Resetting Bearer on OpenWRT

[Aleksander Morgado]

> > > I'm running modemmanager on OpenWRT and am getting this error fairly
> > > regularly such as when I use AT commands to manually change LTE
> > > bands(is there a better way to change/lock bands than using AT
> > > commands?): "Bearer not allowed to connect, not registered in 3GPP
> > > network"
> > >
> >
> > If you're using ModemManager, you can also use mmcli directly, e.g.
> > "mmcli -m 0 --set-current-bands="eutran-2|eutran-3|eutran-4"
> > The list of supported bands should be given in the "mmcli -m 0"
> > output. Changing bands, of course, will trigger a network
> > unregistration and re-registration, and so the device will get
> > unregistered in the process.
> >
> > > Doing "ifup broadband" resolves the issue(at least until it loses the
> > > bearer again) since resetting the interface triggers
> > > proto_modemmanager_setup() to flush and reset the bearer but I don't
> > > see a way to automatically get modemmanager state change notifications
> > > to OpenWRT so that I can trigger the bearer reload directly from the
> > > modemmanger state change event. I assume that's because modemmanager
> > > uses dbus and OpenWRT uses ubus right? I've used a cron job as a
> > > workaround to monitor the bearer and issue an "ifup broadband" when
> > > it's lost.
> > >
> >
> > One thing you need to consider is that, if I'm not mistaken, the
> > netifd integration done in OpenWRT does *not* consider automatic
> > reconnections. So if your modem gets disconnected due to any reason
> > (e.g. unregistered from the network because you changed bands in use)
> > then you'll need to manually trigger a reconnection (e.g. with "ifup
> > broadband" as you were doing). I'm sure this can be changed somehow to
> > autoreconnect, but didn't spend much time on that after doing the
> > initial MM integration in OpenWRT.
> >
> > The script you're using, where you just look for the bearer state in
> > MM and trigger a manual interface up event in netifd if the bearer is
> > disconnected, is one way to have the automatic reconnection logic
> > working, yes :)
> How difficult would it to be to add minimal ubus support for triggering
> automatic re-connections as a compile time option?
> >

Absolutely no idea, because I don't know how netifd expects that to be
reported or how it does connection monitoring. Any info on how that
works would be appreciated.

-- 
Aleksander
https://aleksander.es