Kitz Forum
Broadband Related => Broadband Hardware => Topic started by: Weaver on July 15, 2018, 02:15:59 PM
-
I tried to reset one of my modems using the 20 second pin reset. I was trying to put in a good XML config, it was not an issue of a bad firmware flash image, and I was hoping to get the configuration reset so I could load up a known good backup if the XML. A lot of failures were endured, I was using an iPad, which is all I have these days, and I think I hit a lot of bad bugs in Safari/iOS which are triggered when you are using a setup that has no dhcp server on the LAN and you have static rfc 1918 made-up addresses temporarily.
In the end I managed to ping the modem’s admin i/f at 192.168.1.1 and in a web browser I got the emergency recovery http-server in boot loader thingy. In Safari / iOS on my iPad I could not see any web page at all, so I tried a different iPad web browser, iCab. iCab gave me a half of a web page, the bottom half of the page was missing and there were 0ther things wrong with the display. I just don't have another machine at hand, tried three iPads. I have gone through the procedure successfully before, that is the weird thing, but the iOS version is now different, although I used an older iPad to rule that out. The firmware version is different, is our own johnson's latest jumbo v16-based one, dated late June. I did a recovery using Safari on an iPad with this very same emergency boot rescue http thingy.
Anyway, that modem is now effectively stuffed as I cannot out a good config into it.
The normal http server was no longer working after I fiddled around with the device using the CLI in telnet, trying to get DNS working properly and a few other things.
I could try the 30-30-30 reset, but I am completely exhausted now and cannot think straight.
Does anyone have any suggestions concerning de-bricking?
It does respond to pings and has some life in it as shown by the funny web page appearing, when you put it into the right state at any rate.
Otherwise I might appeal to a very generous volunteer to de-brick it for me?
-
I know that there is a set of header-pins on the PCB, which provides serial port access to the device's console.
I don't know the assignment of TxD, RxD, Gnd & Vcc to the four pins . . .
J1-1, J1-2, J1-3 & J1-5, reading left to right.
Please see johnson's post (https://forum.kitz.co.uk/index.php/topic,21956.msg378293.html#msg378293), below, where the details are now visible for all to see.
-
I would have to buy one of the "get console" hardware devices which can provide an ipad with an RS232-type interface.
Too much expense and too much physical pain and fatigue so I would just write one modem off as a lesson to be learned. Unless I can fiddle about again with the reset procedure and achieve some success, then maybe I should just give the modem to any kitizen who can fix it, FOC.
I don't know why the emergency recovery web ui seemed to be producing those half web pages.
-
So to clarify, you have not been able to flash a firmware via the bootloader recovery page?
The 20 second reset procedure should be clearing any settings you have changed, it does for mine.
On my 1312 it seems picky about when it serves you the recovery page, after the 20 seconds flashing red to solid I have to wait at least 1 minute before it starts responding with a web page on 192.168.1.1.
I have no idea why the page seems corrupted? I have not used the recovery procedure on the v16 firmware (not that firmware version should change anything, the recovery is from the bootloader) but I shall try in a moment to just to make sure.
Edit: There is also 1 port on the modem that will not respond with the recovery console, I forget which atm and need the serial console to find it easily, will get back to you in a minute.
-
I have a bricked 1312 and Iam TJ offered to de-brick it but he hasn't posted on here for a while. Mine was no red light on boot at all so it never got to the recovery web page.
-
So I just tried from the v16 jumbo frames firmware and the recovery is accessible as normal, the port that will not work is labelled 4 on the box. After swapping cables you have to wait a full minute or so before it becomes accessible again.
The full page looks like this:
(https://i.imgur.com/m9ekkhl.jpg)
Which part is corrupted on yours?
I have a bricked 1312 and Iam TJ offered to de-brick it but he hasn't posted on here for a while. Mine was no red light on boot at all so it never got to the recovery web page.
Did he tell you want was required to debrick yours banger? I assume more than just a serial console.
-
He didn't just said he had several and would have a look and it was probably a flash failure. He used to be a regular poster.
It bricked just after power cycling it while online with the power button but had become non responsive before the power cycle.
-
For reference the serial pin out is attached, excuse my poor editing skills ;D
No need to connect anything to to the 3.3V line in most cases, just ground tx and rx and ensure they are reversed, ie rx -> tx, tx -> rx to your serial converter.
-
I can only see the very topmost part of a page, the rest is either absent or covered up or clipped off.
-
The trouble is, you don't know if it's a GUI issue explicitly related to Safari.
Have you tried another browser from your I device?
-
I tried to send a PM Weaver but:
If we cant get the modem functional again (I'm sure with some perseverance we can). I would be happy to try my hand if you'd trust sending it to me. Shouldn't be much postage given it only needs to be the device itself.
But that's a later conversation, I know you have been through the browser mill before trying to troubleshoot things, but have you tried chrome on iOS? It might still be webkit but maybe it avoids whatever is corrupting the page in safari?
-
I can certainly do that Chrome thing. Safari showed no output, iCab showed the corrupted half display. But Safari has worked with this same emergency http thingummy in the past. So this suggests there is something wrong with the modem or it is a bad release of iOS or some of these apps, which is really unlikely given that they work otherwise.
If I can't get it going with a bit of handholding I would like to take you up on your generous offer.
The flags were set wrong it seems, so that is why you were unable to PM me.
-
So have you got any further with the recovery page? Was thinking, how are you connecting to the modem when you try and access it, do you have a lightning to ethernet adapter for the ipad?
-
I have several access points which are indeed wireless-to-ethernet converters. I just plugged the modem into my main LAN switch and set the iPad to a statically configured IPv4 address of 192.168.1.99 netmask 255.255.255.0 gateway 192.168.1.1 (not that it matters). Safari just did not work and iCab came up with this unreadable webpage. I think there are (once again) bad bugs in iOS and / or Safari when you use static addresses like this and / or when no DHCP server is accessible on the LAN even if it is not supposed to be used. These bugs do not help. I tried plugging in and unplugging the Firebrick from the main switch because the Firebrick is a DHCPv4 server and I wanted to see what effect this had on the iOS + Safari weird behaviour.
I have not done any more with the sad modem yet as I have been really ill and asleep all day due to the Fentanyl (ie heroine++) running out. But I will try now I am feeling a bit more human.
-
Hmm . . . I really think that you have explained the problem. When using the Broadcom firmware upload mechanism (which is basically from the boot-loader) it is essential to use a direct cable link between the terminal device and the VMG1312-B10A.
-
I do not understand. It has worked fine before using the exact same setup, using Safari and this emergency recovery mechanism. No wireless involved as far as the modem knows. :-)
Or have i misunderstood? :-)
-
I appreciate that it worked previously but I am surprised that it did so. It is the transit through the switch (or Firebrick or other intermediate LAN gubbins) that is causing the problem.
If (and I know your situation does not allow it) the VMG1312-B10A could be connected from its LAN1 port to any simple computer's NIC port via an Ethernet patch cable and the computer's NIC port is assigned a static IPv4 address in the 192.168.1.X sub-net (anything from 192.168.1.2 - 192.168.1.254) there should be no problem.
Rather than banging your head against the wall (and disturbing the sleeping kittehs), why not accept the offer that (I understand) has been made by fellow Kitizen johnson?
-
I guess there is no reason it shouldn't work over a switch and access point, as you have proven before, as long as there are no interfering DHCP servers and the addresses do not conflict.
I had thought maybe the "192.168.1.99" was the problem (the bootloader suggests the host should be 192.168.1.100) but I have just tried with my 1312 and the recovery page is still accessible from a machine with 192.168.1.99 address just fine.
What does the rest of your address space look like Weaver? I guess switches are pretty agnostic to different subnets, but what about the wireless access point?
It certainly adds some unknowns and possible faults. A direct connection would be the first thing I'd try to rule them out, though I guess thats easier said than done with your setup.
Edit: and of course the offer still stands if posting isn't a problem :). I would say though, lightning to ethernet adapters can be had for around £10 and might prove useful for other scenarios as well.
-
> It is the transit through the switch (or Firebrick or other intermediate LAN gubbins) that is causing the problem.
I am missing something in what Burakkucat is saying. Everyone who uses an access point and an ethernet switch has the same setup. Switches just forward packets and so do access points and do not cause problems normally. The switch passes packets reliably all the time and so does the AP, it is used routinely for all LAN and internet traffic with no evidence of problems and the emergency recovery thing is no different to anything else in ethernet terms, and it worked before, but my software has changed on the iOS device and in the modem. The switch and WAP are not L3 devices: they have no knowledge of eg IP and just handle any ethernet frames agnostic wrt ethertype as far as I am aware.
The modem when set up as it was went through the main LAN switch, if you thought for some reason that my main HP switch were broken then I could bypass it by going straight into the WAP.
There is no reason why any piece of converter electronics should be more reliable than a WAP that I use every day all day. I could swap the AP and the switch out if there were evidence of general problems.
The value .99 was just something I made up, I used various values, anything that was unique so as not to collide and was in the correct /24 assuming that that was what the modem was expecting (not a /16 as specced for RFC1918).
My feeling is that I have already proven that this is all down to long-term bugs in Safari and possibly some other bugs besides. The fact that the web page appearance is different between the two web browsers iCab (partial, corrupted) Safari (no page) is the basis for this.
Burakkucat is right, there is little point in me killing myself over this, and getting into more trouble with Mrs Weaver hiring her to swap cables around.
I would love to take Johnson up on his more than kind offer. Burakkucat has helped me to set the forum's messaging system flags correctly so johnson should now be able to PM me.
-
I was trying to make it clear that the Broadcom boot-loader firmware-update page should be regarding as something "special". It is a situation where VMG1312-B10A is not in its normal mode of operation.
I have no problem with your LAN configuration when the VMG1312-B10A has been allowed to boot normally.
-
Now fixed, after inspection by our very own Johnson.
-
It's good to know that the problem has now been resolved. :)
-
Perhaps Johnson could report on his findings?
-
It was basically a no fault found. The thing had stopped sulking by the time our good friend got it. I was so utterly stupid, that despite Burakkucat’s telling me to give it a five minute no-power siesta, I was in such a muddle that I forgot to do this essential recheck later on, before my wife posted it to johnson. Duh. What a moron. There was also a lot of buggy stuff going on with iPad Safari and manually configured RFC1918 IPv4 addresses plus no dhcp server - a combination that seems to create all kinds of screwy behaviour with Safari. That all added yet more chaos into the mix too.
-
Johnson spotted something in my config. There is something odd about it, to do with interface groups or the lack of them. It may be that the device as configured does not work properly if the ethernet cable to the router is plugged into LAN4 so if this theory is correct, it was just bad luck before, due to janet picking socket LAN4 without us realising the significance of it.
-
It does seems like this may be the culprit.
For some reason Weavers base config, a slightly modified one from AAISP, has the default interface group missing LAN4.
For reference heres mine:
(https://i.imgur.com/uImPIY5.jpg)
Notice LAN1 is missing from the Default group as its assigned to the VDSL connection, in Weavers the Default list is missing LAN4, but with no second group owning it.
Maybe this is due to a previous other interface group that has since been removed? Not sure.
LAN4 is also non functional when attempting to access the recovery page after booting into recovery mode. With a serial console attached plugging in a cable to LAN4 results in messages saying the port has been disabled.
This would seem to explain both the loss of connectivity to the normal web interface and the problems accessing the recovery page. But as Weaver says, still just a theory, I guess a test with one of the currently functional modems changing the cable to LAN4 would be a good way to prove or disprove.
Edit: In case its useful as the labels are tiny and virtually impossible to see if you have the unit wall mounted, LAN4 is the furthest from power in and closest to the phone cable in.
-
Guys, should we try and fix this?
-
Mine appears to have the LAN group including all LAN sockets, maybe an AA foible?
-
Guys, should we try and fix this?
Fix what?
For some reason Weavers base config, a slightly modified one from AAISP, has the default interface group missing LAN4.
Just don't use any AA settings. You have modified firmware and won't be relying on AA for updating your modem.
Most people make an effort to remove any ISP config from their Zyxel modem. As you bought these retail I have no idea how you have for AA settings in amongst your config.
-
I should have looked at the standard default config that the brand new ones come with, for inspiration.
Rationale: not entirely rational. [Remember I am completely full of drugs. ??? ]
I bought one from AA ages ago when they still had some stock and that was really cheap. I specifically wanted to use the AA config, although with Burakkucat’s help I did disable the TR-069 remote updating, because as you said, I have custom firmware.
AA staff did the donkey work so that it just worked immediately out of the box in modem-only mode as I asked, as they set up the basic ADSL settings.
I also have a slightly tuned config in other respects: (1i) going on a tip from Burakkucat, I have locked it to G.992.3 only, disabling auto-mode because I do not want it to select G.992.5. This optimistic tweak is in hope of slightly higher performance by reducing the input frequency range window and so reducing the total noise ingested. That is the idea, but it does not seem to deliver any improvements from the evidence of experiments. Since it doesn't hurt, I have kept it in though. (2) I also have slightly reduced overhead by changing from RFC2684 LLC to the more efficient VC-MUX, which I can safely do because I am not on 20CN whereas AA would have to do two configurations.
I hoped I could get internet access to the modem going, behind my usual firewalling, but have not got that far. I wanted to get NTP to work and DNS and also ideally I wanted to make it really easy for AA staff to directly log in and have a peep at stats without me having to dig them out and send them to AA, cutting out the round trip. For that reason I wanted some other AA customisations such as correct IP addresses set for AA's NTP server, DNS servers etc.