Python logging module does not work from engines

[slishak]

Using a custom handler in the logging module can cause an engine to hang indefinitely. I think this behaviour has not always existed, as it only started happening when I updated to the latest releases of IPython/Jupyter/ipyparallel. I believe it could be related to jupyter/notebook#1397.

If I have module loggingtest.py as follows:

import logging
import sys

hPrint = logging.StreamHandler(sys.stdout)
logger = logging.Logger('Logger')
logger.addHandler(hPrint)  # commenting out this line fixes the bug
logger.log(1, 'Log initialised')  # engine will hang on this line

def log(message):
    logger.log(1, message)

And then I start a cluster using ipcluster start -n 2 and connect to it using c = ipyparallel.Client(), importing that module will hang indefinitely with either of the following methods:

out = c[0].execute('import loggingtest')
out.wait(timeout=5)

import loggingtest
out = c[1].apply_async(loggingtest.log, 'test')
out.wait(timeout=5)

I'm using Windows 7, Python 2.7.10 32-bit, all IPython modules up to date:

ipykernel (4.3.1)
ipyparallel (5.1.1)
ipython (4.2.0)
ipython-genutils (0.1.0)
ipywidgets (5.1.5)
jupyter (1.0.0)
jupyter-client (4.3.0)
jupyter-console (4.1.1)
jupyter-core (4.1.0)
nbconvert (4.2.0)
nbformat (4.0.1)
notebook (4.2.1)

[slishak]

Probably also related to https://github.com/jupyter/tmp_for_org_transfer/issues/1

[minrk]

Hi! I’m going through and cleaning up old/stale issues on this repo. It looks like this issue is resolved (running the same does not reproduce the error with current versions). Feel free to open a new Issue if you are still having this trouble.