[Pkg-utopia-maintainers] Bug#755568: upgrade to network-manager(-gnome?) in jessie breaks wireless connectivity

[Jim McCloskey]

Package: network-manager
Version: 0.9.10.0-1
Severity: important


It's actually not clear to me whether this bug-report should be filed
against network-manager or against network-manager-gnome.

I upgraded my laptop to testing on Monday July 21st:

  Start-Date: 2014-07-21  09:38:16
  Commandline: apt-get -t testing dist-upgrade

Among the packages affected were network-manager and network-manager-gnome:

    network-manager:amd64 (0.9.8.10-4,  0.9.10.0-1)
    network-manager-gnome:amd64 (0.9.8.10-1, 0.9.10.0-2)

Before the upgrade, my network connections worked really unusually
well (better than my housemates using OsX).  Following the upgrade,
though, I saw the following problematic behavior (on my home network):
Upon reboot, the laptop established a wireless link with the access
point and all seemed normal. After roughly one minute's use, however,
I lost the connection, with no notification from the system and in the
following sense: addresses on the local network remained accessible,
but any attempt to connect beyond the local network (ssh, http ... )
resulted in `no route to host' messages.  I repeated this pattern (by
rebooting) five times.

The problem is not with the network configuration here -- all other
machines (OsX and debian systems that had not been upgraded)
continue to work as expected and as before. Only the system which had
undergone this upgrade was affected.

I forced a downgrade to the version of network-manager in wheezy
(0.9.4.0-10), which in turn forced the removal of
network-manager-gnome, and the problem disappeared. I upgraded again to
version 0.9.10.0-1 (current testing) and the problem returned.

At present, the system reports (through the network icon) that the
laptop is connected; ifconfig shows that wlan0 has an IP address, but
all outbound connections are impossible. It's as if the default route
cannot be configured (or that that information is somehow lost after
the initial connection).

Here is a relevant part of syslog after the most recent such failure:

--------
Jul 21 17:10:21 lapdog4 dhclient: DHCPDISCOVER on wlan0 to 255.255.255.255 port 67 interval 7
Jul 21 17:10:21 lapdog4 NetworkManager[4634]: DHCPDISCOVER on wlan0 to 255.255.255.255 port 67 interval 7
Jul 21 17:10:21 lapdog4 dhclient: DHCPREQUEST on wlan0 to 255.255.255.255 port 67
Jul 21 17:10:21 lapdog4 dhclient: DHCPOFFER from 192.168.1.1
Jul 21 17:10:21 lapdog4 NetworkManager[4634]: DHCPREQUEST on wlan0 to 255.255.255.255 port 67
Jul 21 17:10:21 lapdog4 NetworkManager[4634]: DHCPOFFER from 192.168.1.1
Jul 21 17:10:21 lapdog4 dhclient: DHCPACK from 192.168.1.1
Jul 21 17:10:21 lapdog4 NetworkManager[4634]: DHCPACK from 192.168.1.1
Jul 21 17:10:21 lapdog4 dhclient: bound to 192.168.1.108 -- renewal in 35719 seconds.
Jul 21 17:10:21 lapdog4 NetworkManager[4634]: <info> (wlan0): DHCPv4 state changed preinit -> bound
Jul 21 17:10:21 lapdog4 NetworkManager[4634]: <info>   address 192.168.1.108
Jul 21 17:10:21 lapdog4 NetworkManager[4634]: <info>   plen 24 (255.255.255.0)
Jul 21 17:10:21 lapdog4 NetworkManager[4634]: <info>   gateway 192.168.1.1
Jul 21 17:10:21 lapdog4 NetworkManager[4634]: <info>   server identifier 192.168.1.1
Jul 21 17:10:21 lapdog4 NetworkManager[4634]: <info>   lease time 86400
Jul 21 17:10:21 lapdog4 NetworkManager[4634]: <info>   nameserver '192.168.1.1'
Jul 21 17:10:21 lapdog4 NetworkManager[4634]: <info> Activation (wlan0) Stage 5 of 5 (IPv4 Configure Commit) scheduled...
Jul 21 17:10:21 lapdog4 NetworkManager[4634]: <info> Activation (wlan0) Stage 5 of 5 (IPv4 Commit) started...
Jul 21 17:10:21 lapdog4 avahi-daemon[2603]: Joining mDNS multicast group on interface wlan0.IPv4 with address 192.168.1.108.
Jul 21 17:10:21 lapdog4 avahi-daemon[2603]: New relevant interface wlan0.IPv4 for mDNS.
Jul 21 17:10:21 lapdog4 avahi-daemon[2603]: Registering new address record for 192.168.1.108 on wlan0.IPv4.
Jul 21 17:10:21 lapdog4 NetworkManager[4634]: <info> (wlan0): device state change: ip-config -> ip-check (reason 'none') [70 80 0]
Jul 21 17:10:21 lapdog4 NetworkManager[4634]: <info> Activation (wlan0) Stage 5 of 5 (IPv4 Commit) complete.
Jul 21 17:10:21 lapdog4 NetworkManager[4634]: bound to 192.168.1.108 -- renewal in 35719 seconds.
Jul 21 17:10:21 lapdog4 NetworkManager[4634]: <info> (wlan0): device state change: ip-check -> secondaries (reason 'none') [80 90 0]
Jul 21 17:10:21 lapdog4 NetworkManager[4634]: <info> (wlan0): device state change: secondaries -> activated (reason 'none') [90 100 0]
Jul 21 17:10:21 lapdog4 NetworkManager[4634]: <info> NetworkManager state is now CONNECTED_LOCAL
Jul 21 17:10:21 lapdog4 NetworkManager[4634]: <info> NetworkManager state is now CONNECTED_GLOBAL
Jul 21 17:10:21 lapdog4 NetworkManager[4634]: <info> Policy set 'branci40 2' (wlan0) as default for IPv4 routing and DNS.
Jul 21 17:10:21 lapdog4 NetworkManager[4634]: <info> Activation (wlan0) successful, device activated.
Jul 21 17:10:21 lapdog4 NetworkManager[4634]: <info> startup complete
Jul 21 17:10:23 lapdog4 ntpd[2686]: Listen normally on 10 wlan0 192.168.1.108 UDP 123
Jul 21 17:10:23 lapdog4 ntpd[2686]: peers refreshed
Jul 21 17:10:23 lapdog4 nm-dispatcher: Script '/etc/NetworkManager/dispatcher.d/01ifupdown' exited with error status 1.
Jul 21 17:10:23 lapdog4 nm-dispatcher: Dispatching action 'up' for wlan0
Jul 21 17:10:23 lapdog4 NetworkManager[4634]: <warn> (2) 01ifupdown failed (failed): Script '/etc/NetworkManager/dispatcher.d/01ifupdown' exited with error status 1.

-------

The other experiment I've been running is this: network-manager on the
laptop is currently at the version in testing (0.9.10.0-1), but I
purged network-manager-gnome almost exactly three hours ago. The
problem has not manifested itself in those three hours,

Jim