Add gRPC health check service support to HC filter

[futangw]

Hi, I'd like to setup the gRPC health check between two envoys, but looks envoy as a server (upstream) does not generate gRPC health check response?

I had client envoy setting this: health_checks: [{
timeout: 0.5s,
interval: 5s,
unhealthy_threshold: 2,
healthy_threshold: 3,
grpc_health_check: { service_name: "ic" }
}]

and I had the server envoy setting this: http_filters:

• name: envoy.health_check
  config:
  pass_through_mode: false
  endpoint: /grpc.health.v1.Health/Check

But looks the server envoy recognizes 'this is a health check' because the statistics "http.ingress_http.tracing.health_check" is creasing, but on the client envoy, the cluster.grpc.health_check.failure is creasing too.

The tcpdump of client envoy health check request as below:

The tcpdump of server envoy health check response as below:

when I set "pass_through_mode: true", I did see the HC message pass to my application.

the envoy version is: envoy version: 5a49ba651b283ace69ff809fe48243ba687375d3/1.6.0-dev/Clean/RELEASE

thank you!

[mattklein123]

That's correct that right now Envoy itself does not support the gRPC health check service for responding to health checks. I'm going to go ahead and mark this as a feature request.

[futangw]

@mattklein123 Matt, thanks for the clarification and willing to take enhancement.