Re: NM vs. suspend/resume

[Dan Williams <dcbw redhat com>]

On Fri, 2010-09-24 at 09:37 -0400, Neal Becker wrote:
> Jirka Klimes wrote:
> 
> > On Friday 24 of September 2010 13:00:58 Neal Becker wrote:
> >> NetworkManager-0.8.1-6.git20100831.fc13.x86_64
> >> knetworkmanager-0.9-0.20.20100603.fc13.x86_64
> >> 
> >> When resumed, NM does not notice wired eth0 if when suspended it was
> >> running on wifi.  Restarting NM fixes it.
> >> 
> >> Or perhaps the problem is knetworkmanager not noticing the change?
> >> 
> > 
> > Could you grab NetworkManager logs (var/log/messages)?
> > Do you see the device in nm-tool output?
> > 
> > Jirka
> 
> See attached.  nm says nothing about eth0 when resumed.  Then nm is 
> restarted and everything works.

So that log indicates that NM isn't getting the suspend/resume signals
at all, actually.  Unless that log doesn't show NM actually being
suspended and resumed?

NM >= 20100831 has two mechanisms for suspend: pm-utils (which has
historically had these types of issues) and listening to upower signals.

I spent a bunch of yesterday testing wifi suspend/resume on an
up-to-date F13 and could  not reproduce these issues.  I'll next test
wired suspend/resume.  However, if you don't see suspend/resume messages
in /var/log/messages, then it's likely a problem with communication
between pm-utils and/or upower.

Dan