conntrack Better handle clashes
ghananigans opened this issue · comments
Currently, we always insert the original direction's tuple and make a best-effort to insert the reply-tuple.
(linux always inserts the reply and makes a best effort for the original).
We should investigate improving our approach to clashes.
should we drop packets for connections that conflict with an existing tuple?
A friendly reminder that this issue had no activity for 120 days.
This issue has been closed due to lack of activity.