Re: periodic network disconnects (reason 'ip-config-unavailable')



----- Original Message -----
From: "Dan Williams" <dcbw redhat com>
To: "Chuck Anderson" <cra WPI EDU>
Cc: networkmanager-list gnome org
Sent: Wednesday, May 1, 2013 10:30:15 PM
Subject: Re: periodic network disconnects (reason 'ip-config-unavailable')

On Wed, 2013-05-01 at 16:17 -0400, Chuck Anderson wrote:
On Mon, Apr 29, 2013 at 11:07:51AM -0500, Dan Williams wrote:
On Sat, 2013-04-27 at 13:56 -0400, Chuck Anderson wrote:
On Fedora 18, I'm seeing periodic network disconnects with these
messages in the log:

Apr 27 11:01:11 a NetworkManager[833]: <info> (wlan0): device state
change: activated -> failed (reason 'ip-config-unavailable') [100 120
5]
Apr 27 11:01:11 a NetworkManager[833]: <warn> Activation (wlan0) failed
for connection 'Sanctuary 2.4GHz'
Apr 27 11:01:11 a NetworkManager[833]: <info> (wlan0): device state
change: failed -> disconnected (reason 'none') [120 30 0]
Apr 27 11:01:11 a NetworkManager[833]: <info> (wlan0): deactivating
device (reason 'none') [0]
Apr 27 11:01:11 a dbus-daemon[736]: dbus[736]: [system] Activating
service name='org.freedesktop.nm_dispatcher' (using servicehelper)

What does reason 'ip-config-unavailable' mean and how can I
troubleshoot what is happening here?

I do have IPv6 on this network and found this bug report, but that
appears to be for an older version and should be fixed in the version
in Fedora 18, NetworkManager-0.9.8.1-1.git20130327.fc18.x86_64:

What are the messages just before that?  Is it IPv6 that's timing out?
There should be some log messages about that.  You can also run NM with
verbose IPv6 logging which might tell us more about what's going on.  So
for now I'll suspect IPv6, and the issues there are that the router may
not send advertisements quickly enough to prevent timed-out DNS servers.

sudo dbus-send --system --print-reply
--dest=org.freedesktop.NetworkManager /org/freedesktop/NetworkManager
org.freedesktop.NetworkManager.SetLogging string:"debug" string:""

grab the syslog output from that during a failure, then we'll see what
the actual issue is and how to deal with it.

This was caused by my router's radvd configuration being non-RFC
compliant.  I was troubleshooting my Android device's IPv6
connectivity so I had temporarily lowered the AdvDefaultLifetime,
AdvValidLifetime and AdvPreferredLifetime on radvd.conf.
Unfortunately, radvd lets you set these values to ones that violate
contraints required by RFC4861 (MUST be higher than the
MaxRtrAdvInterval).

After fixing radvd.conf, NetworkManager is now behaving correctly.

Ah, that explains it.  Good to know, thanks for the follow-up.  It might
be a good idea to make NM complain (once) in the logs if this is the
case for a particular network.

The problem is that MaxRtrAdvInterval is not published over the network. But I thought radvd was complaining 
and wouldn't start. If it is a recent version of radvd, it might deserve fixing. We'll be reworking the IPv6 
autoconfiguration using the libndp library and we can turn some of the messages to warnings/errors.

Cheers,

Pavel


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]