1.28 version information and update actions

Update before master

The following table shows the actions that you must take before you update the Kubernetes master.

Update after master

What do I need to do?

The steps you need to take to ensure that your VPC cluster worker nodes continue pulling images from the Container Registry depend on your network security setup.

• If you use the default network rules for all security groups, network ACLs, and network policies, you do not need to take any action.
• If you have a customized network security setup that blocks certain TCP connections within the VPC, you must take additional actions before updating to or creating a new cluster at version 1.28. Make the adjustments in the following sections to ensure that connections to the new VPE Gateway for Registry are allowed.

Regardless of whether you need to take additional steps, if you keep other clusters in the VPC that do not run version 1.28 you must refresh the cluster master on those clusters. This ensures that the correct updates are applied to the non-1.28 clusters so that traffic to the new VPE is allowed.

I have custom security groups. What do I change?

The necessary allow rules are automatically added to the IBM-managed kube-<cluster_ID> cluster security group when you update to or create a cluster at version 1.28. However, if you created a VPC cluster that does NOT use the kube-<cluster_ID> cluster security group rules, you must make sure that the following security group rules are implemented to allow traffic to the VPE gateway for registry. If the rules are not already implemented in your custom setup, add them. Each of these rules must be created for each zone in the VPC and must specify the entire VPC address prefix range for the zone as the destination CIDR. To find the VPC address prefix range for each zone in the VPC, run ibmcloud is vpc-address-prefixes <vpc_name_or_id>.

Add the following rules to your custom security group.

To make these rules more restrictive, you can set the destination to the security group used by the VPE Gateway or you can specify the exact VPE Gateway reserved IP address. Note that these IP addresses can change if all cluster workers in a VPC are removed.

I have custom ACLs. What do I change?

If VPC networks ACLs that apply to your cluster workers have been customized to only allow certain egress and ingress traffic, make sure that the following ACL rules, or equivalent rules, are implemented to allow connections to and from the VPE Gateway for Registry. If the rules are not already implemented, add them. Each of these rules must be created for each zone in the VPC and must specify the entire VPC address prefix range for the zone as the source (for outbound rules) or destination (for inbound rules) CIDR. To find the VPC address prefix range for each zone in the VPC, run ibmcloud is vpc-address-prefixes <vpc_name_or_id>. The priority for each rule should be higher than any rule that would otherwise deny this traffic, such as a rule that denies all traffic.

Add the following rules to your custom ACLs.

I have custom network policies. What do I change?

If you use Calico policies to restrict outbound connections from cluster workers, you must add the following policy rule to allow connections to the VPE Gateway for Registry. This policy must be created for each zone in the VPC and must specify the entire VPC address prefix range for the zone as the destination CIDR. To find the VPC address prefix range for each zone in the VPC, run ibmcloud is vpc-address-prefixes <vpc_name_or_id>.

apiVersion: projectcalico.org/v3
kind: GlobalNetworkPolicy
metadata:
  name: allow-vpe-gateway-registry
spec:
  egress:
  - action: Allow
    destination:
      nets:
      - <entire-vpc-address-prefix-range> # example: 10.245.0.0/16
      ports:
      - 443
    protocol: TCP
    source: {}
  order: 500
  selector: ibm.role == 'worker_private'
  types:
  - Egress