Print

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

[William Grimsley]

I know they're low, but the errors spike dramatically in the space of a few minutes and then the line drops out so something's wrong.

Surely, if the line is dropping out away from the night then there's something wrong and it's not a DLM action?

Also, the Last Retrain Reason was 2 which is a LOS which is slightly concerning.

[kitz]

As I said the other day, its not unusual to see CRCs (OH Frame Errors) during the sync up process.
I had an adsl2+ line that never got any CRCs - except during sync up and thunderstorms.   

I dont know for certain, but a theory is they could happen during the Initialisation process, probably during channel analysis and before DLM profile is applied.

[William Grimsley]

As I said the other day, its not unusual to see CRCs (OH Frame Errors) during the sync up process.
I had an adsl2+ line that never got any CRCs - except during sync up and thunderstorms.   

I dont know for certain, but a theory is they could happen during the Initialisation process, probably during channel analysis and before DLM profile is applied.

I understand that. Sorry, if I'm not making sense. But, the OH Frame Errors are not occuring during sync, they are happening before the sync. Anyway, why would my line drop out when DLM didn't do anything?

[kitz]

Surely, if the line is dropping out away from the night then there's something wrong and it's not a DLM action?

No.   DLM performs resyncs at any time of the day now.   When my line was in DLM recovery it would usually resync at about 11/12 o'clock (just before lunch)  - that was well over a year ago. 
Since then Ive seen DLM force retrains later than that.   The night time only rule no longer applies to DLM.   

[William Grimsley]

Surely, if the line is dropping out away from the night then there's something wrong and it's not a DLM action?

No.   DLM performs resyncs at any time of the day now.   When my line was in DLM recovery it would usually resync at about 11/12 o'clock (just before lunch)  - that was well over a year ago. 
Since then Ive seen DLM force retrains later than that.   The night time only rule no longer applies to DLM.

Ah, so now DLM works on the fly for FTTC like ADSL? I'd just like to clarify, the Last Retrain Reason of 2, is that DLM?

[kitz]

But, the OH Frame Errors are not occurring during sync, they are happening before the sync.

How do you know that they aren't occurring during initialisation.   Are your stats on MDWS?
Channel Analysis occurs before setting the sync speed, so they will be there showing as soon as line comes up.  They could even occur during the exchange process which is when sync is set.
This is the only reason why I could think my adsl2+ line would repeatedly show a tiny amount of CRCs each and every time a resync was performed.   This behaviour could be repeatedly observed without deviation over about 6 years.     

Anyway, why would my line drop out when DLM didn't do anything?

Cant answer that one

[Moderator edited to fix a broken BBC [/quote] tag.]

[William Grimsley]

But, the OH Frame Errors are not occurring during sync, they are happening before the sync.

How do you know that they aren't occurring during initialisation.   Are your stats on MDWS?
Channel Analysis occurs before setting the sync speed, so they will be there showing as soon as line comes up.  They could even occur during the exchange process which is when sync is set.
This is the only reason why I could think my adsl2+ line would repeatedly show a tiny amount of CRCs each and every time a resync was performed.   This behaviour could be repeatedly observed without deviation over about 6 years.     

Anyway, why would my line drop out when DLM didn't do anything?{/quote]
Cant answer that one

The reason is in the above screenshots. One shows since link time 0 OH Frame Errors have occured. But, I may be wrong. They are possibly occuring during sync up.

[kitz]

Ah, so now DLM works on the fly for FTTC like ADSL? I'd just like to clarify, the Last Retrain Reason of 2, is that DLM?

No it doesnt work on the fly.   
What I think is more likely, is that as more people have got FTTC, then its gradually spread out to later in the day.
DLM cant resync all the lines at the same time.   The first time I noticed an afternoon resync was soon after g.inp was introduced.

[kitz]

[Moderator edited to fix a broken BBC [/quote] tag.]

Gosh you were quick off the ball with that one b*cat.   Thanks.
the { just goes to prove that I type BBC codes rather than use the buttons 

[William Grimsley]

Ah, so now DLM works on the fly for FTTC like ADSL? I'd just like to clarify, the Last Retrain Reason of 2, is that DLM?

No it doesnt work on the fly.   
What I think is more likely, is that as more people have got FTTC, then its gradually spread out to later in the day.
DLM cant resync all the lines at the same time.   The first time I noticed an afternoon resync was soon after g.inp was introduced.

Oh, I see. Now, that is interesting.

[burakkucat]

Gosh you were quick off the ball with that one b*cat.   Thanks.

You, of all people, should know that kittehs can move at speed close to the velocity of light . . . if the inclination (or motivation) is present! 

[William Grimsley]

Gosh you were quick off the ball with that one b*cat.   Thanks.

You, of all people, should know that kittehs can move at speed close to the velocity of light . . . if the inclination (or motivation) is present! 

 

[William Grimsley]

I've just remembered something that may be important. The last time my BT Home Hub 5 (Type B) had a DLM resync was 04:11 so I'm pretty sure my DLM resyncs are still happening early in the morning.

[Chrysalis]

What I find unusual is how high the SES is vs the ES.

But over 20 SES whilst banded AND interleaved, suggests you never going to have much joy on a unbanded fast path profile without g.inp.

note 20 ES is pretty much nothing but 20 SES isnt nothing.

[William Grimsley]

What I find unusual is how high the SES is vs the ES.

But over 20 SES whilst banded AND interleaved, suggests you never going to have much joy on a unbanded fast path profile without g.inp.

That's what struck my attention as well (like everything else). G.INP is enabled and I've never had a fast path profile so I'm not too concerned.