View a markdown version of this page

Calling ingress settings - Wickr Enterprise
ConsiderationsReference architectures

This guide provides documentation for Wickr Enterprise. If you're using AWS Wickr, see AWS Wickr Administration Guide or AWS Wickr User Guide.

Calling ingress settings

Wickr supports a calling ingress setting, allowing a client to connect to any calling node within the cluster and have the call route to the correct calling server. Wickr supports four calling ingress types:

  • LoadBalancer (default)

    • The LoadBalancer will be provisioned by the cloud provider (fully on-premise installations will require additional configuration). After the LoadBalancer has been provisioned, the KOTS config must be updated again to provide the load balancer's hostname or IP addresses.

  • NodePort

    • Exposes a NodePort service on each calling node that will serve as the entry point for calling traffic. A hostname resolving to one or more nodes or an IP address of one or more nodes must be provided. You can choose a port range from 30000-32767 for UDP and optionally TCP traffic.

  • Existing NLB

    • Attaches the calling ingress service to an existing NLB. You will need to supply the target group ARN for UDP and optionally, TCP traffic.

  • No Service

    • Select this if you don't need an additional Kubernetes service to allow ingress traffic. This will typically be used with the host network setting to route calling ingress traffic directly to your calling nodes.

Considerations

  • To ensure backwards compatibility with older clients and federated networks without calling ingress, when calling ingress is enabled, the legacy calling mode is still available (direct connection to the calling servers). If changing any default ports, ensure that you do not have any port collisions on the calling nodes.

  • Dual-stack NLBs serving UDP traffic must have IPv6 backend targets. For more information, see Network Load Balancer Target Groups.

  • If you require STIG compliance, you must disable the host network option for calling. If the nodes are configured in dual-stack mode, but the cluster is not, you may lose IPv6 connectivity (assuming an IPv4 cluster).

  • Calling ingress requires pre-defined host names or IP Addresses. Scaling nodes or providing custom routing may require modifying configuration.

  • Default calling ingress ports are 8443 for TCP and 16384 for UDP. Make sure firewalls and security groups allow traffic for these ports, or alternative ports if the defaults are overridden.

Reference architectures

Ingress with load balancer

This option exposes a single load balancer as the entry point for all calling traffic.

  1. For Calling Ingress Type, choose either Load Balancer or Existing NLB. For more information about Existing NLB, reference the NLB stack in the Wickr Enterprise CDK Sample on GitHub.

  2. Do one of the following, depending on the Calling Ingress Type:

    • For Existing NLB, provide the target group ARNs for UDP and TCP traffic and the hostname of the NLB.

    • For Load Balancer, provide the hostname after it is provisioned by Kubernetes.

    Alternatively, for either Calling Ingress Type, you can provide the IP addresses of the load balancer or a custom hostname that points to the load balancer.

  3. (Optional) To combine messaging and calling traffic under a single NLB, choose Existing NLB in the Ingress section, and provide an HTTPS target group.

Ingress with NodePort

This option is useful if host networking is disabled and you don’t want to expose an additional load balancer.

Note

Make sure your firewalls and security groups allow traffic for the NodePorts.

  1. For Calling Ingress Type, choose NodePort.

  2. Add the calling node hostnames or IP addresses.

  3. Disable Calling Host Network.

Direct ingress with HostNetwork

This option does not expose any additional Kubernetes service and allows calling ingress traffic to connect directly over the calling nodes’ host network. This approach is preferred if IPv6 connectivity is required.

  1. For Calling Ingress Type, select No Service.

  2. Add the calling node hostnames or IP addresses.

  3. Enable Calling Host Network.