Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[MrMike]

I currently have a Raspberry Pi in Docklands hosted by the Mythic Beasts company running Ubuntu (no GUI, not interested) to which I SSH over IPv6. The Pi is IPv6-only.

Just curious as to why you're running the Pi on IPv6 only. Purely for testing, or to get around NAT?

[Weaver]

@MrMike If Mythic Beasts have IPv4 addresses available for your device they don't advertise the fact. I just didn't get any IPv4 allocated to my hosted Pi. The Pis can act as IPv4 cheap web servers though because Mythic Beasts have some sort of proxy box available for handling incoming http which is a NATing protocol converter, converts to IPv6 on the Pi’s side and handles the reverse side of session in the usual NAT fashion.

If they decide to start selling IPv4 addresses - assuming they can even scrounge any on the various markets - then I might consider getting one.

My own network at home enjoys native IPv6 from AA so I am all good accessing the Pi to administer it but there are some very old-fashioned boxes that I have that which only speak IPv4 and would like to talk to that Pi. I don't have a 4to6 converter at my end and Mythic Beasts’ NAT system will not handle the case where the Pi initiates a conversation. It’s meant to be the other way around for triggering their NAT in http. But their system might be useful in a lot of cases, I just need to read up on it.

[Weaver]

@spring The First page (Zen) seems to be talking about the WAN (DSL) interface, yes? If so then that isn't me as I’m talking about getting internet access from across the LAN-side ethernet ports where the admin i/f lives and which are connected back to the router.

The second page AA is a good find! Thanks.    It seems to say that static routes don't work or something. It's a bit confusing, AA documentation is sometimes a bit unclear in places, as I can’t tell when they are talking about a B10A used as a router and one used in modem-only mode.

I was doing something almost the same as that but without their excellent metric -1 setting. A clever idea.

I just don’t know then why my commands seem to do nothing, well, no change is visible in the routing table and tests show that it has not actually worked anyway, while failing to show the correct changed state.

[Weaver]

Returning to the ‘first look’ theme of this thread properly, I have to say how impressed I am with this B10A modem. I didn't even know what I was getting and it turns out I’ve got a lot more than I ever knew.

*Monitored tones. It looks like this modem has the monitored tones feature. Is that correct? I take it the DLinks did not have such capability but I don’t know why I say that and I can’t be sure. It seems to me essential anyway, never mind the cost, otherwise you could be on a downward slide over time heading inexorably to an eventual resynch. Is that correct? If my understanding is correct then not having it means a setup that is not sustainable long term and will have a definitely limited connection lifetime. I assumed that I would get this.

* PhyR - I had forgotten about the very existence of this, didn’t know it had been deployed in BT 21CN, assuming that my understanding is correct, and didn't know who the mfr of the DSLAM at the NSBFD exchange was, which is crucial because I believe this is a Broadcom-to-Broadcom only thing? So if I have got this right then an extremely pleasant surprise - big reliability boost and possibly a speed boost too.

* Latency is reduced for some reason. Perhaps this is to do with an indirect consequence of having PhyR. I have a very low downstream interleave D value (see G.992.3) of only 2 (unfortunately). I don't know why but perhaps some software somewhere has decided that given the presence of PhyR then I am adequately protected against burst noise/spikes even with a low D value. Could that be right?

* And just plain speed: an extra downstream 200-400kbps compared with the DLink modems, and at a higher actual SNRM too, so the speed would be even higher if I could make the SNR values the same. Unfortunately on the one line I've tried, the upstream is 10% worse, 515 kbps instead of 566 kbps before.

* Stability: the SNRM downstream is not all over the place, starting out lower than the target and then drooping much further before wobbling up and down slowly, sometimes ending up down at 0.6 dB downstream from a 3 dB downstream target. Upstream is unchanged at a controlled and stable 6 dB staying close to the 6 dB upstream target.

[burakkucat]

* PhyR - I had forgotten about the very existence of this, didn’t know it had been deployed in BT 21CN, assuming that my understanding is correct, and didn't know who the mfr of the DSLAM at the NSBFD exchange was, which is crucial because I believe this is a Broadcom-to-Broadcom only thing? So if I have got this right then an extremely pleasant surprise - big reliability boost and possibly a speed boost too.

I believe that we have established -- within the last week or so? -- that the NSBFD DSLAM line-cards use a Broadcom chip-set.

[Weaver]

@Burakkucat indeed we have. ?I was wanting a sanity check though concerning my belief that ‘PhyR’ is a Broadcom-only thing, that was all. :-)

[spring]

Did you try disabling all QoS things in the router and turning the firewall off? [disabled state]

[Weaver]

Remember this is just a modem, not a router. QoS may work but I am not sure whether the firewall could be relevant? Could someone answer those two questions about a straight modem?

I'm assuming that the modem just copies everything through the PPPoE interface from one side to the other, bridging indeed. AA staff put some QoS entries into the config, but I don't know if that was just a mistake.

[spring]

Why does it matter to try, if it's switchable back on or reversable in config 

edit: so the reason i suggested it is, in the aa page they said the firmware might prevent static routes in web interface when QoS is on, in the context of a firmware bug. besides, you should clear d-rom and use factory zyxel config

[Weaver]

Quite true spring. I was just trying to understand, you made an important point.

[spring]

That is the reason nowadays from bios, windows, router config, even the network cable, its best to take the thorough and "best" route [and more often as time passes, the "safe" route. hardware has become a joke in terms of damaging itself even through software]

10 years ago that wasnt needed but times changed

ps if ssh still doesnt work for you it could be the new firmware. for me Busybox ["sh" in telnet, is that ssh?] worked but they removed some commands, same on other newer zyxel firmwares from other routers.

also try adding routes from supervisor first

[Weaver]

Second device has now arrived courtesy of ebay, all the way from Perthshire, as it happens. Pristine-looking in box as new, looks like it has never been used.

Pushed a copy of the XML config used by the first device into the new one and all was well. That is good to know, that I can just share config XML files with no changes. I practice I may want to make changes to the contents per modem, so as to set differing LAN-facing admin interface IP addresses

Custom 1508 MTU supporting firmware uploaded and flashed into the device. Cleared ROM-D. Thanks to Burakkucat for the magic incantation to do this. The supervisor password was discovered, again with the help of the Kuro Neko, and was the same as for the first device. Perhaps that is because I had already pushed the same XML config file into it?

All ready to install in place of the DLink modem and connect to the phone line tomorrow morning. Only thing remaining to do is to set the device’s admin IP address to be distinct so it doesn't clash with the other modems. (I am not using the sophisticated AA method of setup where all modems are access for admin on the same IPv4 address in their various scopes and you access them by http on differing TCP port numbers and the Firebrick router NAT does dest port rewriting/translation.)

Waiting on two more ebay modems from France. The wonders of parcel tracking info in the web tell me that they are languishing in Inbhir Nis until some idle or penny-pinching white van man man decides he will honour us with a trip across to the West Coast. The delivery services from Inbhir Nis can be very slack, need a rocket up their backsides from the customers who bring them all their business.

[banger]

Nice when someone takes care of their gear I have had a couple that looked as if they had never been used as they still had the sellophane on them.

[Weaver]

I intend to put this new arrival on the slowest line - in both directions - cwcc@a.3. For some reason this line lost 157 kbps or ~5.5% downstream rate on the 13th of April when it dropped from 2848 to 2691 kbps d/s sync rate, and its upstream sync rate also dropped from 499 to 436 kbps the former rate being unusually high and the latter rate being the norm for this line which is bad on upstream compared to the other two lines which have been achieving around ~500k and 566kbps upstream sync. Why this big performance drop happened I simply don't know. I can't see how a new interference source or crosstalk source could be responsible as this did not happen to the other two lines. Downstream later improved slightly to 2706 kbps sync.

I am hoping the new B10A will do a really good job on this line and it will be interesting to see if it remains worse than the others with the superior modem installed. Will the size of the downstream improvement be, or less the same as already noted on line cwcc@a.1? Alternatively, probably being unrealistically optimistic, there might I suppose even be a greater downstream improvement, if there is some kind of line problem and it is one that these superior modems alone can handle.

It will be very bad news if the upstream performance worsens with the B10A since this line has always been relatively weak in upstream performance apart from during a recent brief period mentioned earlier.

[Weaver]

The one big hole in the bits per bin at 2 bits instead of say 10 has moved from bin 79 340.6875 kHz to bin 57 245.8125 kHz when I repeated the measurement some days later.

If that is the ‘pilot tone’, then that explains why I can’t find likely noise sources despite lots of googling around for any mention these frequencies or those close by and a list of broadcast radio stations comes up with nothing.

I really don't understand pilot tones at all, or why it might vary.