[POC] Many metadata-proxy pods stuck in CrashLoopBackOff in scale out kubemark from one TP, looks good from another TP

[Sindica]

What happened:
In 2x2 scale out kubemark case:

$ ./cluster/kubectl.sh --kubeconfig test/kubemark/resources/kubeconfig.kubemark.tp-2 get pods -AT | grep -v Running
TENANT   NAMESPACE     NAME                                                         HASHKEY               READY   STATUS             RESTARTS   AGE
system   kube-system   kubernetes-dashboard-848965699-ktpbt                         1860350797080663147   0/1     CrashLoopBackOff   9          44m

$ ./cluster/kubectl.sh --kubeconfig test/kubemark/resources/kubeconfig.kubemark.tp-1 get pods -AT | grep -v Running
TENANT   NAMESPACE     NAME                                                         HASHKEY               READY   STATUS             RESTARTS   AGE
system   kube-system   kubernetes-dashboard-848965699-ww7kj                         3223766609900507535   0/1     CrashLoopBackOff   9          47m
system   kube-system   metadata-proxy-v0.1-5rm2s                                    2833962277662696395   1/2     CrashLoopBackOff   11         37m
system   kube-system   metadata-proxy-v0.1-mbbbf                                    5739051913249442897   1/2     CrashLoopBackOff   11         37m

What you expected to happen:
All pods stable in running status