Python logging not printing to console or file (tested)

Python logging that prints nothing, or never creates its log file, almost always comes down to one of five things. We measured each on Python 3.13.5.

1. Nothing configured: only WARNING and above show

import logging
log = logging.getLogger("app")
log.debug("debug message")
log.info("info message")
log.warning("warning message")
print("effective level:", logging.getLevelName(log.getEffectiveLevel()))

Output:

warning message
effective level: WARNING

Measured on Python 3.13.5: with no configuration, the effective level is WARNING, so debug() and info() print nothing. The warning that does appear comes from logging's built-in "last resort" handler (bare message, no level prefix, on stderr). Call logging.basicConfig(level=logging.INFO) once at startup.

2. basicConfig() silently does nothing the second time

import logging
logging.warning("an import or library logged first")   # implicitly calls basicConfig()
logging.basicConfig(level=logging.DEBUG, format="%(levelname)s %(message)s")   # silently ignored
logging.getLogger("app").info("info message")
logging.basicConfig(level=logging.DEBUG, format="%(levelname)s %(message)s", force=True)  # fix
logging.getLogger("app").info("info message after force=True")

Output:

WARNING:root:an import or library logged first
INFO info message after force=True

Measured: any call like logging.warning() on the root logger (often inside an imported library) runs basicConfig() implicitly. After that, your own basicConfig(level=DEBUG, ...) is a no-op with no error, so your level and format never apply. force=True (3.8+) removes the existing handlers and applies yours.

3. Setting the handler's level isn't enough

import logging, sys
log = logging.getLogger("app")
h = logging.StreamHandler(sys.stdout)
h.setLevel(logging.DEBUG)
log.addHandler(h)
log.info("handler is DEBUG, logger is not set")      # dropped: logger level wins first
log.setLevel(logging.DEBUG)
log.info("now the logger is DEBUG too")

Output:

now the logger is DEBUG too

Measured: a record must pass the logger's level before any handler sees it. The handler was DEBUG but the logger was still at the inherited WARNING, so the first info() was dropped. Set both: log.setLevel(logging.DEBUG).

4. Not writing to a file

import logging, os
os.makedirs("/tmp/lg/run", exist_ok=True); os.chdir("/tmp/lg/run")
logging.warning("something logged before setup")
logging.basicConfig(filename="app.log", level=logging.INFO)    # ignored: root already has a handler
logging.info("should go to app.log")
print("app.log exists:", os.path.exists("app.log"))
logging.basicConfig(filename="app.log", level=logging.INFO, force=True)
logging.info("goes to app.log")
logging.shutdown()
print("app.log:", open("app.log").read().strip())
print("relative to cwd:", os.path.abspath("app.log"))

Output:

WARNING:root:something logged before setup
app.log exists: False
app.log: INFO:root:goes to app.log
relative to cwd: /tmp/lg/run/app.log

Measured: basicConfig(filename="app.log") after an earlier log call creates no file at all, for the same reason as section 2; force=True fixes it. A relative filename is resolved against the current working directory, not the script's folder, so under cron, systemd or an IDE the file may be somewhere you aren't looking. Use an absolute path. Output is written as each record is logged; logging.shutdown() (run automatically at exit) closes the file.

5. The opposite: every message printed twice

import logging, sys
logging.basicConfig(level=logging.INFO, stream=sys.stdout, format="%(name)s: %(message)s")
log = logging.getLogger("app")
log.addHandler(logging.StreamHandler(sys.stdout))   # one more handler
log.info("printed twice")
log.propagate = False
log.info("printed once")

Output:

printed twice
app: printed twice
printed once

Measured: a handler on app plus a handler on the root logger prints each record twice, because records propagate up to the root. Add handlers in one place only, or set propagate = False on the child logger.

Written with AI assistance. Measured on 2026-10-09 with Python 3.13.5 on Linux: the default WARNING level and last-resort handler, basicConfig ignored after an implicit call and force=True, logger vs handler levels, a file handler that never creates its file, relative log paths, and duplicate output from propagation. Every block was run verbatim and its output pasted from that run (stdout and stderr together). PlainHeaders is our extension; everything else here is the standard library.