Re: [gentoo-user] Re: losing network IP address

On 03/21/2017 03:28 PM, Alan McKinnon wrote:
> On 21/03/2017 23:05, thelma@×××××××××××.com wrote:
>> On 03/21/2017 01:25 PM, Kai Krakow wrote:
>>> Am Tue, 21 Mar 2017 11:08:49 -0600
>>> schrieb thelma@×××××××××××.com:
>>>
>>>> On 03/21/2017 10:21 AM, Ian Zimmerman wrote:
>>>>> On 2017-03-21 06:59, thelma@×××××××××××.com wrote:
>>>>>
>>>>>> Sometimes it does, the IP stays for a several days but sometimes it
>>>>>> boots with correct IP and the IP changes after few hours to
>>>>>> something like 192.168.xxx (which is not what my firewall
>>>>>> provides.)
>>>>>
>>>>> Maybe you have another (misconfigured) DHCP server on the network?
>>>>> tcpdump could be your friend.
[snip]
>>>
>>> What's strange about these logs?
>>>
>>>> This must be from "apcupsd"
>>>> Could this service causes some problem with my network?
>>>
>>> That service is only victim of the IP address changing. It's not the
>>> source of the problem.
>>>
>>>> Shortly after I could not receive any message from the system, I
>>>> think the network connection got lost. --
>>>
>>> You wrote that this happened in different locations, I guess you mean
>>> different networked sites with each their own DHCP server.
>>>
>>> So, I my best guess is that you're running a local DHCP server by
>>> accident.
>>>
>>> Which service does configure your network?
>>>
>>> If it's systemd-networkd, you could check those logs after the problem
>>> occurred:
>>>
>>> # journalctl -b -u systemd-networkd.service | fgrep -i dhcpv4
>>>
>>> Apparently, it doesn't tell you what the DHCP server is by MAC address,
>>> but the reported "via" at least tells you the IP.
>>>
>>> It looks like this:
>>>
>>> Mar 17 19:21:33 jupiter systemd-networkd[680]: enp5s0: DHCPv4 address
>>> 192.168.4.102/24 via 192.168.4.254
>>>
>>> Other network management daemons should log similar lines. Try first
>>> without fgrep.
>>
>> I run DHCP on my DD-WRT router, and all the boxes are running OK,
>> except this one. The IP the box suppose to get is: 10.10.0.7 (static
>> IP ) and here is an entry from last night long:
>>
>>  eden dhcpcd[7478]: net0: adding address fe80::a98a:8875:2106:64bc
>>  eden dhcpcd[7478]: DUID 00:01:00:01:1a:bc:bc:ca:00:30:18:ad:ed:b4
>>  eden dhcpcd[7478]: net0: IAID 18:ad:ed:b4
>>  eden dhcpcd[7478]: net0: soliciting an IPv6 router
>>  eden dhcpcd[7478]: net0: soliciting a DHCP lease
>>  eden dhcpcd[7478]: net0: probing for an IPv4LL address
>>  eden dhcpcd[7478]: net0: using IPv4LL address 169.254.156.68
>>  eden dhcpcd[7478]: net0: adding route to 169.254.0.0/16
>
> That's an APIPA address. Think of it as a no-network-config-required
> fallback. The machine did that because it didn't get a real address.
>
> If you've checked you don't have some static network config local to the
> box, and your mac address is correct in the dhcp server, then you
> probably have physical layer 1 problems.
>
> You say you've checked the cables and two network points. OK, that rules
> out your machine, cable and RJ sockets.
>
> Next check the switch, use a different port.
> Switches are high susceptible to having ports damaged, it comes from
> induced spikes from the long cable runs plugged into them. Homo sapiens
> doesn't know how to fix that, no matter what marketing departments print
> on boxes

Good point, I'll take an extra switch to remote location as all my ports
are almost taken so I'll need extra switch anyway.

It looks like the PC wasn't getting the correct IP so I can not rule-out
a port on a switch.

--
Thelma