Setting stream_name in handler used in basicConfig causes all downstream modules to not log to CloudWatch

[et304383]

If I have a root module with code like this:

import mymodule

cw_handler = watchtower.CloudWatchLogHandler(
    log_group=os.environ['LOG_GROUP'],
    stream_name=os.environ['INSTANCE_ID'],
    send_interval=5,
    log_group_retention_days=14
)
console_handler = logging.StreamHandler()

logging.basicConfig(
    level=os.getenv('LOG_LEVEL', logging.INFO),
    handlers=[cw_handler, console_handler]
)

logger = logging.getLogger(__name__)

And then in mymodule.py:

logger = logging.getLogger(__name__)

CW logs are emitted for the root module but nothing shows up for any of the submodules under the stream.

If I omit the stream name parameter, then obviously it works, but logs go to a separate stream per module, which I do not want.

Edit: upon looking for help in other possibly similar issues, I discovered this comment. I set use_queues to False and it works.

Seems this queueing logic is causing a lot of problems.

[petero-dk]

We have this exact issue, confirmed on Django 1.11 running on Python 2.7 (yes old we know)

This however did not show up on Django 1.2.

This also mostly seem to be a problem when running Celery via Django or Django via uwsgi.