F23 i686; Old Gateway NetBook; b43; NetworkManager; network; hassles . .

[Philip Rhoades]

Rick,


On 2016-02-25 09:49, Rick Stevens wrote:
> On 02/24/2016 01:53 PM, Philip Rhoades wrote:
>> People,
>> 
>> I have installed the b43 WiFi driver (from memory I think I had 
>> problems
>> with the b43-fwcutter RPM and had to install it manually) but in any
>> case it works generally with this old NetBook both in X and from the
>> console - _IF_ I log in from the netbook first . . if I reboot the
>> NetBook remotely and try and ssh to it when it comes back up, the
>> network isn't accessible.  So I thought I would go through the 
>> exercise
>> of getting rid of NetworkManager and going back to a traditional 
>> network
>> setup and I followed the instructions here:
>> 
>> https://onemoretech.wordpress.com/2014/08/18/disabling-networkmanager-step-by-step
>> 
>> 
>> - which didn't actually work - the problem is, as soon as I do:
>> 
>>    systemctl stop NetworkManager
>> 
>> the little LED WiFi light goes out on the keyboard and it seems there 
>> is
>> nothing I can do to get it on again - the manual slider switch does 
>> not
>> work (I just get kbd unrecognised key errors).  lsmod reports that all
>> the modules are still loaded as for NetworkManager.  As soon as I
>> restart NetworkManager, the WiFi light comes back on - so the question
>> is: what is NM doing to enable the WiFi switch?  I think if I could 
>> find
>> that out, I could get the traditional networking to function . .
>> 
>>  From the attached /var/log/messages, you can see the result of:
>> 
>>    systemctl stop NetworkManager
>> 
>> before the:
>> 
>> "================================================================"
>> 
>> and  the result of:
>> 
>>    systemctl start NetworkManager
>> 
>> after.  You can see the wlan0 deauthenticating and "link is not ready"
>> messages.
>> 
>> Not sure where to go from here - suggestions?
> 
> IIRC, stopping NetworkMangler makes it disable the wireless (and
> bluetooth and mobile broadband if you have it) via rfkill. I think
> that's overstepping its rights a bit (no, actually overstepping its
> rights by a TON), but there's a hell of a lot I dislike intensely about
> systemd and NetworkMangler (especially NetworkMangler's lack of
> documentation).
> 
> Ok, turning off my rant valve, you can walk down the /sys/class/rfkill
> tree, look for the symlink that points at your wireless, and poke the
> "state" file to 1 to turn the radio back on. In my case:
> 
> 	echo 1 >/sys/class/rfkill/rfkill1/state
> 
> I also have a /sys/class/rfkill/rfkill3 symlink, but it points at the
> bluetooth device.


I have:

/sys/class/rfkill/rfkill0/device -> ../../../acer-wmi
/sys/class/rfkill/rfkill1/device -> ../../../acer-wmi
/sys/class/rfkill/rfkill2/device -> ../../phy0

So I am presuming 0 is the WiFi.


> An easier way might be installing the rfkill RPM and using its tools to
> re-enable the wireless. First, see if wifi is disabled
> 
> 	rfkill list wifi


Doing:

   ffkill list

Gives:

0: acer-wireless: Wireless LAN
	Soft blocked: no
	Hard blocked: no
1: acer-bluetooth: Bluetooth
	Soft blocked: yes
	Hard blocked: no
2: phy0: Wireless LAN
	Soft blocked: no
	Hard blocked: no


> If you see "Soft blocked: yes" then that's how it got disabled. "Hard
> blocked: yes" indicates the physical switch is turned off and doing
> the
> soft unblock below won't help.
> 
> To re-enable soft-blocked devices:
> 
> 	rfkill unblock wifi


Doing:

   systemctl stop NetworkManager

turns off the WiFi light but does not change the Soft blocked result.

Making changes to /etc/sysconfig/network-scripts/ifcfg-* as recommended 
and then doing:

   systemctl start network

turns the WiFi light back on (!) but fails with a job error - doing:

   systemctl status network

shows:

   Bringing uup interface Billion_1:  command failed: Network is down 
(-100)
   Failed to start LSB: Bring up/down networking

Doing:

   ifup Billion_1

Gives:

   wlan0 (phy #0): failed to connect, status: 1: Unspecified failure
   RTNETLINK answers: File exists

I seem to be making progress but am still not quite there yet.  Even 
though I have since had pointed out to me a way of getting 
NetworkManager to work reliably . . now I have started this exercise, I 
would like to get the old network method working . .

Thanks,

Phil.
-- 
Philip Rhoades

PO Box 896
Cowra  NSW  2794
Australia
E-mail:  phil at pricom.com.au