Add health check handler for worker service

[tsurdilo]

Looks as currently worker service https://github.com/temporalio/temporal/tree/master/service/worker
does not expose a gRPC handler and the health check method like matching and history services do.

Please add this if possible.

[yiminc]

Worker does not have a gRPC service endpoint. It would be good to have sdk to implement the health check so whoever running a worker would have it.

[paymog]

While we wait for this to get implemented, does anyone have a workaround for those of us running workers in kubernetes? We recently deployed a broken docker image for our worker and didn't realize for a few days.

[tredmon]

While we wait for this to get implemented, does anyone have a workaround for those of us running workers in kubernetes?

I've just started playing with temporal in kubernetes, but a simple liveness probe could look like (assuming your image has sh and pidof installed and that your worker has a binary which is unique on your pod runtime):

livenessProbe:
  exec:
    command:
      - /bin/sh
      - -c
      - pidof your-worker-binary >/dev/null

For the readiness probe I'm considering creating a file in an emptyDir volume when the worker is ready, and creating another file when the worker receives a termination signal. Then the readiness probe could look for the existence of said files.