In order to prevent data corruption, if an oplock/lease break request sent from the smb server to the client does not get a response, (if the channel failed) the server needs to resend the break request over other channels as long as there are intact channels left in the (multi-channel) session. This is required in order to move the multichannel feature from experimental to supported. The purpose of this bug is to track this piece of technical debt.
*** Bug 12376 has been marked as a duplicate of this bug. ***
This is fixed with 883657a7cfe0788b8c82b58a0379a8ea724df72a in 4.13.0rc1
(In reply to Stefan Metzmacher from comment #2) > This is fixed with 883657a7cfe0788b8c82b58a0379a8ea724df72a in 4.13.0rc1 Thanks a lot for fixing this, Metze! Unless I'm missing something, we could now consider removing, or at least toning down, the "this feature is still considered experimental" warning in the manpage. What do you think?
(In reply to David Disseldorp from comment #3) We still have these before we can thing about it: https://bugzilla.samba.org/show_bug.cgi?id=11899 https://bugzilla.samba.org/show_bug.cgi?id=14433 https://bugzilla.samba.org/show_bug.cgi?id=14449 I hope to get to them for 4.14. We also need people doing real world testing with real hardware and windows clients. See also https://www.samba.org/~metze/presentations/2020/SDC/ for my status update