2👍
✅
This issue is a bug in billiard, the multiprocessing module fork used by Celery. Here is a link to the issue.
You may confirm this by running python -c "from billiard.connection import Pipe; p = Pipe(duplex=False)"
which should produce the same error.
Cogniva has just pushed a fixed of this to their fork of billiard. At the time of writing this post this fix has been submitted though not merged into celery/billiard. This issue will be resolved in version 3.4.0.0 of billiard.
To install the fix :
$ pip uninstall billiard
$ git clone git@github.com:cogniva/billiard.git
$ cd billiard
$ git checkout bb6169ff1bdd8bc7c5520f7c47085063bc51d4c7
$ python setup.py install
To confirm this fix works run python -c "from billiard.connection import Pipe; p = Pipe(duplex=False)"
. This time you should not see an error message.
Source:stackexchange.com