Expose a liveness or readiness probe for off-cluster services that expose TCP endpoints.
The tcp-healthz container leverages the exechealthz server maintained by the Kubernetes community. This server exposes a /healthz endpoint that is driven by the successful execution of a given command.
The tcp-healthz container's entrypoint is still exechealthz. Using arguments, we tell exechealthz to run the tcp-healthz binary with the relevant ports.
In the case that all TCP endpoints are up and accessible, tcp-healthz will exit with a zero status. This results in successful responses to incoming /healthz requests.
In the case that a TCP endpoint is deemed unaccessible, tcp-healthz will exit with a failure status. This results in non-successful responses to incoming /healthz requests.
See the examples directory for usage examples.