Print

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

[Chrysalis]

I am sitting here wondering how a lift and shift fixes faults.

and of course why some customers manage to get one and others dont (inconsistencies).

Isnt a lift and shift akin to just eg. swapping a ethernet cable from one port to another? if so why is swapping ports having such affects on service? is the dslam equipment prone to leaking noise to adjacent ports (and cause crosstalk)?

To the OP if your attainable hasnt budged but sync moved it would suggest DLM is slowing the line down and the lift and shift (temporarily) reset the profile.

The bad news for you recoveries from line banding seem to be much slower than interleaving, DLM seems much more cautious to remove it.  Its also possible isp requested it, as I do believe isp's can request banding.

[mikehiow]

To answer your question why a lift and shift is used, you need to understand that not all engineers understand the technology completely.

Mine believed that the port was broken because he could only sync at 60mbps with his JDSU at the cab. I made a brief attempt to explain to him that's because it's banded and a DLM reset would be fine, but he was very insistent. He did leave me with a brand new HG612, though!

He didn't understand that I had line stats from the ECI in front of me and told me not to pay too much attention to speedtests, despite the visible max attainables, powers and SNRs on the screen.

A very helpful and nice chap, but I don't think he's been trained in the technology properly.

[mikehiow]

Another sync drop this morning, this time at 9am.

A massive drop in SNR proceeds it.

Full monty (24h)
http://vxr.in/share/line_stats__FULL_MONTY_P-20130723-1001.png

Current:
http://vxr.in/share/line_stats-P-20130723-1001.png

[Bald_Eagle1]

Oh dear 


I suppose a 'faulty' modem can't be ruled out, but just look how much US Line & Signal attenuation changed at this morning's resync.

Do you have any record of your usual US attenuation values?


Which part of the country do you live in?

Kitz has mentioned possibly the same issue from the Blackpool area.


BTW, the link to the current stats doesn't work.

[mikehiow]

Hmm, current stats problem is weird

I'm on the Isle of Wight.



Historic, although a little patchy:
http://eci.vxr.in/atn.txt

Fairly solid.

I'm wondering if perhaps it's something in the neighbours house as nothing has changed here since the issues begun, not that I can think of, anyway.

Any idea what could cause a drop in SNR like that?

[Bald_Eagle1]

Hmmm.

Looks like a definite physical problem to me (Hlog graph etc.)

Looking at your ECI attenuation stats & assuming they are comparable to the HG612's records, I am guessing the values of 0.1 for the U0 band was due to your connection not needing to use it previously.

Apparently the U0 band is primarily used by longer connections with much lower attainable rates.

It now appears that, for whatever reason, your connection is now attempting to use the U0 band.

Attenuation in that lowest frequency band is higher than the other US bands. It is usually the lowest (for those connections that normally 'need' to use it).

I presume your bitloading graph from the ECI modem also looked more healthy until quite recently (no gaps etc.)?

Are you 100% convinced your internal wiring hasn't been damaged or swapped around?

[mikehiow]

It's possible, like I said - I have a 5-10m run of CAT5 between the faceplate and the modem with RJ11 plugs crimped straight on solid core cat5.

I shall bring a socket back from the office today to go where the modem is and connect it to the IDC terminals on the NTE5/Faceplate end whist perhaps swapping pairs.

Although I wonder if it might be easier to just move the modem back to the NTE5 for now.

[mikehiow]

I wonder if I found the cause -

Moved the modem back to next to the socket and saw this (it was unreachable for nearly half hour before I remembered I had to create a route on the router now)


Full monty: http://vxr.in/share/line_stats__FULL_MONTY_P-20130723-1829.png

[Bald_Eagle1]

Early signs are indeed promising.

I assume that now US attenuation is much lower, the U0 band is no longer needed as DLM no longer thinks you have a long D-side?

Hopefully it isn't simply that you resynced at a 'quiet' time. 

If the connection remains steady, DLM may eventually decide to let you have the lost sync speed back again 

[mikehiow]

Plusnet have emailed me saying they want to book an engineer visit for another DLM reset since it dropped back to 60, so that shouldn't be an issue. I got a visit the next day last time.

I've simply asked to wait a few days to conduct more tests to ensure I've found the cause.

Hopefully it's just a case of re-terminating the CAT5 to put it back where it was. Of course, I guess it could be anything from the UPS or a PSU where it was located or even the PSU that was powering it (I used a new one when moving it).

When I try putting it back, I may relocate it further away from all the other bits and bobs.

[mikehiow]

Bald_Eagle1 - I only ask here as I can't seem to find the Modemstats thread.

Have you already or at least considered releasing the source code for modemstats?

I'd love to update my monitoring stuff to handle the HG612 while adding some functionality (Modemstats has been very useful to me, but I don't usually have a windows machine running 24/7) and having something to work off would be extremely helpful when it comes to figuring how to get all the data I want out of the HG612 (Kind of like the modified eDMT code I had for the ECI).

I'm more than happy to share my work when it's done

[Bald_Eagle1]

I'm due to release the first update to the Windows programs in the next few days.

Assuming feedback regarding reliability is favourable, I may well release the c source code soon, but it is currently very Windows specific at the moment.

It needs tidying up & possibly adding Linux compatibility back into it, but that is likely to take me quite a while.

[mikehiow]

Well I'm only interested in the harvesting bits, and even then mostly only the telnet commands and what you're reading back - it just saves a hell of a lot of research and testing when writing a module for my stats app for the HG612.

If you're willing to share, I'd be most grateful.

[mikehiow]

Well, it's starting to recover just two days after it dropped and less than a day since I fixed it.

Sync rose to 73999 at 0630 this morning.

It'll be interesting to see if it goes the rest of the way tomorrow.

[mikehiow]

Now recovered back to 79999 all by it's self

Just need Plusnet to give my profile a nudge on their on and I'm back to normal