process stdout/stderr is swallowed?

[zzlk]

Using the docker runtime, if you try to launch an executable that does not exist, then docker itself will throw an error and you get a nice error message in stdout from the hydroplane daemon:

docker.errors.APIError: 400 Client Error for http+docker://localhost/v1.42/containers/d72eebbf516c56ae5981d7d0a82f46133230aef82f4713f0db4b0b1e86ffe290/start: Bad Request ("failed to create shim task: OCI runtime create failed: runc create failed: unable to start container process: exec: "./kvs_benchmark": stat ./kvs_benchmark: no such file or directory: unknown")

But if the executable exists but is called with the wrong arguments, then from docker's point of view launching the exe was a success, but it will usually terminate instantly with some usage error message on it's own stdout. This is not logged to the hydroplane stdout so you only get this message:

2023-02-27 23:32:48,905 - INFO - [uvicorn.access] 127.0.0.1:43262 - "POST /process HTTP/1.1" 201

Indicating that the process was created successfully but in reality it was not. I guess more complex programs would have their own logging/persistence infra that you could then go and inspect but for smaller programs like the hydro example ones, it's not easy to debug them without a useful error message.