is there any extra config needed in order to have std_out from a ShellTask show up in the server logs? When I declare a task like this I see the output fine

@task(log_stdout=True)
def sample_print_task():
    print("hello")

When I declare a shell task like this

task_shell = ShellTask(return_all=True, log_stdout=True, log_stderr=True, stream_output=True)
with Flow("test") as flow:
    print_test = task_shell(command="echo hi", task_args={"name": "hi"})

I don't see anything printed after I register and run the flow. I have

log_to_cloud

set to True on the agent I am running

Ah that would make for a good enhancement if you want to open an issue for it on the repo! 🙂 Currently the shell task looks to not do anything with the base task class’s

log_stdout

so you would need to do something like:

@task(log_stdout=True)
def show_output(lines):
    print(lines)

# OR

@task
def show_output(lines):
    from prefect import context
    logger = context.get("logger")
    <http://logger.info|logger.info>(lines)

task_shell = ShellTask(return_all=True, log_stderr=True, stream_output=True)
with Flow("test") as flow:
    print_test = task_shell(command="echo hi", task_args={"name": "hi"})
    show_output(print_test)

that works, but only if the output is small enough. One of my tasks returned an error with

Failed to write log with error: 413 Client Error: Payload Too Large for url: <http://host.docker.internal:4200/graphql>

That could be due to debug logs not being on by default. For your agent try setting the env var

--env PREFECT__LOGGING__LEVEL=DEBUG

(this will pass the env var to all flow runs it creates) or set it on the single flow’s run config directly

I have both

PREFECT__LOGGING__LEVEL=DEBUG

and

PREFECT__LOGGING_LOG_TO_CLOUD=True

on my agent, still no logs from ShellTasks

Interesting 🤔 could you open an issue with how you are running everything / a reproducible example on the repo because that sounds like it could be a bug

ok so when in my efforts to create a reproducible example, everything was actually working fine. Turns out my issue was in my docker image there was a different version of prefect installed vs. what my server was running on