You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
When listing directories, it looks like connection pooling isn't being used.
Since I am listing a lot of directories, I am cycling through the ephemeral port list around 20 times, before doing anything else.
I would probably significantly reduce time and load on both ends, if I was able to reuse a single connection, instead of spawning 200000.
I believe the issue why connection pooling isn't being used automatically (as advertised by requests/urllib3) is that the connections created are always streaming, even if it's just a PROPFIND, which doesn't really benefit from the streaming.
My proposed solution would be to make the streaming flag in the request conditional upon the method being called. stream=True for actual up/downloads, and stream=False for list
The text was updated successfully, but these errors were encountered:
When listing directories, it looks like connection pooling isn't being used.
Since I am listing a lot of directories, I am cycling through the ephemeral port list around 20 times, before doing anything else.
I would probably significantly reduce time and load on both ends, if I was able to reuse a single connection, instead of spawning 200000.
I believe the issue why connection pooling isn't being used automatically (as advertised by requests/urllib3) is that the connections created are always streaming, even if it's just a PROPFIND, which doesn't really benefit from the streaming.
My proposed solution would be to make the streaming flag in the request conditional upon the method being called.
stream=True
for actual up/downloads, andstream=False
forlist
The text was updated successfully, but these errors were encountered: