kubernetes/kops · error

unhandled LoadBalancer type %q

Error message

unhandled LoadBalancer type %q

What it means

The GCE API load balancer Build method (pkg/model/gcemodel/api_loadbalancer.go:422) switches on spec.api.loadBalancer.type and only implements the supported cases (e.g. public/internal); any other value falls to the default branch and returns 'unhandled LoadBalancer type'. It is a compile-time-style enum guard ensuring the model never silently skips LB creation.

Source

Thrown at pkg/model/gcemodel/api_loadbalancer.go:422

			return err
		}
		// We always create the internal load balancer also;
		// it allows us to restrict access to only the nodes.
		if err := b.createInternalLB(c); err != nil {
			return err
		}

		return b.addFirewallRules(c)

	case kops.LoadBalancerTypeInternal:
		if err := b.createInternalLB(c); err != nil {
			return err
		}

		return b.addFirewallRules(c)

	default:
		return fmt.Errorf("unhandled LoadBalancer type %q", lbSpec.Type)
	}
}

View on GitHub (pinned to 4c8573c808)

Solutions

1. Set spec.api.loadBalancer.type to a valid GCP-supported value ('public' or 'internal')
2. Remove the api.loadBalancer block entirely to use DNS-based endpoints instead
3. Dump the spec with 'kops get cluster -o yaml' to spot typos, casing, or empty values
4. After fixing, run 'kops update cluster' to regenerate the GCE LB tasks

Example fix

// before
api:
  loadBalancer:
    type: ""
// after
api:
  loadBalancer:
    type: internal

t := cluster.Spec.API.LoadBalancer.Type
switch t {
case kops.LoadBalancerTypePublic, kops.LoadBalancerTypeInternal, "":
	// "" only valid if the whole loadBalancer block is omitted
default:
	return fmt.Errorf("GCP: unsupported api.loadBalancer.type %q", t)
}

func supportedGCPAPIType(spec *kops.ClusterSpec) bool {
	lb := spec.API.LoadBalancer
	return lb == nil || lb.Type == kops.LoadBalancerTypePublic || lb.Type == kops.LoadBalancerTypeInternal
}

When it happens

Trigger: Building the GCE model for a cluster whose spec.api.loadBalancer.type is not one of the values handled by the switch — usually an empty string (loadBalancer block present but type omitted), or an invalid/foreign enum value pasted into the spec.

Common situations: Copy-pasting a loadBalancer block from AWS clusters into a GCP spec; leaving 'type:' blank after editing; schema drift between kOps versions changing the LoadBalancerType enum.

Related errors

• unhandled LoadBalancer type %q
• instance group %q has %d zones, which is not yet supported f
• found multiple instance groups matching MIG %q
• error creating kops config template: %w
• error creating gcp machine template: %w

AI-assisted analysis of kubernetes/kops@4c8573c808 (2026-09-05). Data as JSON: /api/errors/936bedf091e73583. Report an issue: GitHub.