Print

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

[Weaver]

Superb. I have regained access to the normal web server !!

Thank you so much to all who have helped me. And I also have a reminder of Burakkucat’s serious reset procedure for the future.

But this easy reset procedure could all be done remotely with having to beg Mrs Weaver for help with physical intervention

[burakkucat]

A purrfect result. 

[Weaver]

 

Not so fast, I’m afraid. I loaded up the latest config and stuffed everything up; I now have no access of any kind - no http on port 80, no telnet and not even Johnson’s http 8000 api works properly - can’t get stats from it any more on port 8000. So with no telnet it will have to be the full Burakkucat reset.

Damn - I went a long way backwards.

I need to difference that config against other versions too. Perhaps I should export a config from another modem and then difference that against what I’m using for this modem #1 - to see if I’ve done something really stupid.

A thought - maybe that latest config has evil in it backed up, the evil I introduced by hand is stored in that config!

I difference the config taken out of a good modem, and cannot see the problem.

I noticed that I cannot even ping the sickly modem #1.

It is however working as a modem, doing its job ok.

[Moderator edited to merge successive "tweet-like" posts.]

[johnson]

What are the differences between the known good and one you uploaded?

I assume just IP? Super basic thought would be the sickly modem is no longer at 192.168.1.1 if it doesnt even respond to pings, but on a conflicting IP to one of the other modems.

[Weaver]

There are quite a few differences, 53 differences in all: one or two which I don’t understand; encrypted passwords are different; lots of things to do with Wi-Fi as well; one thing to do with DSL config; some missing entries to do with VDSL profiles which would appear unimportant - I remember seeing them in the UI; there is one entry concerning versioning that has a zero length content in the one version and content which is the text string "(null)" in the other; there are a couple of dates which are different.

I have searched for all IPv4 addresses using a regex, nothing obviously wrong. There is a MAC address associated with the WAN i/f and that is different.

The only other address in that LAN is .254 which is the firebrick.

Can’t even ping it from my iPad. Nor can the firebrick ping it directly either.

I can either wrestle with this one with the Burakkucat full reset procedure and see what happens or I can pull a backup one and program that.

Perhaps I just did something really stupid like loading up the wrong config file, one for the wrong ‘slot’, which would put it in eg 192.168.2.1 instead of .1.1. Perhaps picked the wrong file with my poor eyesight.

The two files I’m considering are here attached.

[Moderator edited to merge successive "tweet-like" posts.]

[Chrysalis]

The other modems have stats firmware and are fine. The modem that is giving problems was ok until I fiddled around with it, doing medical experiments in the CLI and at some point I made it unhappy; I was trying to fix the networking config so it had a proper defined default gateway and sort dns out and so on. There is no way I can see that it’s anything to do with Johnson’s enhanced custom build.


I’ll try this good tip for recovery when I have got Janet available to help me then.

ok thanks for clarifying.

[Weaver]

There are other very minor differences: one date is in the distant past, a date in ISO format is terminated with a Z and has no time zone offset given in it.

It makes me wonder if there is some possibility that a very minor syntactic transgression, or illegality in the content of some field could crash the thing on startup when it digests the config file?

I noticed that you can specify four IPv4 addresses for the modem’s own admin i/f; I wonder if I could exploit that by having every modem at every one of four addresses. I would need to ensure that the firebrick’s own address in those lans is within the intersection of all of the netmasks. I’m not sure where the netmask is specified and I’m wondering if it defaults to /24. If so that’s not helpful with things laid out as they are. If I could set the netmask to something wide enough then that would be good. I don’t know what is going on with the IPv4 config in the thing. Maybe there are other XML elements which we should be using but aren’t currently.

When I try again with the full reset procedure or a fresh modem, I’ll use the config that is known-good pulled out of happy modem #2, I’ve converted all the addresses in that good modem #2 config file appropriately to 192.168.1.* for modem slot#1 instead of .2.* (using a thoroughly tested Shortcuts iPad program which uses regexes and whose results have been diffed to check them).

[burakkucat]

b*cat nods in acknowledgement.

I have downloaded the two configuration files and compared them using a "sideways diff". Basically --

sdiff -s -w160 config-file-1 config-file-2 | less

I can see nothing intrinsically wrong . . . 

[Weaver]

I will get hold of sdiff for the Pi / Debian Buster if I can then. I used the Kaleidoscope app for iOS as a sideways diff, which is very nice - not as powerful as Examdiff Pro which I have used a lot though.

[burakkucat]

You could, of course, just use the -y flag to the standard diff utility --

diff -y config-file-1 config-file-2 | less

[Weaver]

So I’m wondering - either there’s something intrinsically odd about modem 1, or else the config file (attached to an earlier post) is just plain bad and must crash the modem.

Perhaps I should just program a new modem?

I wonder what the story is about the MAC address stored in the config file?

[Weaver]

I keep begging Mrs Weaver for some help with the physical business of resetting a modem using Burakkucat’s procedure given at :
    https://forum.kitz.co.uk/index.php/topic,20731.msg361030.html#msg361030

I don’t really want to replace the firmware though as it already has a version that I’m quite happy with.

Is there a way that I can skip that part of the above procedure?

Or will I be unable to regain control over it without going through the ritual in its entirety?

I want to put it into a known state regarding config. A though occurs to me. I have not even power-cycled it yet. I wonder what that would do? One way to find out. I can beg Mrs Weaver to do that for me at least. As things are, the main port 80 http server isn’t working, and the Johnson stats server on port 8000 is, I think, working but not properly - as I get a short prompt via http on port 8000, but the full port 8000 graphing functions are not working, a failure which I suspect is possibly due to a dependency no longer working.

[burakkucat]

I don’t really want to replace the firmware though as it already has a version that I’m quite happy with.

Is there a way that I can skip that part of the above procedure?

Or will I be unable to regain control over it without going through the ritual in its entirety?

I want to put it into a known state regarding config.

I mentioned a simplified procedure in the second paragraph of Reply #6, above. To reiterate --

If I am correctly remembering what johnson mentioned, the act of taking the VMG1312-B10A to a solid red power LED is sufficient to revert the device back to the default configuration of the currently loaded firmware image.    That should be all that needs to be done as the current firmware image is good, the mangled configuration is not good.

[Weaver]

Brilliant - I just overlooked that at the time. Thank you very much indeed.    Free of pain now and so much clearer in my mind.

[Chrysalis]

In same boat I think :/

The one I got from ebay, solid red light.

Got in with reset pin