Print

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

[burakkucat]

Actually, the QLN graph from HG612 modem stats shows a very odd and quiet line

Yuck is all I can say. 

Very odd. I would like to be able to explain what we are seeing . . . but I am unable to do so.

Am I correct in believing your G.993.2 service terminates on a (cabinet installed) Huawei MSAN?

[j0hn]

Indeed that's a huge difference from your usual QLN on MDWS. Possibly a crosstalker is offline or has ceased their service.
Vectoring can also cause such symptoms, particularly the better use of bits/tone in D1 and D2. Less likely but worth  checking the vectoring state anyway.

[Sheepie]

yes my service terminates at cabinet - Huawei MSAN, some 270m away.

Vectoring is not enabled.

I thought maybe it was a power cycle at the cab and then me be one of the first to sync, but wouldn't the max attainable drop back to what it was previously (or there abouts).

I need to compare actual QLN values direct from the modem to see which stats app is correct, DslStats and HG612 stats look very different even though I think they are meant to be based on "adsl info -qln"

I might have to resync manually before the DLM does, packet loss is becoming a problem as it is at the moment.

Edit: MDWS Quiet Line stats are incorrect, uploaded from DslStats, it is not showing current values from info --QLN, some example tones:
   2000         -127.5000
   2001         -127.5000
   2002         -127.5000
   2003         -127.5000
   2004         -127.5000
   2005         -127.5000
   2006         -127.5000
   2007         -127.5000
   2008         -128.0000
   2009         -128.0000

[j0hn]

I thought maybe it was a power cycle at the cab and then me be one of the first to sync, but wouldn't the max attainable drop back to what it was previously (or there abouts).

Yes, that's the usual behaviour. The upstream SNRM dropping 3dB at the same time suggests it's neither a simple DSLAM reboot or a lower downstream SNRM target set by DLM.

A before and after QLN might help work out what it is, we may never know though. A missing crosstalker or a physical change (improvement) to the line are just 2 of the umpteen possible causes. There's no change in attenuation or reported power levels either so it certainly leaves one scratching their head as to what has occurred.

[Sheepie]

Just rebooted the modem and it's back to 6db and normal looking QLN, hopefully lower ES and packet loss too!

It's got to be a power cycle at the MSAN, incorrectly working out bit loading based on QLN before anyone else tried to sync (or before most people anyway).

[burakkucat]

Yuck is all I can say. 

The above comment was based solely on sight of the QLN plot. The DS bands had become very "deep" (in a vertical sense) whilst the US1 (or was it the US2?) appeared to have a sine wave superimposed upon it.

I never looked at any of the other plots or data . . . so I shall have to rely on j0hn's analysis. 

[displaced]

I had a remote-triggered resync (RDI) early this morning at 00:58.  First one I've ever seen.

Perhaps it was the cab being updated with the new DLM settings? One can only hope!

(I'm on the Gravesend exchange, which I stubbornly refuse to call by it's new 'Ebbsfleet' name.  Ebbsfleet's not even a real place -- it's just a roundabout and a train station.  Grrr.)

[edit] I should probably add that my target SNRM is still currently showing as 6.0.  After the resync I gained a handful of kilobits in either direction, 1.5Mbit DS above my reported attainable, which is normal for the line -- it always syncs a bit quicker that it says it can!

[BigJ]

I had an RDI today at 00:58 too. Over the last few days I've been watching my downstream SNRM stepping downwards in 1dB increments, 6.4, 5.4, 4.4 but instead of 3.4 it dropped to 1.8dB. That seemed strange but I was surprised to see my upstream drop by about 2dB along side it. No idea what this signifies.

attached image from MyDSLWebStats for easy reference.

EDIT: Rebooted the modem and SNRM is back at 4.4dB. I guess my modem went a little wonky during a DSLAM reboot. Bad modem, bad!

[burakkucat]

Looking at the "small print" in the upper section of your SNRM plot, I see that the synchronisation speed is (essentially) 80/20 Mbps DS/US. I would not expect to see much of a fall in terms of the SNRM value purely because you are at the upper limit (synchronisation speed) for the product.

Any SNRM below 3 dB should be regarded with suspicion. If such a state is maintained, I would predict that a remote (DSLAM) initiated re-train of the G.993.2 link will eventually occur.

Without studying your circuit's previous history, I shall assume that a 4 dB target SNRM will be the preferred minimum.

[displaced]

Oooo.

Fingers crossed that I see an SNRM drop overnight tonight!  My line's reasonably well-behaved.  No SESes and only a handful of ESes every day, and the SNRM hovers around .5dB of the target most of the time. 

Let's hope the DLM God (I call him 'delemicus')  has recognised my line's piousness.

[j0hn]

BigJ your line is a 100% perfect example of the new xdB DLM target SNRM. It shows every characteristic I would expect to see.

1st is that only the downstream is affected. Upstream will show the usual slight variation but should remain roughly the same.
2nd is that the line doesn't jump straight to 3dB. Every example I've seen show the line should drop in 1dB stages.
3rd is that the max attainable rises with the drop in target SNRM.

With a lower SNRM caused by a DSLAM reboot it usually affects both downstream/upstream. The max attainable usually drops back down after all your crosstalkers reconnect. It also doesn't go 1dB intervals.

Your line appears to be on a 4dB target SNRM at the moment. If DLM considers that your line is coping with the current target then I would expect the line to resync either tomorrow or the 16th between 7-8am with a 3dB target. Let's see if that prediction comes to be 

[BigJ]

Thanks for the info guys. I'm slowly starting to get a grip on how this all works.

Hope Delemicus answers your prayers displaced Great name by the way

Hmmm, I'll take a punt on it dropping to 3 tomorrow!

EDIT: spelling. removed question about exchange as displaced answered it already.

[Dray]

I see you're not using the "default" HG612 

[WWWombat]

I see that the synchronisation speed is (essentially) 80/20 Mbps DS/US. I would not expect to see much of a fall in terms of the SNRM value purely because you are at the upper limit (synchronisation speed) for the product.

Looking at the full process on MDWS, it looks like the line only reached 80/20 speeds because of the final event in the sequence - jumping from 65/20, to 69/20, to 72/20, and finally to 80/20.

The first two look like bona-fide steps in the 5-4-3dB process, but the 3rd looks like an unfortunate DSLAM reset. A close zoom into the SNRM graph shows the first reading after that resync is at 4.9dB at 01:00. 1 minute later, it has dropped to 1.7dB. The knock-on effect is obvious: A massive increase in FEC rates, which fortunately didn't translate into many retransmissions or CRCs. Thankfully, this was stopped by the manual resync back to 4.4dB.

This event likely shows the biggest risk to the 5-4-3dB process: A DSLAM reset can easily knock 3dB lines out of any decent behaviour - but hopefully it gets seens as a "wide area event".

[BigJ]

Thank you for the insightful comment. As suggested, could this odd behaviour be explained by me using a Billion 8900AX instead of a HG612? The HG612 is just more resilient.

Also, if I hadn't done a manual modem reboot, would DLM have eventually noticed the SNRM was way off target and triggered a re-sync?