No services are advertised if any service is missing an EndpointSlice

[thxer]

What happened?

When any Service in the cluster is missing an EndpointSlice, kube-router stops advertising all Services’ VIPs. The error from processing the problematic Service bubbles up and short-circuits the whole getVIPs() pass, resulting in no VIPs being advertised cluster-wide.

What did you expect to happen?

Kube-router should continue advertising VIPs for Services that are healthy/ready even if one Service has a configuration problem (e.g., missing EndpointSlice). Errors for a single Service should be handled per-service (logged and skipped), not abort the entire reconciliation.

How can we reproduce the behavior you experienced?

Steps to reproduce the behavior:

1. Deploy kube-router with service VIP advertising enabled.
2. Ensure most Services are healthy and have EndpointSlices.
3. Create (or break) one Service so that it has no EndpointSlice (e.g., misconfigure it or temporarily stop the EndpointSlice controller).
4. Observe kube-router logs and advertised routes/VIPs: VIP advertisement halts for all Services due to the error from the single problematic Service.

System Information (please complete the following information)

• Kube-Router Version (kube-router --version): 2.6.0
• Kubernetes Version (kubectl version) : 1.32.9
• Cloud Type: on premise
• Kubernetes Deployment Type: kubeadm
• Kube-Router Deployment Type: daemonset
• Cluster Size: [e.g. 200 Nodes]

[aauren]

Thanks for reporting this! I feel a bit bad that a bug of this size made it into the release. I'll try to get to this one in the next day or two. In the meantime, I'm guessing that this bug won't be a problem on a <2.6.0 release of kube-router, so hopefully downgrading will resolve the issue immediately until we get a fix out.