VTAdmin fails on IPv6-only clusters

[jordangarside]

VTAdmin fails to connect to vtctld on IPv6-only Kubernetes clusters due to improper address formatting.

Error:

W1219 23:35:57.273443  1 component.go:39] [core] [Channel #1 SubChannel #3]
grpc: addrConn.createTransport failed to connect to {
  Addr: "fd01:10:100:1a01::c0f0:15999",
  ServerName: "",
}. Err: connection error: desc = "transport: Error while dialing: dial tcp: address fd01:10:100:1a01::c0f0:15999: too many colons in address"

Root Cause:

The discovery config uses simple string formatting:

vitess-operator/pkg/controller/vitesscluster/reconcile_vtadmin.go

Lines 289 to 303 in 839329a

	"vtctlds": [
	{
	"host": {
	"fqdn": "%s:%d",
	"hostname": "%s:%d"
	}
	}
	],
	"vtgates": [
	{
	"host": {
	"hostname": "%s:%d"
	}
	}
	]

"hostname": "%s:%d"

With an IPv6 ClusterIP and port, this produces:
fd01:10:100:1a01::c0f0:15999

Expected Behavior:

IPv6 addresses must be bracketed per RFC 2732:
[fd01:10:100:1a01::c0f0]:15999

Suggested Fix:

Use net.JoinHostPort() which handles both IPv4 and IPv6 correctly:

"hostname": net.JoinHostPort(vtctldServiceIP, strconv.Itoa(vtctldServiceGrpcPort))

Environment:

• Kubernetes: IPv6-only (single-stack)
• Operator image: planetscale/vitess-operator:latest
• Digest: sha256:459c3570105bb7fa986b6049b733833def40aa47bc787d83d3f656d9082f9737