Ever since updating from kde4 to plasma on my experimental machine, the NetworkManager-dispatcher.service has stopped detecting the "pre-down" state and skips to "down" state instead. This indicates a forcible disconnection and results in any CIFS shares left unmounted, hanging the shutdown process for 90 seconds. Reproducible: Always Steps to Reproduce: 1. configure NetworkManager-dispatcher.service [1] 2. put a script into /etc/NetworkManager/dispatcher.d/pre-down.d and configure permission accordingly [2] 3. manually disconnect from the network 4. restart 5. stay connected to the network 6. restart Actual Results: second shutdown will hang at "unmounting <your cifs share>" Expected Results: the shares should have been unmounted on both restarts At first I suspected SDDM, but I have now tried it with LXDM as well and it shows the same behavior. The pre-down state sometimes does trigger when logging out of plasma instead of restart/shutdown, but never on restart/shutdown. With plasma's shutdown scripts not working, I'm out of options to automatically unmount any shares before disconnecting. [1] https://wiki.archlinux.org/index.php/NetworkManager#Network_services_with_NetworkManager_dispatcher [2] https://wiki.archlinux.org/index.php/NetworkManager#Use_dispatcher_to_handle_mounting_of_CIFS_shares
Actually, it doesn't seem like the 'down' action is executed at all. I'm guessing that means NetworkManager is just killed instead of letting it shutdown correctly.
That's a system service, we (Plasma) don't manipulate when they go up/down at all. I'm afraid that's entirely the job of your init system (initd, systemd, upstart). Hope that helps. Sorry.