Reusing muc_nickname when resetting a jitsi-videobridge instance

The jitsi-videobridge documentation states that

The muc_nickname must be unique across all jitsi-videobridge instances

However, from this it is not clear if

  1. the muc_nickname must be unique across all jitsi-videobridge instances that are active at the same time, or
  2. the muc_nickname must be unique across all jitsi-videobridge instances ever.

My context

I run Jitsi Meet in Kubernetes and want to have multiple video bridges. My idea was to use a StatefulSet consisting of (e.g.) 3 Pods with hostnames jvb-0, jvb-1, jvb-2. This way, the Pods can derive their MUC_NICKNAME and their USERNAME from their hostname. The former seems to be the default with docker-jitsi-meet, and the latter can be achieved using:

          env:
            - name: JVB_AUTH_USER
              valueFrom:
                fieldRef:
                  fieldPath: metadata.name

The problem

Whenever I delete a jitsi-videobridge Pod, e.g. jvb-0, Kubernetes creates a completely new Pod with the same name but with a new IP. No state is shared between the old and new jvb-0, so the new one is effectively a new jitsi-videobridge instance. Now, even after the new jvb-0 is initialized, the jicofo health check keeps failing:

[...]
Jicofo 2021-02-07 03:21:18.297 WARNING: [145] org.jitsi.jicofo.bridge.JvbDoctor.log() Health check timed out for: jvbbrewery@internal-muc.meet.jitsi/jvb-0
Jicofo 2021-02-07 03:21:33.298 WARNING: [145] org.jitsi.jicofo.bridge.JvbDoctor.log() jvbbrewery@internal-muc.meet.jitsi/jvb-0 health-check timed out, but will give it another try after: 5000
Jicofo 2021-02-07 03:21:53.298 WARNING: [145] org.jitsi.jicofo.bridge.JvbDoctor.log() Health check timed out for: jvbbrewery@internal-muc.meet.jitsi/jvb-0
Jicofo 2021-02-07 03:22:08.299 WARNING: [145] org.jitsi.jicofo.bridge.JvbDoctor.log() jvbbrewery@internal-muc.meet.jitsi/jvb-0 health-check timed out, but will give it another try after: 5000
[...]

This problem happens not only for jvb-0, but for any jitsi-videobridge instance which I recreate. It persists until I restart jicofo. After restarting jicofo, the health-checks for all jitsi-videobridge instances succeed.

I don’t know the reason, yet. But I thought maybe it’s because I’m reusing the hostname and thus reusing the muc_nickname. Do you know whether my problem is expected and I shouldn’t be reusing the muc_nickname?