kubernetes/kops · error

unexpected amount of ipv6 prefixes on interface %q: %v

Error message

unexpected amount of ipv6 prefixes on interface %q: %v

What it means

After finding the node's single ENI, the controller expects exactly one IPv6 prefix delegated to it, because it assigns that single /80 prefix as the node's podCIDR. This error is thrown when DescribeNetworkInterfaces reports a number of Ipv6Prefixes other than 1 on that interface.

Source

Thrown at cmd/kops-controller/controllers/awsipam.go:141

			Filters: []ec2types.Filter{
				{
					Name: new("attachment.instance-id"),
					Values: []string{
						instanceID,
					},
				},
			},
		})
		if err != nil {
			return ctrl.Result{}, err
		}

		if len(eni.NetworkInterfaces) != 1 {
			return ctrl.Result{}, fmt.Errorf("unexpected number of network interfaces for instance %q: %v", instanceID, len(eni.NetworkInterfaces))
		}

		if len(eni.NetworkInterfaces[0].Ipv6Prefixes) != 1 {
			return ctrl.Result{}, fmt.Errorf("unexpected amount of ipv6 prefixes on interface %q: %v", *eni.NetworkInterfaces[0].NetworkInterfaceId, len(eni.NetworkInterfaces[0].Ipv6Prefixes))
		}

		ipv6Address := aws.ToString(eni.NetworkInterfaces[0].Ipv6Prefixes[0].Ipv6Prefix)
		podCIDRs := []string{ipv6Address}
		if err := patchNodePodCIDRs(r.coreV1Client, ctx, node, podCIDRs); err != nil {
			return ctrl.Result{}, err
		}
	}

	return ctrl.Result{}, nil
}

func (r *AWSIPAMReconciler) SetupWithManager(mgr ctrl.Manager) error {
	return ctrl.NewControllerManagedBy(mgr).
		Named("aws_ipam").
		For(&corev1.Node{}).
		Complete(r)
}

View on GitHub (pinned to 4c8573c808)

Solutions

1. Ensure the subnet and instance are configured for IPv6: subnet has an IPv6 CIDR and the ENI has an IPv6 prefix delegated (check in EC2 console / aws ec2 describe-network-instances --query ...Ipv6Prefixes).
2. Wait/requeue briefly — prefix delegation can lag instance launch; treat 0 prefixes as transient rather than fatal if the node is new.
3. Verify the cluster is actually running with the IPv6 pod-CIDR IPAM mode; this controller should only be enabled for kops IPv6 clusters.
4. If multiple prefixes are delegated intentionally, adjust the controller to pick the first prefix instead of erroring.

Example fix

// before
if len(eni.NetworkInterfaces[0].Ipv6Prefixes) != 1 {
	return ctrl.Result{}, fmt.Errorf("unexpected amount of ipv6 prefixes on interface %q: %v", *eni.NetworkInterfaces[0].NetworkInterfaceId, len(eni.NetworkInterfaces[0].Ipv6Prefixes))
}
// after
prefixes := eni.NetworkInterfaces[0].Ipv6Prefixes
if len(prefixes) == 0 {
	klog.Warningf("no ipv6 prefixes yet on interface %q; requeueing", *eni.NetworkInterfaces[0].NetworkInterfaceId)
	return ctrl.Result{RequeueAfter: 10 * time.Second}, nil
}
if len(prefixes) > 1 {
	return ctrl.Result{}, fmt.Errorf("unexpected amount of ipv6 prefixes on interface %q: %v", *eni.NetworkInterfaces[0].NetworkInterfaceId, len(prefixes))
}

resp, _ := ec2Client.DescribeNetworkInterfaces(ctx, &ec2.DescribeNetworkInterfacesInput{Filters: eniFilters})
if len(resp.NetworkInterfaces) == 1 && len(resp.NetworkInterfaces[0].Ipv6Prefixes) == 0 && recentlyLaunched(instance) {
	// prefix delegation not propagated yet — retry later instead of failing
}

func hasSingleIPv6Prefix(iface ec2types.NetworkInterface) bool {
	return len(iface.Ipv6Prefixes) == 1 && iface.Ipv6Prefixes[0].Ipv6Prefix != nil
}

result, err := r.Reconcile(ctx, req)
if err != nil && strings.Contains(err.Error(), "ipv6 prefixes") {
	klog.Warningf("ipv6 prefix count mismatch, requeueing with backoff: %v", err)
	return ctrl.Result{RequeueAfter: 30 * time.Second}, nil
}

When it happens

Trigger: eni.NetworkInterfaces[0].Ipv6Prefixes has length 0 (the ENI was not allocated an IPv6 prefix yet, or IPv6 prefix delegation is disabled on the subnet/instance) or length > 1 (multiple /80 prefixes delegated, e.g. after kubelet/controller restarts or manual assignment), while patching expects exactly one.

Common situations: Cluster not fully configured for IPv6 prefix delegation (subnet without an IPv6 CIDR, 'assignIpv6AddressOnCreation' or prefix delegation not enabled); node launched before prefix assignment completed; manual EC2 changes adding or removing IPv6 prefixes; running an IPv4-only cluster while the IPv6 IPAM controller is enabled.

Related errors

• unexpected number of network interfaces for instance %q: %v
• error deleting EgressOnlyInternetGateway %q: %v
• error listing EgressOnlyInternetGateway: %v
• error listing EgressOnlyInternetGateways: %v
• found multiple EgressOnlyInternetGateways matching tags

AI-assisted analysis of kubernetes/kops@4c8573c808 (2026-09-05). Data as JSON: /api/errors/8fe78fd4ed264a35. Report an issue: GitHub.