Closed pevogam closed 10 months ago
Do you have a simple reproducer for this issue? I remember similar issues on cleanup in another project when the process was interrupted and gc was already cleaning the objects that were in the process of manual __del__
execution (and were incomplete)
I think the reason is in not closing the sessions manually and using a _session_cache
class attribute, then ending the script. I would have expected that the default __del__
of the sessions would properly close them but perhaps I am wrong and using close()
explicitly is actually needed. What would solve the problem on my side is if I added this (ignore the private access for illustration purposes) to the end of the python script:
for session in pool.TransferOps._session_cache.values():
session.close()
in order to purge the cache of sessions and close each previously created one.
Since this problem no longer appears in any of my use I am closing this. Perhaps what we have documented so far might be useful if we encounter anything similar in the future.
This is an odd issue since
open
must be replaced or somehow modified but during deletion of aSpawn
instance I keep gettingI haven't seen this problem before so it might be recent, not sure if anyone else has seen this.