grpcurl | Like cURL , but for gRPC : Command-line tool

Since the error is

error getting request data: message type RateRequest has no known field named base

And the json is

According to this GitHub anwser

It is likely a macOS-specific threading issue

On the other hand, according to tkinter's doc. The library may have special event handling logic and threading restrictions, and gRPC Python uses a thread pool to handle incoming requests with gRPC Core handling the IO.

By digging into that doc, we find

A Python interpreter may have many threads associated with it. In Tcl, multiple threads can be created, but each thread has a separate Tcl interpreter instance associated with it. Threads can also create more than one interpreter instance, though each interpreter instance can be used only by the one thread that created it.

Each Tk object created by tkinter contains a Tcl interpreter. It also keeps track of which thread created that interpreter. Calls to tkinter can be made from any Python thread. Internally, if a call comes from a thread other than the one that created the Tk object, an event is posted to the interpreter’s event queue, and when executed, the result is returned to the calling Python thread.

Although I'm not an expert on platform-specific threading models, I do know that macOS is using a "Grand Central Dispatch" task-queue system that hides away a lot of thread-pool details, which is different from Windows.

Thus, without digging into gRPC's thread-pool model, I can't say it's safe to call Tkinter widgets, who keep track of thread that create them, from the volatile gRPC's I/O threads.

You need to actually use a shell, since that's a shell feature. sh -c "foo | bar" or whatever. Also remember that all the relevant commands need to be available in the target image.

You may want to post other parts of the trace log. To debug this issue, we need to:

• Check if the name resolution worked, there should be a log saying the given address has been resolved into ip:port;
• See why each address won't work, like the ip:port is not reachable.

If the problem still can't be solved, I would recommend to post issues to https://github.com/grpc/grpc/issues.

QUARKUS_GRPC_SERVER_HOST should be 0.0.0.0 instead of localhost

I needed to enable HTTP/2 Connections in Cloud Run for this to work.

gRPC Reflection requires bidirectional streaming, so make sure to check the Enable HTTP/2 option (--use-http2) while deploying. That will enable bi-di streaming.

Also make sure to use the :443 port and authenticate to the server if needed by adding Authentication metadata (see Service-to-Service authentication documentation).

Interesting question. I've not tried streaming messages with (the excellent) grpcurl.

The documentation does not explain how to do this but this issue shows how to stream using stdin.

I recommend you try it that way first to ensure that works for you.

If it does, then bundling various messages into a file (out) should also work.

Your follow-on questions suggest you're doing this incorrectly.

• chunk_data is the result of having split the file into chunks; i.e. each of these base64-encoded strings should be a subset of your overall image file (i.e. a chunk).

• your first message should be { "info": "...." }, subsequent messages will be { "chunk_data": "" } until EOF.