As suggested by Guido I am posting a potential solution to the following
problem for review.
Problem <https://bugs.python.org/issue38119#msg351960>
A quick fix for this problem is to prevent resource tracker from handling
destruction/unlinking of shared memory. A PR
<https://github.com/python/cpython/pull/15989> for the same was opened by Davin
(@applio), and at this moment it is the only implemented solution to fix some
of the critical issues in shared memory.
But, as soon as resource tracker is removed from the picture cleanup of shared
memory is dependent on the user itself and this also makes Linux’s shared
memory behaviour inconsistent with windows.
Windows uses a reference counting mechanism to count the number of processes
using a shared memory segment. When all of them are done using it, Windows
simply unlinks and frees the memory allocated to the shared memory segment. On
the other hand, in Linux, even after all the processes are done using the
shared memory segment, the segment will continue to exist unless it is cleaned
by the respective processes.
Having said that, the current implementation of Linux’s shared memory is also
not consistent with windows. So, I think it makes sense to go ahead with the
removal of resource tracker for now until we reach a better solution.
On that note, I have mentioned some of the possible solutions here
<https://bugs.python.org/msg351445>. Out of these solutions I think 5 is the
best. This issue and solution 5 is currently being discussed here
<https://bugs.python.org/issue38119>.
_______________________________________________
Python-Dev mailing list -- python-dev@python.org
To unsubscribe send an email to python-dev-le...@python.org
https://mail.python.org/mailman3/lists/python-dev.python.org/
Message archived at
https://mail.python.org/archives/list/python-dev@python.org/message/O67CR7QWEOJ7WDAJEBKSY74NQ2C4W3AI/
Code of Conduct: http://python.org/psf/codeofconduct/
