gdong42 / grpc-mate Goto Github PK

Currently grpc-mate supports only unary calls. It would be nice to support streaming as well. Need to address how to workaround client/server streaming with HTTP/1.1

Also, add instructions on how to install and run via docker.

Currently grpc-mate returns HTTP status code 502 when backend gRPC server does not support reflection. It might be more user-friendly to also give a descriptive message about this fact.

Is it possible to set custom headers?

I can't interact with grpc-mate by browser

Maybe provide something like auto inject container to pods that have specific metadata tags. Like grpc-mate.enabled, or something better.

Currently grpc-mate provides /actuator/services endpoint that returns all services, methods and types. An example response:

{
   "services":[
      {
         "name":"grpc.examples.echo.Echo",
         "methods":[
            {
               "name":"UnaryEcho",
               "input":"grpc.examples.echo.EchoRequest",
               "output":"grpc.examples.echo.EchoResponse",
               "route":"/grpc.examples.echo.Echo/UnaryEcho"
            },
            {
               "name":"ServerStreamingEcho",
               "input":"grpc.examples.echo.EchoRequest",
               "output":"grpc.examples.echo.EchoResponse",
               "route":"/grpc.examples.echo.Echo/ServerStreamingEcho"
            },
            {
               "name":"ClientStreamingEcho",
               "input":"grpc.examples.echo.EchoRequest",
               "output":"grpc.examples.echo.EchoResponse",
               "route":"/grpc.examples.echo.Echo/ClientStreamingEcho"
            },
            {
               "name":"BidirectionalStreamingEcho",
               "input":"grpc.examples.echo.EchoRequest",
               "output":"grpc.examples.echo.EchoResponse",
               "route":"/grpc.examples.echo.Echo/BidirectionalStreamingEcho"
            }
         ]
      },
      {
         "name":"grpc.reflection.v1alpha.ServerReflection",
         "methods":[
            {
               "name":"ServerReflectionInfo",
               "input":"grpc.reflection.v1alpha.ServerReflectionRequest",
               "output":"grpc.reflection.v1alpha.ServerReflectionResponse",
               "route":"/grpc.reflection.v1alpha.ServerReflection/ServerReflectionInfo"
            }
         ]
      },
      {
         "name":"helloworld.Greeter",
         "methods":[
            {
               "name":"SayHello",
               "input":"helloworld.HelloRequest",
               "output":"helloworld.HelloReply",
               "route":"/helloworld.Greeter/SayHello"
            }
         ]
      }
   ],
   "types":[
      {
         "name":"grpc.examples.echo.EchoResponse",
         "template":{
            "message":""
         }
      },
      {
         "name":"grpc.reflection.v1alpha.ServerReflectionRequest",
         "template":{
            "host":"",
            "fileContainingExtension":{
               "containingType":"",
               "extensionNumber":0
            }
         }
      },
      {
         "name":"grpc.reflection.v1alpha.ServerReflectionResponse",
         "template":{
            "validHost":"",
            "originalRequest":{
               "host":"",
               "fileContainingExtension":{
                  "containingType":"",
                  "extensionNumber":0
               }
            },
            "errorResponse":{
               "errorCode":0,
               "errorMessage":""
            }
         }
      },
      {
         "name":"helloworld.HelloRequest",
         "template":{
            "name":""
         }
      },
      {
         "name":"helloworld.HelloReply",
         "template":{
            "message":""
         }
      },
      {
         "name":"grpc.examples.echo.EchoRequest",
         "template":{
            "message":""
         }
      }
   ]
}

The resulting output could be quite long and hard to read for a real-world gRPC server that has loads of services. It would be nice to filter these results by service name and method to only show relevant parts.

e.g.

• /actuator/services?name=grpc.examples.echo.Echo returns only service with name grpc.examples.echo.Echo, its methods, and types that are relevant.
• /actuator/services?method=ServerStreamingEcho returns services that have method name as ServerStreamingEcho, this method, and types that are relevant.

Of course, both name and method can be specified at the same time.

Not sure if filtering type is a valid use case as well. Let's not add the type parameter for now. Keep it simple.

Hi.
I have been having issues discovering services that supposedly have ServerReflection enabled. I am running the helloworld example from the grpc official repository.

However, when I try to access it via grpc-mate, it seems like it is unable to discover it, and I have been unable to find any issues or guides that would help me figure out what is not working.

Thank you before hand for your time.

For users that develop gRPC services alongside gRPC Mate, a web UI of the latter is handy. User can check with services/methods they are working on, and quickly shoot a gRPC call with JSON payload through the web UI and check the results.

Although one can achieve the same using CLI commands, it would be more efficient to click around using a UI while he/she is debugging.

Features Must Have:

• Show information of the backend gRPC service it connects to, e.g. the address, the port, etc
• List all services and methods the backend exposes.
• Easy to construct a request with selected service/method, with optional input message ( better pre-filled with sample values)
• Show the result status code and result message, or error message if the request is not successful

Features Nice to Have:

• Health check status

Some considerations on implementation:

• it should be opt-in by an optional command line flag, e.g. -ui
• the static resources including the template files should be packaged into the standalone executable binary.

I keep getting the following error from my gRPC server:

2022/07/10 16:42:10 http: TLS handshake error from 172.22.224.1:60652: tls: first record does not look like a TLS handshake

Any chance of TLS being supported in the future? I'm not sure what I'll try next. I've been using grpc-mate from within Docker.

Inspect if the target gRPC service has a health check service registered, monitor the service result, if it returns non-OK status, gRPC Mate can show something like DOWN or 503 status to indicate the health status.

The health check probe should be scheduled at a fixed rate, with frequency set to 10s or so, and the frequency should be adjustable.