Not able to query grpc healthcheck endpoints using grpcurl

[keshavnandan]

trafficstars

I created a gRPC service and added grpc healthcheck endpoints as per the gRPC Health Checking Protocol and I am getting the status as SERVING when using the grpc_health_probe command. However, when I am calling the healthcheck endpoints using grpcurl command, its not working.

➜  healthcheck git: ✗ grpcurl -v -plaintext -protoset ./healthcheck.protoset -d '{"service": "healthcheck"}' localhost:8080 healthcheck.HealthCheck/Check

Resolved method descriptor:
rpc Check ( .healthcheck.HealthCheckRequest ) returns ( .healthcheck.HealthCheckResponse );

Response headers received:
(empty)

Response trailers received:
content-type: application/grpc
Sent 1 request and received 0 responses
ERROR:
  Code: Unimplemented
  Message: method Check not implemented

Following is my proto file which has both Check and Watch methods

syntax = "proto3";
option go_package = "healthcheck/healthcheck";
package healthcheck;

message HealthParams {
  string service = 1;
}

message HealthCheckRequest {
  string service = 1;
}

message HealthCheckResponse {
  enum ServingStatus {
    UNKNOWN = 0;
    SERVING = 1;
    NOT_SERVING = 2;
    SERVICE_UNKNOWN = 3;  // Used only by the Watch method.
  }
  ServingStatus status = 1;
}

message Health {
  // The health status
  string status = 1;
}

service HealthCheck{
  rpc Get(HealthParams) returns (Health) {}
  rpc Check(HealthCheckRequest) returns (HealthCheckResponse);
  rpc Watch(HealthCheckRequest) returns (stream HealthCheckResponse);
}

What's working is the following

➜  healthcheck git: ✗ grpc-health-probe -addr=localhost:8080
status: SERVING

[mdable2]

I am having a similar issue. I am running a gRPC API on localhost:50051 and when I try to do grpcurl localhost:50051 grpchealthv1.Health/Check I get Failed to dial target host "localhost:50051": dial tcp 127.0.0.1:50051: connect: connection refused. I am able to see the host via telnet and can connect and hit the endpoint using a Python client.

I am also using the gRPC Health Checking Protocol (just changed the package to be grpchealthv1). Any help on what I am doing wrong is greatly appreciated.

[jhump]

@keshavnandan, sorry, but grpcurl does not do anything special with health-checking. That call would just be a normal RPC. One thing I do note that is suspicious: the gRPC health-checking protocol uses a service named grpc.health.v1.Health, but you have provided a proto where the name is healthcheck.HealthCheck. Why the difference? If you've followed the directions at that other doc page, I think that means you are using a health check impl that is part of an official gRPC runtime, which means the problem is likely this name mismatch. Try again but use the actual proto for the health-check service?.

@mdable2, "connection refused" is a typical TCP connection error -- nothing magical about grpcurl in that respect. Are you sure the server was actually up and running when you tried that? I don't see how grpcurl could get that error and another program not, unless the server was not accepting connections when grpcurl was invoked (not up or not listening on socket).