marketplace:
      failed to list bundles: rpc error: code = Unavailable desc = connection error:
      desc = "transport: Error while dialing dial tcp: lookup redhat-marketplace.openshift-marketplace.svc
      on 172.30.0.10:53: no such host"'
```

I took some time to understand the message, but it was clear it wanted to use a default catalogsource that was still available.

first we tried delete the olm-operator pod:

```
oc -n openshift-operator-lifecycle-manager delete pod olm-operator-b86687bd8-hjzj4
```

We created the subscription again, but it failed with the same message.

Then we restarted all the pods:

```
oc -n openshift-operator-lifecycle-manager delete $(oc -n openshift-operator-lifecycle-manager get pods -o name)
```

Then the installation of the operator continued without recreating the subscription, the message was gone.

4.12

Not easily reproduced

1.
2.
3.