Hi all,
I am new to this forum; not quite a newbie, but not really experienced either. Hope somebody can help me with what I find to be a tough nut to crack:
I've got two laptops at home, both (still) running SUSE11.0, both installations from the same DVD. They share one line to a satellite broadband modem via a network switch (not a router). Living out in the sticks, I have had satellite broadband for years, without any problems. Recently an old modem died and I got a new one. I know that the modem and connection to the satellite are fine; they have been tested with a Windows machine (by the technician who installed the modem). Connection to the modem is via network cable, standard stuff (unchanged to what's been in use for years). My expectation was that the two linux boxes will just pick up the new connection and get on with the job. Now to the tricky bit: ONE laptop is doing exactly that, the other does NOT reconnect to the new modem! Having two systems, obviously I can compare. I have looked into the network settings using YaST2 (on gnome) and cannot find any differences.
I have rebooted the machine that does not connect and also restarted the network manager. The reaction to restarting the network manager is that linux reports (via the icon on the toolbar) "Requesting a network address from the wired network", but then it comes back with "Disconnected - The network connection has been disconnected". I don't know whether that means that the machine is actually connected in the first place and then, without an IP (using DHCP), disconnects or whether it does not get any response from the modem in the first place. I find this bizarre, given the fact that the other machine (a clone in terms of the OS, but a different hardware model) connects without me having to do anything... What else can I test/try? Could there be a leftover entry somewhere in a network configuration file that is not being overwritten properly?
Any help would be much appreciated. There are similar threads, but I have not found any one that helps me with this particular problem.
Cheers,
Michael

OK, there is a lot here, but having two computers really ought to make things easier, because you can compare one with the other.

The other thing with networking, in particular, is that you should think of it like an onion; lots of layers, and while each of the layers may be easy on its own, you need to get all of them right, and, however much you mess with the wrong layer, it won't help. So, you need to start doing some investigating, and getting out the right tools.


So, both computers used a wired ethernet connection. They are both plugged in to identical (nominally) ethernet ports on the switch, everything uses the same baud rate ethernet?

Having established those, we can then move on to;

• Presumably, the computer that is working picks up a sensible ethernet address (you might want to say what address range you are expecting to use, what the 'working' computer is getting, and what pool of addresses DHCP is set up to hand out - there is the possibility that it could be set up to only hand out one - that would be a bit idiosyncratic, but, technically, it is a possibility, and would mean that only the first computer, whichever it was, would work).
• What is happening with the non-working computer; does it get an ethernet address at all? Is the ethernet interface up?
• Now, can you ping anything? Either the modem or the other computer?
• (You mentioned dhcp as if you know that dhcp is being used. It could be mDNS. What do you have in resolv.conf, as, if that gets overwritten, it usually says who has overwritten it.)

You could have to look at other tools as well. I know that you usually go to Yast for sys admin, but you might have to go a bit further today.

ifconfig, for instance

Another thing that you could look at, is whether both computers are running the same processes (services). Yast can give you this information, but you would probably be better served by ps, because it is easier to post the output. In any case, are there any differences in whether the two run nm, nm-applet (or maybe, gnome-applet, or something), dhcpd and that kind of thing? You'd expect them to be identical, but the process numbers probably won't be, but the more important question is whether you've got the same processes running, and whether any missing ones could make any difference to networking.

It would be worth having a look at dmesg, just after boot, in case any services have shut down for any reason.

Well, if it is dhcp, or any of the mDns family, it should wait for an address to be given to it. Presumably, if, after the timeout period, it hasn't had anything that it would accept, then it probably will announce that the network connection has been disconnected. These error messages can be a nit confused/confusing, and you are better off if you can confirm that they mean what you think they mean.

ipv6 is off, isn't it?

And note that 'it didn't work' isn't anywhere near as helpful as 'I used exactly this command, and here is the error message'. So, when you try ping, if it doesn't work, paste the command and error message, because that is where the information lies.

When the rest of the stuff is working, there is still dns; if you can ping ip adresses, but you can't ping hostnames, then there is probably something wrong with name lookup, and very probably something in resolv.conf is stopping you from using a sensible nameserver.

OK, many thanks! Here is some more information, to try and answer your questions.

Regarding various layers: Yes, agreed. The thing that is so surprising is that one computer picks things up correctly, while the other doesn't, although both were working prior to the outage. And I have not changed anything.

Correct, everything uses the same baud rate internet. The only difference between the two is that they are different models Dell laptops. But obviously there must be a difference in the OS setup somewhere (that I was not aware of and now cannot find).

Regarding IPs: I doubt the numbers themselves will help you, but yes, I can see, in the setup of the computer is working, the two entries (as I expected them) for DNS1 and the gateway. I also know the IP address of the modem, but canNOT ping it from the computer that is not connecting properly.

The ping output of the machine that does not connect is as follows:
>ping [modem IP]
connect: Network is unreachable

Regarding handing out a single IP: No, that is not possible. The computer that is not connecting is the one with which I tried to connect FIRST, without network switch and without the second computer. It was then the first and should have been allocated whatever number/connection. As far as I can verify my ISP has given the computer that DOES connect the same (static) IP address that it had prior to the outage. The one that does not work also had a fixed address (I can find that number somewhere in my notes, I am sure). I can take the computer that does not connect off the network switch, but my tests suggest that this plays no role. It always behaves the same way, irrespective of whether it is the first, second or only computer connected to the modem (without the switch).

/etc/resolv.conf does not tell me anything:

### BEGIN INFO
#
#
### END INFO
#
search site

As for comparing the two machines, here a copy of the two ifconfig outputs. First the computer that does NOT connect:

eth0 Link encap:Ethernet HWaddr 00:14:22:EC:C5:F3
inet6 addr: fe80::214:22ff:feec:c5f3/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:2 errors:0 dropped:0 overruns:0 frame:0
TX packets:11 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:128 (128.0 b) TX bytes:2222 (2.1 Kb)
Interrupt:17

lo Link encap:Local Loopback
inet addr:127.0.0.1 Mask:255.0.0.0
inet6 addr: ::1/128 Scope:Host
UP LOOPBACK RUNNING MTU:16436 Metric:1
RX packets:230 errors:0 dropped:0 overruns:0 frame:0
TX packets:230 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:0
RX bytes:17908 (17.4 Kb) TX bytes:17908 (17.4 Kb)

wlan0 Link encap:Ethernet HWaddr 00:1B:77:44B:91
UP BROADCAST MULTICAST MTU:1500 Metric:1
RX packets:0 errors:0 dropped:0 overruns:0 frame:0
TX packets:0 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:0 (0.0 b) TX bytes:0 (0.0 b)

wmaster0 Link encap:UNSPEC HWaddr 00-1B-77-44-DB-91-00-00-00-00-00-00-00-00-00
-00
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:0 errors:0 dropped:0 overruns:0 frame:0
TX packets:0 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:0 (0.0 b) TX bytes:0 (0.0 b)

Now the computer that DOES connect:

eth0 Link encap:Ethernet HWaddr 00:1C:23:A9:0B:E7
inet addr:114.129.162.34 Bcast:114.129.162.35 Mask:255.255.255.252
inet6 addr: fe80::21c:23ff:fea9:be7/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:26998 errors:0 dropped:0 overruns:0 frame:0
TX packets:25056 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:15922774 (15.1 Mb) TX bytes:4202259 (4.0 Mb)
Interrupt:17

lo Link encap:Local Loopback
inet addr:127.0.0.1 Mask:255.0.0.0
inet6 addr: ::1/128 Scope:Host
UP LOOPBACK RUNNING MTU:16436 Metric:1
RX packets:3200 errors:0 dropped:0 overruns:0 frame:0
TX packets:3200 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:0
RX bytes:160400 (156.6 Kb) TX bytes:160400 (156.6 Kb)

wlan0 Link encap:Ethernet HWaddr 00:1C:BF:33:7F:BF
UP BROADCAST MULTICAST MTU:1500 Metric:1
RX packets:0 errors:0 dropped:0 overruns:0 frame:0
TX packets:0 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:0 (0.0 b) TX bytes:0 (0.0 b)

wmaster0 Link encap:UNSPEC HWaddr 00-1C-BF-33-7F-BF-00-00-00-00-00-00-00-00-00
-00
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:0 errors:0 dropped:0 overruns:0 frame:0
TX packets:0 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:0 (0.0 b) TX bytes:0 (0.0 b)

Clearly, the machine that does not connect does not have an IP address allocated to it. Other than that I don't know how to read this output.

Don't know how to look into /bin/dmesg (it is a binary file). But I have had a look into boot.msg and see there that the network manager has been started correctly. I don't see anything that might relate to networking that has crashed (such as e.g. eth0).

Yes, I agree with your interpretation of the disconnect error message. The timeout period must be something like 5 or 10 seconds.

Both machines have ipv6 ENABLED. Should I disable it?

Hope I haven't forgotten anything... Ta.

OK, wish you had posted that earlier, as it indicates that something fairly fundamental is wrong, and it is happening relatively early on in the process.


That's all wrong (essentially, it is blank), but if the rest was working, it would get overwritten, so that possibly isn't a problem.

and

Except for experts, choose one of ipv6 and ipv4 for the network and stick to it. For most people that will be ipv4. So, disable ipv6.

That computer does not have an ipv4 address; I think the first thing to try is to give it an ip address manually (say, whatever ip your other computer gets +1...which would be:114.129.162.35 ). You could do this in Yast (I think it is 'control interface by traditional ifup/ifdown method' and manually add the address), although you could also do the ip address from the command line.

At that point, you should be able to ping the modem or the other computer, and maybe even more than that starts working (so DNS may or may not start working then).

If this is a 'static' (desktop computing, with a fixed position) computer, this may be all that you need, although it isn't very elegant, when NM ought to assign the ip for you. For a computer that you take from one network to another, maybe this is a real problem.

If everything works apart from DNS, you have two choices
(Probably)you can set your modem as your DNS resolver, and provided it gets its resolutions from upstream, that should work
If either you don't want to do this, or the modem doesn't have that facility, you can use your ISPs Dns server(s), Google's, openDNS, etc, etc. Google probably isn't the absolute fastest, but it is pretty universally available.

From http://code.google.com/speed/public-dns/docs/using.html

"Google Public DNS IP addresses

The Google Public DNS IP addresses (IPv4) are as follows:
8.8.8.8
8.8.4.4 "

and, you could use, eg, the Google resolvers in your modem (probably) and tell both of you computers to use the modem as their source of resolves. If your modem doesn't allow you to set upstream DNS servers, then just ensure that your individual computers know which resolvers to use.

That should be enough to get you going (i hope), but it doesn't really address the question of why NM hasn't sorted this out for you. Let's see what happens, and proceed from there.

(I keep forgetting check the cables...I know it is a bit daft, but cables do go wrong.)


For dmesg, you just type 'dmesg' at the command line. It gives the latest messages, and shortly after boot (maybe 'shortly' means up to 36 hours...) the buffer won't have overwritten any of the messages.


What the ISP ought to be doing is to give an ip address to the external interface (the internet-facing port) of the modem. They could do this via dhcp or ppp, but, really, they have no business messing with anything else...however they might find that messing with their clients' internal networks minimises support calls, so you can never tell what they think is appropriate.

I've probably forgotten some useful stuff, so tell me how you get on.

I disabled ipv6, but in yast could not figure out how to manually set the IP address (unless it is called "DHCP client identifier"). Some input fields are not active. Anyhow, I failed. Whatever I tried, I still cannot ping the modem.

But that gave me another idea, which I think led to significant new insights - although I tested everything up to and including the network cable connected to the computer, I am starting to suspect that I may have a hardware problem.

Why? Because I remembered that I still have a knoppix 5.3 DVD here and I booted my machine from that. Then entered the Control Centre and looked up the network settings in there. It finds both wireless (correct!) and wired (eth0) inactive. I went into administrator mode, chose SUSE9.1 (the closest I could find to my SUSE11.0) and could then edit the eth0 settings - there I could find easily how to enter an IP address. Then I tried to activate eth0, without success. I tried to ping the modem from the command line; same error message as before (network unreachable). This means that both operating systems fail to connect (well, are they independent of each other? knoppix tells me that it is working with the settings made by SUSE). If knoppix tries to connect to the network independent of any bad settings in SUSE, my suspicion is that this could indicate a network card failure (or possibly the cable socket). I had disconnected the network cable during the long outage of 2 months, but the card may yet have died while not in use (I have used the computer regularly offline). The computer is almost 5 years old... I had checked the hardware detection entries in the boot log, where the network card was found, but it does not say anything about it being functional. Later on, eth0 is being started successfully (I might need to re-check this), but it fails to connect. Any opinion about this? I hope what I am writing makes sense...

You won't believe this, but whatever happened (possibly setting an IP address using knoppix linux) in the end did make a change. Next time booting the laptop that previously refused to connect, it went straight online. Don't know what I've done, but I've done it... Many thanks for the help!