DNS servers unreachable

[#1]

One of the reasons that I decided to go with Plusnet was their claims of exceptional customer service. My experience so far is that they are no better than TalkTalk. Two weeks after opening a ticket (and BTW, it wasn't a 'question', it was a fault report), I received the following:

As you rightly state there are some issues with DNS look ups on the HUB ONE and we cannot assign one on the HUB ONE manager.
Our work around for this is to change the DNS on the devices being used.
Ticket Closed
Our suppliers have closed your problem report so we're closing this ticket

I would bring Plusnet's attention to the following:

It's outrageous that an IT company would close a fault ticket without the customer's agreement.

The fault is not in the Hub One router - I experienced the same problem with a Huawei router.

Changing the DNS setting on connected devices is not a reasonable workaround. What about visiting guests? What about smart TVs?

Will Plusnet change all their advertising to add the caveat that smart devices (IOTs) other than PCs and smartphones, e.g. smart TVs, are not supported?

Oh, and my sync speed has dropped again to 68Mbps.

[#2]

I'm sorry to see your fault ticket was closed. Looks like it was done by the last adviser which responded to you. I've provided feedback to them now.

Unfortunately there isn't anything we can do as the Hub One doesn't have the option to change DNS.

Smart devices should connect just fine to the router without you needing to change the DNS server. If you're having problems with the wireless signal, can you run through the steps here and let us know how you get on?

With regards to your speeds, as the sync rate of 68mbps is above your estimates of 49mbps, I'm afraid that there's not much we can do. Have you tried plugging your router into the test socket to rule out any internal wiring?

Apologies for any inconvenience or frustration caused.

[#3]

---

---

@Gandalf wrote:

Unfortunately there isn't anything we can do as the Hub One doesn't have the option to change DNS.

---

Of course there's something Plusnet can do: fix the underlying problem - which is something to do with UDP routing in BT's network.

---

@Gandalf wrote:Smart devices should connect just fine to the router without you needing to change the DNS server.

.....and if the DNS server is unreachable? How's that going to work?

---

@Gandalf wrote:If you're having problems with the wireless signal, can you run through the steps here and let us know how you get on?

I'm not having problems with the wireless signal.

---

@Gandalf wrote:Apologies for any inconvenience or frustration caused.

It's not about inconvenience! There's an admitted fault in the system which Plusnet refuse to address!

---

---

---

---

[#4]

Do you experience the same problems using a wired connection?

[#5]

@Gandalf wrote:

Do you experience the same problems using a wired connection?

If you're talking about the speed drop, then the question is moot - I'm looking at the sync rate in the router.

If it's the DNS problem you're thinking of, I don't believe there is any mechanism described by 802.11 which could differentiate between UDP and TCP.

Whilst writing this I see that CS have opened a new ticket. As the author makes a point of calling it a 'ticket' rather than a 'question' I suspect it was a result of my previous post and @Gandalf's influence. Thank you. However, the new ticket ( 173174878 ) appears to be closed. It refers to 'open' ticket 171928378 which doesn't exist.

Moderator's note by Dick (Strat): Quote fixed.

---

[#6]

If you're talking about the speed drop, then the question is moot - I'm looking at the sync rate in the router.

Have you tried the test socket?

If it's the DNS problem you're thinking of, I don't believe there is any mechanism described by 802.11 which could differentiate between UDP and TCP.

I'd still recommend trying a wired connection as this will help narrow down where the problem lies.

[#7]

The next time this happens it would be useful to rule out the DNS proxying that the router performs. This is easy enough to do by specifying the DNS server for lookups, e.g:
nslookup bbc.co.uk 213.120.234.42
If that works, but the same command without the IP address doesn't, then it's likely the router is misbehaving in some way. If not, then I agree something probably needs looking into.

[#8]

I originally mentioned the line speed problems in this thread merely as an aside, I thought it was just because I was messing with different routers, but I've been watching the router log and I can see that over the last few days the line has been up and down like a yo-yo. I'm now certain that there are two problems: a flaky line and a weird UDP problem.

Last night the line went down. It came back up with a new IP address and the UDP problem came back. The most obvious symptom of this is that DNS requests fail. I can resolve this, for some devices, by assigning a different server in the device config. I can't fix the problem of my VPN client failing to connect over UDP.

Edited router log (reordered to read from top to bottom) :

04:18:58, 19 Apr.    (60047.570000) PPPoE is down after 998 minutes uptime [Waiting for Underlying Connection (WAN Ethernet 7 -? Down)]
04:20:21, 19 Apr.    (60130.950000) PPPoE is up -? Down Rate=76998Kbps, Up Rate=17565Kbps; SNR Margin Down=6.0dB, Up=5.9dB
06:40:58, 20 Apr.    (154964.660000) PPPoE is down after 1580 minutes uptime [Waiting for Underlying Connection (WAN Ethernet 7 -? Down)]
06:42:22, 20 Apr.    (155049.020000) PPPoE is up -? Down Rate=68110Kbps, Up Rate=17383Kbps; SNR Margin Down=6.1dB, Up=5.9dB
16:07:34, 24 Apr.    (534552.580000) PPPoE is down after 6324 minutes uptime [Waiting for Underlying Connection (WAN Ethernet 7 -? Down)]
16:09:03, 24 Apr.    (534641.850000) PPPoE is up -? Down Rate=39800Kbps, Up Rate=6348Kbps; SNR Margin Down=6.1dB, Up=5.7dB
22:06:18, 24 Apr.    (556076.780000) PPPoE is down after 357 minutes uptime [Waiting for Underlying Connection (WAN Ethernet 7 -? Down)]
22:07:49, 24 Apr.    (556167.640000) PPPoE is up -? Down Rate=38750Kbps, Up Rate=6367Kbps; SNR Margin Down=6.2dB, Up=5.8dB
05:36:58, 25 Apr.    (583116.690000) PPPoE is down after 449 minutes uptime [Waiting for Underlying Connection (WAN Ethernet 7 -? Down)]
05:38:27, 25 Apr.    (583205.960000) PPPoE is up -? Down Rate=38889Kbps, Up Rate=8978Kbps; SNR Margin Down=6.1dB, Up=6.1dB

I rebooted the router at around 11:00 on the 25th, after plugging it into the test socket. (However, there has never been a phone plugged into this line and there is no internal wiring.)

01:22:20, 26 Apr.    (51173.120000) PPPoE is down after 850 minutes uptime [Disconnected]
01:22:52, 26 Apr.    (51205.190000) PPPoE is up -? Down Rate=68893Kbps, Up Rate=17409Kbps; SNR Margin Down=5.9dB, Up=6.5dB
02:50:39, 27 Apr.    (142869.970000) PPPoE is down after 1527 minutes uptime [Disconnected]
02:51:11, 27 Apr.    (142901.710000) PPPoE is up -? Down Rate=68893Kbps, Up Rate=17409Kbps; SNR Margin Down=5.8dB, Up=6.5dB

The last message before the line went down was:

02:38:52, 27 Apr.    IN: ACCEPT [57] Connection closed (Port Forwarding: TCP [192.168.11.39]:22 <-?-?> [80.189.103.7]:22 -? -? -? [103.89.90.105]:61163 CLOSED/SYN_SENT ppp3 NAPT)

The significant part is the IP address, 80.189.103.7. This has been in effect since the 25th.
When the connection is restored, we get a new IP address, 80.189.200.135, and the DNS problem returns as shown by TR-069 errors and DDNS update failures immediately after:

02:51:35, 27 Apr.    (142926.240000) CWMP: session closed due to error: Could not resolve host
02:52:34, 27 Apr.    (142985.530000) Updating NoIP service with IP 80.189.200.135 for host rsmvpn.ddns.net
02:52:49, 27 Apr.    (143000.550000) No response from server

I think that this is a problem with the BT network whereby certain packets can't find their way back to the source when the source IP address is in certain IP address ranges, which has been reported in the past.

Just to recap - DNS requests to 213.120.234.42 fail but succeed to 212.159.6.9. My VPN client can't connect via UDP (to any server in any country) but can via TCP.

What I would like to happen now is that the line problem is fixed and the UDP problem handed to someone who understands IP protocols and is familiar with packet routing in BT's network.

[#9]

Is anyone investigating this? The last I heard from customer support was a week ago when they opened a new ticket 171928378 - which doesn't seem to exist.

I first reported this problem several weeks ago and nothing appears to have been done to resolve the issue.

[#10]

Rebooted the router to get a new IP address (80.189.157.46) and the UDP problem has gone again.

Will someone look at this please, because it will come back at some point, depending on the IP address allocated the next time.

[#11]

Is your line problem being looked at? If that is ongoing I imagine it will make diagnostic work on DNS behaviour more difficult.

[#12]

Nothing is being looked at.

The line seems to be stable at the moment, it hasn't dropped recently and the sync rate has stayed at 68Mbs. The line problem has nothing to do with the UDP problem.

It is now more than three weeks since I reported this and no one has investigated the real issue, which it that UDP packets are being lost in BT's network in circumstances which are affected by the IP address allocated.

[#13]

I've escalated this topic to raise its profile.

[#14]

Does the UDP problem seem to come when you have IPs from within certain ranges?   Or does it seem random?

[#15]

I don't have enough data to pin the problem to a specific IP range. All I can say at the moment is:

80.189.103.7 OK
80.189.157.46 OK
80.189.159.83 OK
80.189.200.135 Not OK

What is the lease time for an IP address? It doesn't seem to change with every reconnect or router restart - I've had 80.189.200.135 for 13 days.