Debug level is not propagated to service threads

[BorjaEst]

Description

Logger output on predict and train functions with lower level than WARN is not printed on stderr.

Steps to Reproduce

1. Download demo_app (or any other use case).
2. Create a logger in the api module and log in get_meatada and predict using debug, for example:

import logging
logger = logging.getLogger(__name__)

def get_metadata():
    logger.debug("Some debug information")
    return {"status": "All ok, debug was printed"}

def get_predict_args():
    return {}

def predict(**kwargs):
    logger.debug("Some debug information")
    return {"status": "Debug was not printed"}

3. Create a configuration file which set debug level to "DEBUG" (./deepaas.conf):

[DEFAULT]
debug = true
log_date_format = %Y-%m-%d %H:%M:%S

4. Run using the confg file: deepaas-run --config-file deepaas.conf

Expected behavior:

When calling predict / train method, debug information should be printed.
(like in ´get_metadata`)

Actual behavior

Only logger print levels higher or equal than WARN are printed on predict / train method.

Versions

$ deepaas-run --version
2.1.0

Extra information

I see the functions predict and train are executed by workers. However, the logger level at those workers seems to not be inherited from the original logger level.

[BorjaEst]

@IgnacioHeredia, I think this issue is related to what we spoke today. I was testing with @vykozlov and there might be a race condition somewhere:

• Starting with --debug and no config, prints debug info at the beginning but not from methods.
• Starting without --debug but config, does not print debug at the beginning but does for methods.

Also, when I created this issue, I was testing with the config and not with --debug, it might happen that aiohttp does not propagate logger to workers if arg --debug is not set? Not sure how much is worth value to investigate if a new version is comming with different web framework.