tensorboard

• file_level = cfg_logging.get("file_level", 10)
console_level = cfg_logging.get("console_level", 20): retrieve file level and console level
  • when set level = n, only show the information’s level that higher than this n. For example if set level=20, will not print DEBUG information

Level Name	Value	Description
DEBUG	10	Most detailed logs, used for debugging
INFO	20	General runtime information
WARNING	30	Warning, but the program can still run
ERROR	40	Error, part of the program failed
CRITICAL	50	Severe error, the program may crash

• root_logger = logging.getLogger() : the logging top level controller
• root_logger.handlers.clear() : clear all handlers, for example: FileHandler, StreamHandler, avoid repetitive log
• root_logger.setLevel(min(file_level, console_level)): set the minimum for root logger, anything lower than root logger will be entirely ignored, even though it been set up in handlers.

set up handler:

• what handler: logging.FileHandler(logging_file_path)
• what format of this handler: xx_handler.setFormatter(formatter_object)
• what level of this handler: xx_handler.setLevel(level_number)
• add this handler to root logger: root_logger.addHandler(xx_handler)

handler:

• write into file: logging.FileHandler()
• write to terminal: logging.StreamHandler()

other loggers:

• get root logger: logging.getLogger()
• get specific logger: logging.getLogger("package_name"), for example: logging.getLogger("matplotlib")
• set level for specific logger: logging.getLogger("matplotlib").setLevel(logging.INFO)

  logging.INFO -> number 20