lambda Invoked once, code logs twice, when called multiple times

Question

I have a lambda on aws running a dockerized python code. when I run the lambda 2 times back to back, it will double log, but when ran 3 times or more, or only once, it only logs one time.

now, what do I mean by double logging?

START RequestId: <UUID>
{"message": "starting to do something"...}
{"message": "starting to do something"...}
{"message": "another log message with context data in it"...}
{"message": "another log message with context data in it"...}
REPORT RequestId: <UUID>
END RequestId: <UUID>

as you can see, lambda is logging its system logs only once, but my code logs are logged twice. I also checked monitoring charts for my lambda and it definitely ran once, but the code is double logged.

I do log my code like the following:

def make_logger():
    logger = logging.getLogger('my_name')
    ...
    return logger


def handler(event,context):
    logger = make_logger()
    logger.info("starting to do something")
    message = json.loads(event['Records'][0]['body']
    # some other code and logic
    logger = logging.LoggerAdapter(logger, message)
    logger.info("another log message with context data in it") 

I'm suspecting that something fishy is going on there with how I'm making my logger. I don't think it's the loggerAddapter because logs before it is also duplicated.

update1: while adding print statements to my code as suggested in the comments, I could not get the print statement results, but something interesting happened:

I passed in data that would break the lambda code before the adapter was run to see the effect, and the lambda retries, and then I got up to 8 times the logs, in one run for the retries (second retry).

Answer 1

the answer was in a small part I left out in the question:

def make_logger():
    logger = logging.getLogger('my_name')

    logger.propagate = False
    formatter = UtilFormatter(get_default_log_format())
    log_handler = logging.StreamHandler()
    log_handler.setFormatter(formatter)
    logger.addHandler(log_handler)

    logger.setLevel(logging.INFO)
    return logger

when a lambda is called multiple times back to back, it will run in the same lambda instance. so they will be running make_logger function and will be adding a handler to the list of handlers we already have because these handlers won't be cleared as we expect them. the fix was a simple if statement:

def make_logger():
    logger = logging.getLogger('my_name')

    logger.propagate = False
    if not logger.handlers:
        formatter = UtilFormatter(get_default_log_format())
        log_handler = logging.StreamHandler()
        log_handler.setFormatter(formatter)
        logger.addHandler(log_handler)

        logger.setLevel(logging.INFO)
    return logger