The TG582n FTTC that keeps thinking it's a printer - SOLVED

[#1]

Hi all,
Finally (no thanks to Openreach taking 3 weeks of cancelling and rearranging appointments to deliver a modem) got fibre this week, and it is glorious, so started dropping services off the Plusnet-supplied Technicolour router this evening and picking them up on a Debian box I've got running bind9 and isc-dhcp. (It's actually been running my network with a bridge to next door's wireless for a couple of weeks now!)
Best I've been able to get is about 10 minutes out of it. The problem could be in a million and one areas, including my DNS config (but I doubt it) but the one that I keep coming back to is the slightly schizophrenic nature of the router - at times I can ping all the devices on my internal network by hostname (so internal DNS working then) but I can't reach the router's web interface by IP or hostname. I can ping it, and I can get on via the 10-address, but if I fire up nmap, I get the following:

Nmap scan report for 192.168.13.254
Host is up (0.017s latency).
Not shown: 998 closed ports
PORT    STATE SERVICE
139/tcp open  netbios-ssn
515/tcp open  printer

Where normally when it's working, I get:

Nmap scan report for 192.168.13.254
Host is up (0.0077s latency).
Not shown: 994 filtered ports
PORT     STATE SERVICE
21/tcp   open  ftp
23/tcp   open  telnet
53/tcp   open  domain
80/tcp   open  http
443/tcp  open  https
1723/tcp open  pptp

Has anyone else experienced this sort of problem?
Thanks in advance,  
Weasel

[#2]

Hi,
I'm sure this is frustrating, I'm also sure someone will be able to help you, but let's not use serious mental health conditions to describe issues with electronic equipment. 
Hopefully you get this issue sorted.
Cheers,
Wo0x.

[#3]

I had this issue with the supplied router, you need to do a FACTORY RESET on it (again), then start removing the services you don't want. DON'T USE A BACKUP config as this will make it the same. Know been there done that.
I got 30 min then it would through its toys out the pram, no telnet, no web access to the box.
You can do a reset by powering off then press the reset button and hold it while you power up, keep it pressed for about 10-20sec.

[#4]

Wo0x - Topic title changed to avoid any further upset; please accept my apologies for any offence caused by my late-night rambling. Possibly a little oversensitive to my mind, but I take the point. I didn't set out to insult or offend, it was rather late last night when I posted, I was rather tired and had been staring at bind9 debug logs for several hours, and it described the anomalies I was seeing rather well.
Hairy McBiker - Great stuff, cheers! I didn't have any luck trying to reset it by the method you described, but I managed to get into it and reset it that way. Hopefully second time's the charm!

[#5]

No worries, no offence caused, not to me at least. The title stood out to me whilst browsing this morning, my reply was more a ''come on now, we can be a little more thoughtful'' rather than ''how dare you.'' To be fair, I would probably have described it the same way amongst friends. On an open forum, however, I would choose my words more carefully.
Glad it appears you've fixed your issue.

[#6]

If only that were the case, but unfortunately for now I'm giving up. The only way it seems I can get a reliable connection is to let the router deal with my network, so I'm now back to running with no internal DNS and the usual mediocre DHCP setup. *sigh*
May have to invest in a USB ethernet interface for my network services box and remove the router from the equation entirely.

[#7]

Personally, if the 582n does what you need before failing, I would get a low cost (gigabit) router - such as:
Dual band: TP-Link TLWDR3600 (£49 or less), IPv6 capable
N300: TP-Link TL-WR1043ND (£34 or less)
If you don't want gigabit ethernet , then you could go as low as £16 for a N300 device
(Prices based on a quick scan of Amazon)
Phil

[#8]

Or put an extra NiC in your Debian box and run pppoe and have it as an all in one box?

[#9]

PeeGee - It does what I need in that it has the option to disable DHCP, but then fails utterly in that doing this causes it to die. I was also rather hoping not to have to buy more hardware due to the router being so utterly useless if you want to do anything vaguely advanced with it.
dvorak - That's looking like the way forwards tbf. I've got a bit of a hack to try later with the backed-up ini file, but I've ordered a USB->Ethernet adaptor (the Debian box is the gutted remains of a 901 EeePC - cool, silent running, and very low power consumption), so I can either remove the router entirely from the equation, or alternatively plug the router into the adaptor, let it hand out whatever IP it deems necessary, then set my network up as a subnet from the Eee... It makes port forwarding a bit more of a faff (two layers of NAT to traverse) but on the flip-side it does mean I can use the router's inbuilt wireless as a guest network, allowing internet access but keeping all my stuff safely hidden from it.
Far from ideal, but there we go!

[#10]

You should be able to disable the dhcp & dns SERVERS on the box, just make sure you don't disable the CLIENTS, as they are needed for the pppoe.
What you could do, is point the server to you YOUR internal dns server, so instead of looking at the PN servers it queries yours instead.
telnet in
menu
dns/server/forward
then remove the existing pn servers and add your server. or add yours as a higher priority. I did this on my 589 to add the google/open dns as 0 metric and the pn as 20.

[#11]

Not to pop the cork early, but I think I just nailed it.
Firstly, thank you all for your help and advice. Hugely appreciated.
The Problem
Content Sharing. I think I must've got one of the last routers with a USB port for printer/file sharing, as I think they've now discontinued this particular version. The issue was that as a direct result of this, two (damn near identical) MAC addresses were exposed to my internal network:

9e:97:**:**:**:**
9c:97:**:**:**:**

I just pulled this out of the ARP table on my laptop, and to the eye they looked identical - only the second character was different; I had to copy them into Gedit to make sure I wasn't going mad before I spotted the difference. The underside of the router only lists one MAC address (the latter) but when I first set about trying to get all this jazz working, the only MAC I could find on my network was 9e*, so I put it down to a typo on the underside of the router. The font on the underside being typical sans-serif, 'e' and 'c' can look very similar, and trying to differentiate between the two in a terminal window where my default profile uses 7pt monospace fonts is damn near impossible unless you're looking for it, especially given the sheer quantity of bind9 and DHCP logs I've had to scan-read.
Also I couldn't easily find interface MAC addresses in the firmware, hence why I never thought to look for the elusive second MAC address.
So of course, the DHCP record for *.254 had been given the wrong MAC by my server. The rest is history really.
The Solution
How can I say this without making myself look like a complete and utter fool? Put the correct MAC address in DHCP so it doesn't give the gateway IP to the Content Sharing interface. Simple, really.  There, I think I managed it!
If any Plusnet bods are reading, I'm now getting about 2/3 the speed as before - as I'm still in the exchange training period, would this be due to the extensive network downtime I've had? I don't think the router being powered off would cause the modem to disconnect (if the destructions are to be believed.. First time with a PPPoE setup) but would that show up in the RADIUS logs? Also, could you please please put somewhere in the documentation that this version of the router has two MAC addresses! 😉
Thank you all again for your help, greatly appreciated. 🙂
Cheers,
Weasel

edit
If any kind moderator could mark this as [SOLVED] that'd be great... It would appear I'm past the expiry time to modify my OP to update the topic.  Thanks.

[#12]

Have done

[#13]

I don't use the 582 and keep forgetting about the usb port on it, it would agree why you only see the printer side, it is an lpr queue.