Print

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

[Weaver]

There is going to be a lot of confusion on my part that will be revealed in this post; apologies in advance, do put me straight as needed.

A question: if your are looking at the stats from your modem and you look at the CRCs count, if you have L2ReTx of some sort - PhyR in my case with my ZyXEL VMG 1312-B10As and ADSL2 - or G.INP, what happened when there is an error that is not FEC-corrected? That counts as a displayed ‘CRC’ statistic count, yes? But what if L2ReTX comes in and a retransmission successfully saves the day? The first transmission was unrecoverably corrupted, but after L2 retransmission, the damage is effectively repaired. How can you tell from the stats whether or not retransmissions are succeeding in overcoming ‘CRC’ corruption? Also how much data rate effective slowdown is there caused by L2 retransmissions?

[Chrysalis]

I think if it succeeds, a CRC isnt tallied.  I dont know what the actual counters are for retransmission though.

[tubaman]

I would tend to agree with Chrysalis but don't have any documentary evidence for it.
For G.INP my Zyxel gives these stats - I assume you get similar for L2ReTx?

Code: [Select]

Retransmit Counters                                     
rtx_tx:         284418942               7645                                   
rtx_c:          35889175                7639                                   
rtx_uc:         86462           0                                                                                                                                                       

G.INP Counters                                         
LEFTRS:         7664            87                                             
minEFTR:        44659           7781                                           
errFreeBits:    4204658280              155049192
, which Kitz explains most of at the bottom of this page - https://kitz.co.uk/adsl/linestats_errors.htm.
What I don't know is how the discrepancy between the corrected and uncorrected errors are accounted for.
There must be some slowdown for retransmissions, but it'll be much less than if they all had to be requested from source again.
 

[Chrysalis]

on g.inp if it has to retransmit you get the interleaving type delay, the difference mainly been that it only occurs when it has to correct and retransmit whilst on normal interleaving the delay is there for every packet.

[Weaver]

On my ZyXEL VMG 1312-B10A using ADSL2 and PhyR L2ReTX I don’t get all of those stats reported as Kitz listed on her page and as tubaman listed above: rtx_tx etc

I get merely

SF:      2555819      465110
SFErr:      326      9
RS:      57505933      48277
RSCorr:      137933      12689
RSUnCorr:   346      0

ReXmt:      181162      0
ReXmtCorr:   181118      0
ReXmtUnCorr:   373      0

Tubaman is I presume using VDSL2 in that set of reported stats?

[tubaman]

....
Tubaman is I presume using VDSL2 in that set of reported stats?

Correct.