NTL + Router = Hell

[Mr Wonka]

Hello all

I'm writing to this list as I feel the current support I am getting will not help to resolve the issue.

I have been receiving NTL Broadband through a NTL:Home 120 Cable modem for the last 18 months. This has been routed through a Netgear WGR614 into my home network which consists of a Dualboot WinXP Pro/Gentoo Desktop, a CentOS (Redhat EL Clone) Mail server, and a WinXP Pro Wireless Laptop.

I will first describe the events as they have occured and then provide the more technical information.

Up until last Tuesday (8th March) all of this equipment had been working just fine together. On Wednesday I awoke to found that connectivity to the internet had ceased but that traffic to the local network was working fine. I reset the router, which has crashed in a similar way in the past, to find that the problem persisted. I rang NTL tech support that morning and was told that maintenance was being carried out and connectivity would return soon. That evening, the connection still was not working and I ran again. Only to be told that the maintenance was carried out from 12am that morning and had finished at 6am. A good 5 hrs before I had rang previously.

Confused, I set about debuging my network setup. Thinking that it was now a local problem. The modem was 'hard reset' to factory settings and the networked computers configured to use DHCP. This did not restore the internet. The router was now flashed to the latest firmware and the process was repeated. Again this was not effective. I proceed to remove the Router from the equation and connected the Modem directly to the Desktop machine. After a modem reset my external connection started to work correctly. Absolving NTL of any, and all responsibility. I had tracked the problem to the router. I proceeded to order a new one thinking that since it was almost a year old it was obsolete and a new one would be better then hours on the phone to the Netgear Tech support line.

Yesterday my new router arrived. A Netgear WGT624. I connected it all together, reset the Modem, powered on the Router, and finally the desktop machine. The result... It still does not allow an external connection.

Cue Tech Support. A call to NTL later, and I had confirmed that the responsibility was not theirs, the modem was functioning correctly according to their remote access software. Read on for reasons why I think NTL have more responsibility then they let on. I spent the better part of 7hrs on the phone to Netgear Tech support yesterday. Running through so many different configurations, firmwares, and combinations of resets on both routers that I have lost count. Currently Netgear have 'escalated' my case number and I am now expecting a call from them. As far as I can tell I've managed to completely stump the India based, read the process of a screen, techs and have now been handed over to someone who may or may not have a qualification in an IT related area.

Now for the technical details. Both routers are apparently able to use the NTL DHCP server to obtain all the correct connection information. On the status pages of both routers is shown the correct gateway and DNS IP's as well as an assigned IP address and netmask. They both appear to have correctly done what they need to do but neither of them route traffic to the internet. Both routers have had MAC masquerading turned on and off. Both have had 'hard resets' and upgrades of firmware. Both have had as many settings fiddled as are to be fiddled with. In short, although they appear to be set up just fine they do not work. Which is what has stumped the Tech support people.

Now for why I belive NTL is possibly responsible.
1. The connection loss occurred right after (or during) a period of maintenance. Up until this point I had had 11 months of reliable service from this setup.
2. A brand new Router, of a different model, failed to connect. With the exact same symptoms (Valid IP, Gateway, and DNS but no connection)
3. This is possibly (In my eyes) the most conclusive evidence yet:

In an attempt to temporarily connect the internet to the network I decided to install a second NIC in the CentOS server and route traffic through it. Redhat by default create everything you need to set this up with the minimum of fuss. Low and behold this new setup is still unable to route traffic to the net. Knowing something about the way Linux works I set about debuging the Network startup scripts. I narrowed the part where it all goes pear shaped down to one command.

[root@serverpie ~]# dhclient eth0
Internet Systems Consortium DHCP Client V3.0.1
Copyright 2004 Internet Systems Consortium.
All rights reserved.
For info, please visit http://www.isc.org/products/DHCP

Listening on LPF/eth0/00:06:4f:00:ce:41
Sending on LPF/eth0/00:06:4f:00:ce:41
Sending on Socket/fallback
DHCPREQUEST on eth0 to 255.255.255.255 port 67
DHCPACK from 10.67.240.1
SIOCADDRT: Network is unreachable
bound to 82.24.5.75 -- renewal in 76257 seconds.


A quick bit of Googling later tells me that "SIOCADDRT: Network is unreachable" is a 'route' command error and that at this part of the dhclient process it is trying add the default route to the gateway. The gateway through which all forwarded traffic should be directed. Guess what? My CentOS box shows all the symptoms of my routers. It is able to create and manage a local network but it is not able to route traffic to the external interface.

I've got a theory. NTL have done something that, intentionally or not, stops certain DHCP clients from being able to add the nessesary routing commands.

This is everything I've got now. I know nothing more then this. If anyone can shed some light on the situation I would be more then grateful.

Thanks
Mr Wonka

P.S. I'm a Computer Science Student in my final year. 1. I don't need this right now. 2. Why do I get talked to by tech support people like I've only just figured out how to send email? :confused:

[Tuftus]

Firstly :welcome:

Secondly, whoosh! that went straight over my head, but I am sure some one will be able to make sense of it soon enough...

;)

[Richard M]

2. Why do I get talked to by tech support people like I've only just figured out how to send email? :confused:

Because the majority of people who phone TS lines are clueless, I know because I've worked at a helpdesk.
I can't work out your problem, hopefully one of the NTL network guys can though.

[Chris W]

if you PM me the mac address of the modem i will have a little look at what is going on...

*cue mod to confirm it is ok to give me the details

[Nemesis]

if you PM me the mac address of the modem i will have a little look at what is going on...

*cue mod to confirm it is ok to give me the details

Yeah, the monkey is safe :D

[Mr Wonka]

I've Just been on the phone to the Netgear L2 support. The router is being RMA'd and I'm getting a new one.

Judging by the fact that a New one already failed I've not got much hope.

Mr Wonka

[xathras]

I use a linksys wireless router for my house, I am hard-wired to the router and receive rare problems, now and again I have to release the dhcp lease on the router and renew it to restore connection.
__________________

it may also be useful to directly connect to your cable modem and use docdiag tool to check your configuration and also log into 192.168.100.1 to see any errors you are getting

[Mr Wonka]

You would not belive how many things have been tried. You name it, it's been done. As far as NTL are concerned it's not their problem - "The modem work fine when directly connected so it's the routers fault". My initial post explains my doubts on this though.

I'm currently waiting for; 1. A new DLink router 2. The numerous DHCP leases my fiddling has created to time out (about a week).

I still belive something has changed in the DHCP dialog of the NTL server but I'll never be able to prove it. Oh well.

If the DLink don't work I'm very tempted to lose NTL. They seem very stubborn in their insistance that nothing is different. Again, oh well.

Mr Wonka

[JohnHorb]

I have to say this sounds more likely to be a problem with your setup, but I stand to be corrected.

Since my *nix knowledge is virtually non-existant, I'll concentrate on the XP side.
1) Are your XP systems successfully obtaining IP addresses from the router? (ipconfig /all from a command window). You should see 192.168.x.y. If the address starts 169, the PCs are not getting addresses from the router for some reason.
2) Can you ping IP addresses on the Internet? (ping 212.58.228.154 - this is bbc.co.uk. Note that you'll need to choose an address which responds to pings - I know this one does). If the answer is no, there is, as you suspect, a problem getting out to the net. Any firewalls involved (XP firewall, zonealarm, etc)?
3) Can you ping internet addresses by name? (ping bbc.co.uk).
If the answer to 2) is yes, but the answer to 3) is no, it is a DNS problem. What DNS server is shown in the ipconfig at step 1)?
You may have already done all this, but, if not, post back the results.

[Mr Wonka]

I have to say this sounds more likely to be a problem with your setup, but I stand to be corrected.

Since my *nix knowledge is virtually non-existant, I'll concentrate on the XP side.
1) Are your XP systems successfully obtaining IP addresses from the router? (ipconfig /all from a command window). You should see 192.168.x.y. If the address starts 169, the PCs are not getting addresses from the router for some reason.

Yes. Static set or otherwise. No conflicts or anything.

2) Can you ping IP addresses on the Internet? (ping 212.58.228.154 - this is bbc.co.uk. Note that you'll need to choose an address which responds to pings - I know this one does). If the answer is no, there is, as you suspect, a problem getting out to the net. Any firewalls involved (XP firewall, zonealarm, etc)?

No. There is no route to the net throught the router.

3) Can you ping internet addresses by name? (ping bbc.co.uk).
If the answer to 2) is yes, but the answer to 3) is no, it is a DNS problem. What DNS server is shown in the ipconfig at step 1)?
You may have already done all this, but, if not, post back the results.

No, but then I can't do 2) either.


As the Netgear L2 Tech. put it "I'm in a bit of a black hole"

Mr Wonka

[JohnHorb]

If you have the PC set to 'obtain IP address automatically' and 'obtain DNS server address automatically' and you do 'ipconfig /all' (from a command window) does it show the router address (probably something like 192.168.2.1) for 'default gateway', 'DNS server' and 'DHCP server'?

[ian@huth]

Sometimes when trying to troubleshoot problems you try several different possibilities and none of them work. It is easy at this stage to then think that it is no use trying them again as you have already proved these don't work. One point to bear in mind is that NTL cable modems will only recognise three different MAC addresses within a four hour period. If you have tried three possibilities with different MAC addresses which don't work and then try a fourth which might have been using the CentOS server within four hours of trying the first then you would not be able to get a connection through the CentOS server. If having tried this once and it not having worked have you tried it again outwith the four hour period?

I would presume that you have tried the obvious things like checking for a faulty cable and turning off any firewalls on the systems.

[Mr Wonka]

Wooha

I received the DLink yesterday. Installed it. Set it to MAC masquerade (Just in case).

and...

It all just works.

As you can imagine I'm quite happy about that. I'm going to wait for the new netgear to turn up and then put that through the same procedure. If it doesn't work then i'll know that NTL don't like netgear. If it does. Then I'll be very confused.

Mr Wonka

[Nemesis]

There are differing DHCP server standards ;)

[MrB]

My Netgear MR814v2 worked straight out of the box - it was only on day 2 (and since) that it refused to allow access to the internet. Therefore hold your breath, cross your fingers for a few days and hope there are still no problems ;-).

Nemesis - you mention different DHCP server standards - it may be a daft question but would different routers get a different response on the same NTL modem (in my case a Pace STB, 1MB). Do some brands work better than others?

Can anyone recommend a wireless router that works well with an NTL Pace STB? I have returned my Netgear as it just wouldn't play nicely (discussed on another thread last week). I only need 11mbs to work with the rest of my kit rather than any of the fandango routers with go faster stripes - there seem to be quite a few of this type/different brands come up at reasonable prices on Ebay.

Many thanks

MrB

[Mr Wonka]

The saga continues....

My Netgear replacement arrived today. It's a brand new (as in sealed) WGR614 v5. Latest hardware edition as far as I know. It comes as a very fetching white and looks very Mac like. Far nicer then the v2 it replaced.

It doesn't work.

I've fiddled to death. MAC masquerading, static IP, Dynamic IP, MTU. You name it, it's been tweaked.

Looks like I'm going to stick with the DLink then. It has all the aesthetics of a 1980's Sony Walkman and the build quality of a Goodmans CD Player, but it works, so I can always shove it out of sight.

Some Technical stuff
It appears that although I can view the NTL Gateway IP through the interface the router does not route traffic to it. For instance; I can ping any IP on my subnet (Assuming the PC on that IP is on). So if my IP is 82.3.10.156, I can ping any of 82.3.10.*. Including the gateway. This is the exact sympton that my CentOS system suffered when it was connected. I think that it all boils down to the DHCP client not being able to correctly retrieve the gateway IP and set the appropriate routing information for it. Which is also what my manual tweaking on that system showed.

Who's To Blame
As the current Media channels seem keen on, blame has to be appropriated. I blame NTL. During the upgrades to higher speeds they have changed the DHCP implementation (at least in my area). Causing an imcompatibility with the DHCP client used by Netgear routers (Models WGR614, WGT814) and the version of DHClient used by CentOS (Redhat 4 EL).

What will be done
Absolutely nothing. NTL 'do not support home networks' so it's not their problem. Netgear can only change their DHCP implementation to fit. Which won't happen (at least any time soon). I will keep my DLink, flog my Netgear on EBay and then when the contracts up, consider moving to ADSL.

I hope some of this information will be of help to someone.

Thanks
Mr Wonka

Note to NTL Tech's
If any of this can be confirmed, that'd be great.

[Chris W]

[/u]If any of this can be confirmed, that'd be great.

Bung me the mac address of the netgear router that couldn't connect

did you try cloning the pcs mac and connecting with this router?

[Mr Wonka]

Unfortunately the Netgears been packed up for selling on eBay. I'm a bit loathed about unpacking it all again since I want to sell it as New. Which it is. For the duration that it's connected I don't have any access to the net either. Which I can't really afford right now.

I was very wary of connecting too many MACs this time around so before the Modem was connected to the router I set it all up to use the primary computer MAC address. As the DLink is currently set up to do. Infact the Netgear was given the exact same settings as the DLink.

Monkeybreath: I'll PM you the MAC address

Mr Wonka