Configure Envoy for Contour

This topic tells you how to configure the Envoy for Contour in Tanzu Application Platform (commonly known as TAP) to eliminate upgrade downtime.

By default, Tanzu Application Platform v1.7 installs the Contour’s Envoy pods as a Deployment with 2 replicas instead of a DaemonSet. If you switch from DaemonSet to Deployment, the Envoy pods must be deleted and recreated, which causes downtime during the Tanzu Application Platform v1.7 upgrade.

When running as a Deployment, Contour includes affinity rules such that no Envoy pod runs on the same node as another Envoy pod. This mimics the DaemonSet behavior, without requiring a pod on every node. As a result, your cluster must have as many nodes as the desired replicas for an Envoy deployment.

To avoid the upgrade downtime for Tanzu Application Platform v1.7 and later, you can keep running Envoy as a DaemonSet by adding contour.envoy.workload.type to your values file and setting it to DaemonSet before performing the upgrade:

contour:
  envoy:
    workload:
      type: DaemonSet