Handle errors and write logs — Python Chapter 8

$ run as a normal user# run as root (or with sudo)

Catch exceptions

try runs code; except handles a specific error; else runs when there was none; finally always runs.

exceptions.py

import json
 
 
def load_config(path: str) -> dict:
    try:
        with open(path, encoding="utf-8") as f:
            return json.load(f)
    except FileNotFoundError:
        print(f"{path} not found, using defaults")
        return {}
    except json.JSONDecodeError as e:
        raise SystemExit(f"{path} is not valid JSON: line {e.lineno}") from e
    finally:
        print("config step done")

Watch out

Never write a bare except: or except Exception: pass — it hides real bugs. Catch only what you can handle.

8.2

Raise your own errors

Stop early with a clear message when input is wrong.

raise.py

def set_replicas(n: int) -> None:
    if not 1 <= n <= 20:
        raise ValueError(f"replicas must be 1-20, got {n}")
    print("ok", n)

8.3

Write logs with logging

The logging module adds time, level and source, and can write to a file and the screen at once.

log_setup.py

import logging
 
logging.basicConfig(
    level=logging.INFO,
    format="%(asctime)s %(levelname)-8s %(name)s: %(message)s",
    handlers=[logging.FileHandler("/var/log/backup.log"), logging.StreamHandler()],
)
log = logging.getLogger("backup")
 
log.info("backup started")
log.warning("disk above %d%%", 80)
try:
    1 / 0
except ZeroDivisionError:
    log.exception("calculation failed")

Level	Use
DEBUG	Details for finding bugs
INFO	Normal steps
WARNING	Something unusual, still working
ERROR	A step failed
CRITICAL	The script can not go on

Remember

log.exception(...) inside except also writes the full traceback.

8.4

Log to the journal

Scripts run by systemd can simply log to stdout/stderr — the journal stores it. Or send to syslog.

syslog_log.py

import logging
import logging.handlers
 
log = logging.getLogger("diskcheck")
log.setLevel(logging.INFO)
log.addHandler(logging.handlers.SysLogHandler(address="/dev/log"))
log.info("diskcheck: / is at 71%")

$journalctl -t python3 -n 5
$journalctl -u diskcheck.service -n 20

8.5

Rotate log files

A long-running script rotates its own log, or leaves it to logrotate.

rotate.py

import logging
from logging.handlers import RotatingFileHandler
 
h = RotatingFileHandler("/var/log/app/app.log", maxBytes=5_000_000, backupCount=5)
logging.getLogger().addHandler(h)

All commands in this chapter

journalctl