Package: network-manager-openvpn Version: 0.9.10.0-1 Severity: grave Hi,
I've been putting of reporting that nm-openvpn in Jessie fails to tear down the tap0 interface when disconnecting from a VPN...well, it's just as well that I put it off, because it turns out that it is a more severe issue than I had imagined. It produces 390G of the following line in /var/log/daemon.log ! : Jun 30 16:27:22 hal9000 nm-openvpn[16077]: read from TUN/TAP : File descriptor in bad state (code=77) Jun 30 16:27:22 hal9000 nm-openvpn[16077]: read from TUN/TAP : File descriptor in bad state (code=77) Jun 30 16:27:22 hal9000 nm-openvpn[16077]: read from TUN/TAP : File descriptor in bad state (code=77) Jun 30 16:27:22 hal9000 nm-openvpn[16077]: read from TUN/TAP : File descriptor in bad state (code=77) Jun 30 16:27:22 hal9000 nm-openvpn[16077]: read from TUN/TAP : File descriptor in bad state (code=77) ... Given that less that a second seems to have elapsed between log entries, it look like there are three problems. 1) nm-openvpn fails to tear down tap0 when disconnecting from VPN 2) there is a race condition somewhere, leading to effect 3) all available disk space is consumed. I have not read more than five minutes of paging through /var/log/daemon.log, but I expect that the time stamp increments at some point. I also expect that this bug is fixed in a newer version of network-manager-openvpn and that the fix can be cherry-picked and backported to 0.9.10.0. Sincerely, Nicholas
