Why does Logger create a new file before the maxBytes limit is exceeded?

Question

Python

logger stops considering maxBytes and starts backupCount

logger_django.conf

ROTATE_LOG_SIZE = 10 * 1024 * 1024
ROTATE_LOG_COUNT = 5

LOGGING = {
    'version': 1,
    'disable_existing_loggers': False,
    'formatters': {
        'verbose': {
            'format': '{levelname} {asctime} {module}:{lineno} {message}',
            'style': '{',
            'datefmt' : '%Y-%m-%d %H:%M:%S',
        },
        'extended': {
            'format': '{levelname} {asctime} {message}',
            'style': '{',
            'datefmt' : '%Y-%m-%d %H:%M:%S',
        },
        'simple': {
            'format': '{levelname} {message}',
            'style': '{',
        },
    },
    'handlers': {
        'debug': {
            'level': 'DEBUG',
            'class': 'logging.FileHandler',
            'filename': os.path.join(LOG_ROOT, 'debug.log'),
            'formatter': 'verbose',
        },
        'logger': {
            'level': 'DEBUG',
            'class': 'logging.handlers.RotatingFileHandler',
            'filename': os.path.join(LOG_ROOT, 'logger.log'),
            'formatter': 'extended',
            'maxBytes': ROTATE_LOG_SIZE,
            'backupCount': ROTATE_LOG_COUNT,
        },
        'banking': {
            'level': 'DEBUG',
            'class': 'logging.handlers.RotatingFileHandler',
            'filename': os.path.join(LOG_ROOT, 'banking.log'),
            'formatter': 'extended',
            'maxBytes': ROTATE_LOG_SIZE,
            'backupCount': ROTATE_LOG_COUNT,
        },
    },
    'loggers': {
        'debug': {
            'handlers': ['debug',],
            'level': 'DEBUG',
            'propagate': False,
        },
        'logger': {
            'handlers': ['logger',],
            'level': 'INFO',
            'propagate': False,
        },
        'logger.main': {
            'handlers': ['logger',],
            'level': 'INFO',
            'propagate': False,
        },
        'banking': {
            'handlers': ['banking',],
            'level': 'INFO',
            'propagate': False,
        },
    },
}

log# ls -l --block-size=K ./
total 12172K
-rw-r--r-- 1 www-data www-data  125K Sep 14 13:01 banking.log
-rw-r--r-- 1 www-data www-data  121K Sep 14 13:01 banking.log.1
-rw-r--r-- 1 www-data www-data  285K Sep 14 12:20 banking.log.2
-rw-r--r-- 1 www-data www-data  333K Sep 14 12:17 banking.log.3
-rw-r--r-- 1 www-data www-data  298K Sep 14 10:29 banking.log.4
-rw-r--r-- 1 www-data www-data  315K Sep 14 10:20 banking.log.5

There are other logs in the folder, 4 and 6 mb, there is plenty of space on the disk. As you can see, the files are 125-315kb each, although 10mb is indicated in the config.
The problem was earlier, restarting supervisorctl helped, its config:

uwsgi.conf

[program:workplace-uwsgi]
command = /usr/local/projects/env/bin/uwsgi --ini /usr/local/projects/conf/uwsgi.ini
user = www-data
stdout_logfile = /usr/local/projects/var/log/uwsgi.log
stdout_logfile_maxbytes = 10MB
stderr_logfile = /usr/local/projects/var/log/uwsgi.error.log
stderr_logfile_maxbytes = 10MB
autostart = true
autorestart = true
redirect_stderr = false
priority = 999
stopsignal = QUIT

But a few days later, everything started all over again.

The logs are written actively and a lot, many users request a block (api), the work of which is written to this log. There is a suspicion, but not certainty, that this is the problem (several workers are trying to write to the file at the same time).

Thanks for the help.

Answer 1

Yes, in workers, the concurrent-log-handler solution or similar.