ClusterIP corruption?

I am seeing an intermittent issue running some of our acceptance tests on a Kind cluster with the istio service mesh.
Looking at a service resource I see that the clusterIP has a value like this

“clusterIP”: “5808a34b54f133ba48eabebb1a411024720a668c98783638fbe9a978cf9161c7”, “clusterIPs”: [ “5808a34b54f133ba48eabebb1a411024720a668c98783638fbe9a978cf9161c7” ], “type”: “ClusterIP”,

In addition I see readiness probes failing as follows, again with an unexpected value

“Readiness probe failed: Get "http://1d30535256aa4ba5736dd40358e70d525a8b3370cf83b3340e0e25f0b4dd5982:15021/healthz/ready": dial tcp 1d30535256aa4ba5736dd40358e70d525a8b3370cf83b3340e0e25f0b4dd5982:15021: connect: connection refused”

There appears to be corruption.

Cluster information:

Kubernetes version: 1.23
Cloud being used: (put bare-metal if not on a public cloud)
Installation method: Kind cluster install with metallb
Host OS: Oracle Linux 7.8
CNI and version:
CRI and version: