Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]


Groups > linux.debian.user > #176002 > unrolled thread

Stretch and WiFi ALMOST working

Started byMark Fletcher <mark27q1@gmail.com>
First post2016-12-28 12:00 +0100
Last post2016-12-31 04:10 +0100
Articles 3 — 2 participants

Back to article view | Back to linux.debian.user


Contents

  Stretch and WiFi ALMOST working Mark Fletcher <mark27q1@gmail.com> - 2016-12-28 12:00 +0100
    Re: Stretch and WiFi ALMOST working Brian <ad44@cityscape.co.uk> - 2016-12-29 17:50 +0100
      Re: Stretch and WiFi ALMOST working Mark Fletcher <mark27q1@gmail.com> - 2016-12-31 04:10 +0100

#176002 — Stretch and WiFi ALMOST working

FromMark Fletcher <mark27q1@gmail.com>
Date2016-12-28 12:00 +0100
SubjectStretch and WiFi ALMOST working
Message-ID<sTk7E-73i-7@gated-at.bofh.it>
Hello

Some of you may remember this thread I started earlier this month:

https://lists.debian.org/debian-user/2016/12/msg00130.html

in which I asked for opinions on the best way to connect a stretch box 
using WiFi. After being educated by the various responses, I went back 
and took another look at the box in question and found that not only was 
NetworkManager installed, but it was being used for the existing 
connections. So, contrary to what I said towards the end of that thread, 
I decided to just adjust the NetworkManager config to get the network to 
work the way I wanted.

After some jiggering about I eventually figured out that I had multiple 
instances of the connection setup to both my main WiFi AP and to the 
extension AP I had added. Some of those were limited to specific users 
while others were open to all users. I binned the specific ones and kept 
the ones that were open to all users, one for each connection. By 
including the WiFi PSK for each connection (one for the main AP and one 
for the extension) in the relevant connection config and changing the 
psk-flags from 1 to 0, I had connections happening automatically at boot 
instead of waiting for a login. And, by adding to the config files a 
line saying autoconnect-priority=X I was able to get the machine to 
favour the extension and fall back to the main AP if the extension is 
not available. All very nice.

However, I now discover that Avahi is not doing its thing properly any 
more, and other machines on my network cannot see this machine by name. 
Its IP address right now is 192.168.11.13 and its name is affinity. I 
have another machine on my network, a jessie box, which is called kazuki 
and happens to be at IP 192.168.11.4 right now. From kazuki I can ping 
affinity by IP address, and likewise from affinity I can ping kazuki by 
IP address, but ping kazuki.local from affinity and ping affinity.local 
from kazuki both fail with "ping: unknown host X" where X is the machine 
I am trying to ping, after a pause that suggests some sort of attempt to 
find the other machine is being made.

Before I made the NetworkManager changes on affinity, this would happen 
sometimes, but if I "woke up" affinity by pinging it by IP address, I 
could then get the name to resolve. Now I can never get the name to 
resolve.

Anyone got any bright ideas how to figure out what is going on here? 
Looking in the systemd journal I can't see anything being written when 
trying to do the lookups. Avahi daemon is starting, healthily as far as 
I can tell, on both machines. Other machines can look up kazuki on the 
network by name without problems, eg my wife's iPad which uses it as a 
print server. So the problem most likely is at the affinity end, but I 
need help to diagnose it.

TIA

Mark

[toc] | [next] | [standalone]


#176045

FromBrian <ad44@cityscape.co.uk>
Date2016-12-29 17:50 +0100
Message-ID<sTM3T-AQ-9@gated-at.bofh.it>
In reply to#176002
On Wed 28 Dec 2016 at 19:54:34 +0900, Mark Fletcher wrote:

> However, I now discover that Avahi is not doing its thing properly any 
> more, and other machines on my network cannot see this machine by name. 
> Its IP address right now is 192.168.11.13 and its name is affinity. I 
> have another machine on my network, a jessie box, which is called kazuki 
> and happens to be at IP 192.168.11.4 right now. From kazuki I can ping 
> affinity by IP address, and likewise from affinity I can ping kazuki by 
> IP address, but ping kazuki.local from affinity and ping affinity.local 
> from kazuki both fail with "ping: unknown host X" where X is the machine 
> I am trying to ping, after a pause that suggests some sort of attempt to 
> find the other machine is being made.

You do have libnss-mdns installed and the line

  hosts:	files mdns4_minimal [NOTFOUND=return] dns

in /etc/nsswitch?

-- 
Brian.

[toc] | [prev] | [next] | [standalone]


#176135

FromMark Fletcher <mark27q1@gmail.com>
Date2016-12-31 04:10 +0100
Message-ID<sUids-5Ie-15@gated-at.bofh.it>
In reply to#176045
On Thu, Dec 29, 2016 at 04:46:22PM +0000, Brian wrote:
> On Wed 28 Dec 2016 at 19:54:34 +0900, Mark Fletcher wrote:
> 
> > However, I now discover that Avahi is not doing its thing properly any 
> > more, and other machines on my network cannot see this machine by name. 
> > Its IP address right now is 192.168.11.13 and its name is affinity. I 
> > have another machine on my network, a jessie box, which is called kazuki 
> > and happens to be at IP 192.168.11.4 right now. From kazuki I can ping 
> > affinity by IP address, and likewise from affinity I can ping kazuki by 
> > IP address, but ping kazuki.local from affinity and ping affinity.local 
> > from kazuki both fail with "ping: unknown host X" where X is the machine 
> > I am trying to ping, after a pause that suggests some sort of attempt to 
> > find the other machine is being made.
> 
> You do have libnss-mdns installed and the line
> 
>   hosts:	files mdns4_minimal [NOTFOUND=return] dns
> 
> in /etc/nsswitch?
> 

Thanks Brian. I have got to think the answer is YES since it *was* 
previously working, and even now works (I discovered after my original 
post) immediately after booting or after restarting NetworkManager, but 
then stops working after that.

I will check facts later when the machine is up and verify. 

Some developments on this. I am noticing that, even when using the WiFi 
extension, the WiFi interface on this box seems to "go to sleep" if left 
idle for a few minutes. So when the machine first comes up, all is 
working as it should, but if left idle for a while, suddenly I can't 
ping the machine any more. If I ping it from another machine on my 
network and leave ping running, after about 30 seconds or so the pings 
start being answered. But from that time I can no longer find the 
machine by name (ping affinity.local will fail to resolve 
affinity.local, ping <ip address> will work).

So Avahi not working could be a symptom rather than a cause, if you see 
what I mean. Or rather, be a secondary issue caused by the real issue, 
which is what I should be solving.

The WiFi interface is a Realtek RTL8723BE device, and the firmware is 
being successfully found and loaded on boot (Checked in journalctl -b). 
Googling around I have found some modprobe options that are suggested 
for this card, so I am going to try those and see if that makes any 
difference. But I'm startuing to think it is a 
stability-of-the-network-interface issue rather than an Avahi or an 
out-on-the-edge-of-WiFi-range issue as I thought it was originally.

Mark

[toc] | [prev] | [standalone]


Back to top | Article view | linux.debian.user


csiph-web