This builds upon the answer by Philippe while using dictConfig
. The contextual filter demonstrated in this answer uses psutil
to log the current CPU and memory usage percentage in each log message.
Save this file in say mypackage/util/logging.py
:
"""logging utiliies."""
import logging
from psutil import cpu_percent, virtual_memory
class PsutilFilter(logging.Filter):
"""psutil logging filter."""
def filter(self, record: logging.LogRecord) -> bool:
"""Add contextual information about the currently used CPU and virtual memory percentages into the given log record."""
record.psutil = f"c{cpu_percent():02.0f}m{virtual_memory().percent:02.0f}" # type: ignore
return True
Note that a filter function didn't work for me; only a filter class worked.
Next, update your logging config dict based on this answer as below:
LOGGING_CONFIG = {
...,
"filters": {"psutil": {"()": "mypackage.util.logging.PsutilFilter"}},
"handlers": {"console": {..., "filters": ["psutil"]}},
"formatters": {
"detailed": {
"format": "%(asctime)s %(levelname)s %(psutil)s %(process)x:%(threadName)s:%(name)s:%(lineno)d:%(funcName)s: %(message)s"
}
},
}
Try logging something, and see sample output such as:
2020-05-16 01:06:08,973 INFO c68m51 3c:MainThread:mypackage.mymodule:27:myfunction: This is my log message.
In the above message, c68m51
means 68% CPU and 51% memory usage.