Skip to main content

Accessing Your Application

This topic describes the networking options for accessing applications deployed on clusters created with the Replicated compatibility matrix. It also describes how to use and manage compatibility matrix tunnels.

Networking Options​

After deploying your application into compatibility matrix clusters, you will want to execute your tests using your own test runner. In order to do this, you need to access your application. Compatibility matrix offers several methods to access your application.

Some standard Kubernetes networking options are available, but vary based on the distribution. For VM-based distributions, there is no default network route into the cluster, making inbound connections challenging to create.

Port Forwarding​

Port forwarding is a low-cost and portable mechanism to access your application. Port forwarding works on all clusters supported by compatibility matrix because the connection is initiated from the client, over the Kubernetes API server port. If you have a single service or pod and are not worried about complex routing, this is a good mechanism. The basic steps are to connect the port-forward, execute your tests against localhost, and then shut down the port-forward.

LoadBalancer​

If your application is only running on cloud services (EKS, GKE, AKS) you can create a service of type LoadBalancer. This will provision the cloud-provider specific load balancer. The LoadBalancer service will be filled by the in-tree Kubernetes functionality that's integrated with the underlying cloud provider. You can then query the service definition using kubectl and connect to and execute your tests over the LoadBalancer IP address.

Ingress​

Ingress is a good way to recreate customer-representative environments, but the problem still remains on how to get inbound access to the IP address that the ingress controller allocates. Ingress is also not perfectly portable; each ingress controller might require different annotations in the ingress resource to work properly. Supported ingress controllers vary based on the distribution. Compatibility matrix supports ingress controllers that are running as a NodePort service.

Compatibility Matrix Tunnels (Alpha)​

All VM-based compatibility matrix clusters support tunneling traffic into a NodePort service. When this option is used, Replicated is responsible for creating the DNS record and TLS certs. Replicated will route traffic from :443 and/or :80 into the NodePort service you defined. For more information about using tunnels, see Managing Compatibility Matrix Tunnels (Alpha) below.

The following diagram shows how the traffic is routed into the service using the compatibility matrix tunnels:

compatibility matrix ingress

View a larger version of this image

Managing Compatibility Matrix Tunnels (Alpha)​

note

Compatibity matrix tunnels are an alpha feature.

Tunnels are viewed, created, and removed using the replicated CLI, GitHub Actions, or directly with the Vendor API v3. There is no limit to the number of tunnels you can create for a cluster and multiple tunnels can connect to a single service, if desired.

Limitations​

Compatibility Matrix tunnels have the following limitations:

  • Subdomains and wildcards are not supported.
  • One tunnel can only connect to one service. If you need fanout routing into different services, consider installing the nginx ingress controller as a NodePort service and exposing it.
  • Tunnels are not supported for cloud distributions (EKS, GKE, AKS).

Supported Protocols​

A tunnel can support one or more protocols. The supported protocols are HTTP and HTTPS. WebSockets, GRPC, and other protocols are not routed into the cluster.

Exposing Ports​

Once you have a node port available on the cluster, you can use the replicated CLI to expose the node port to the public internet. This can be used multiple times on a single cluster.

replicated cluster port expose \
    [cluster id] \
    --port [node port] \
    --protocol [protocol]

For example, if you have the nginx ingress controller installed and the node port is 32456:

% replicated cluster ls
ID          NAME                           DISTRIBUTION    VERSION       STATUS         
1e616c55    tender_ishizaka                k3s             1.29.2        running        

% replicated cluster port expose \
    1e616c55 \
    --port 32456 \
    --protocol http \
    --protocol https
note

You can expose a node port that does not yet exist in the cluster. This is useful if you have a deterministic node port, but need the DNS name as a value in your Helm chart.

Viewing Ports​

To view all exposed ports, use the replicated CLI port ls subcommand with the cluster ID:

replicated cluster port ls 1e616c55

Removing Ports​

Exposed ports are automatically deleted when a cluster terminates. If you want to remove a port (and the associated DNS records and TLS certs) prior to cluster termination, run the port rm subcommand with the cluster ID:

replicated cluster port rm 1e616c55 --port 32456 --protocol http --protocol https

You can remove just one protocol, or all. Removing all protocols also removes the DNS record and TLS cert.