Odd network behavior

[rikona]

I'm seeing some odd network behavior that I don't understand. The
problem started when I decided to improve the wireless coverage in my
house by using an extra router as second access point. Both routers
have the same SSID and the same encryption and password. The second
router has different LAN IP address, a different channel[using 5 and
9], and the DHCP server is turned off in the second router/access
point. This second router/access point does work and does extend the
wireless coverage significantly, so that function seems to be working.

But, I started noticing on my Ubuntu box that I was periodically losing
my network connection about every 5 to 20 minutes. The connection would
go down for a couple of minutes then come back up. When the connection
went down, all the lights next to each CAT5 connector would go off
completely [which seemed odd], not even flickering. After a minute or
two the input CAT5 connector light would go on, followed shortly by the
rest of the CAT5 connector lights, and the network would be up.

At first I was blaming Comcast because their local service is not at
all reliable, but after a day or so I began to consider that the
problem might be at my end. I have another router [primarily used by
someone else as a DMZ] ahead of both of the above routers, and that DMZ
connection did not seem to be going down which showed that the problem
was at my end and not at Comcast.

I thought I might get some clues by looking at syslog, which does show
these disconnects. The log is very large, but I'm hoping the following
excerpts might provide enough clues to understand what is going on.

This snippet shows the disconnect:

"
NetworkManager[918]: <info>  [1501536278.6169] device (enp4s0): link 
disconnected
NetworkManager[918]: <info>  [1501536278.6170] device (enp4s0): state
change: activated -> unavailable (reason 'carrier-changed') [100 20 40]
.....
NetworkManager[918]: <info>  [1501536278.6281] manager: NetworkManager
state is now DISCONNECTED
"
and the following snippet shows the reconnect:

"
NetworkManager[918]: <info>  [1501536319.8205] device (enp4s0): link
connected kernel: [2495080.876510] igb 0000:04:00.0 enp4s0: igb: enp4s0
NIC Link is Up 1000 Mbps Full Duplex
NetworkManager[918]: <info>  [1501536319.8228] manager: NetworkManager
state is now CONNECTING
...
NetworkManager[918]: <info>  [1501536319.8452] policy: set 'Wired
connection 1' (enp4s0) as default for IPv4 routing and DNS
...
nmbd[1466]: [2017/07/31 14:27:07.998589,  0] 
/source3/nmbd/nmbd_become_lmb.c:397(become_local_master_stage2)
nmbd[1466]:   Samba name server CHEWIE is now a local master browser
for workgroup FAMILY
"
NOTE - this computer IS the one that I would like to be the master
browser for the workgroup, so that part is working as I'd like.

The network then functions okay for 5 to 20 minutes, then repeats the
same disconnect/reconnect again and again.

At this point I disconnected the second router/access point and the
behavior changed. It no longer disconnects/reconnects, but the network
behavior in the Ubuntu box is different, and not what I want.

At the next reconnect we get:

NetworkManager[918]: <info>  [1501536890.1845] manager: NetworkManager
state is now CONNECTING
...

nmbd[1466]: [2017/07/31 14:39:06.910864,  0] 
../source3/nmbd/nmbd_incomingdgrams.c:304(process_local_master_announce)
nmbd[1466]: [a different Linux computer ] is announcing itself as a
local master browser for workgroup FAMILY and we think we are master.
Forcing election. 
nmbd[1466]:   Samba name server CHEWIE has stopped being a local master 
browser for workgroup FAMILY

[1] Now this is NOT what I want, but it is not clear to me why this box
has stopped being a local master browser for the workgroup. I would
appreciate any insights as to why that happened, and how I can get the
original 16.04 computer to be the master browser, and stay that way.

[2] Also, it is not clear to me why adding the second router as an
access point caused the first router to periodically stop working. I
would also appreciate VERY MUCH any insights as to why that might
happen.

I can post the entire logs if that would be helpful, but they are quite
large - please let me know if I should do that.

Thanks for reading this...