Print

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

[roseway]

According to Kitz' errors page http://www.kitz.co.uk/adsl/linestats_errors.htm the HEC count is usually the count of HECs which have not been corrected. I'm afraid I can't explain the inconsistencies - as far as I know, the values reported by DSLstats do reflect the values reported by the modem/router, but because different modems report the error figures differently, it's quite possible that I haven't covered all the variations.

For the record, I read SFErr or OHFErr for the CRC values, RSCorr or FEC for the FEC values, and HEC for the HEC values.

[les-70]

  Thanks for that. I was confused again by differing time periods  I think I have it all sorted now.  With a near a simultaneous modem reboot and dsl stats start, things make sense.  My most of my errors seem to be in a few one second bursts a few times a day (no idea what is causing it)  so it is easy to have very different reporting of stats if dsl stats has a different time frame.

[renluop]

For a time this morning the error averages for CRCs and ESs were running absolutely in parallel, i.e. each were showing the same number per day. All is normal now. Peculiar!?

[roseway]

A bit peculiar certainly, but probably just coincidence, I guess. ES includes all types of uncorrectable error, not just CRCs, but there can be multiple errors within one ES. So it's quite possible to get a situation where these two factors balance out, and CRC = ES.

[zhadoom]

Hi Roseway,

Looks like there are a smal bug in pbParams tab. Negative values shows without dash symbol.
I double checked executing "xdslcmd info --pbParams" in HG612 .

From DSLstats 4.44:

Code: [Select]

xdslcmd info --pbParams
xdslcmd: ADSL driver and PHY status
Status: Showtime
Retrain Reason: 0
Max: Upstream rate = 11320 Kbps, Downstream rate = 63668 Kbps
Path: 0, Upstream rate = 5990 Kbps, Downstream rate = 54998 Kbps

Discovery Phase (Initial) Band Plan
US: (872,1203) (1972,2779)
DS: (32,867) (1208,1971) (2784,4051)
Medley Phase (Final) Band Plan
US: (872,1203) (1972,2779)
DS: (32,867) (1208,1971) (2784,4051)
       VDSL Port Details       Upstream        Downstream
Attainable Net Data Rate:      11320 kbps         63668 kbps
Actual Aggregate Tx Power:        6.9 dBm          13.7 dBm
============================================================================
  VDSL Band Status U0 U1 U2 U3 D1 D2 D3
  Line Attenuation(dB): N/A 37.0 54.3 N/A 17.8 45.3 69.0
Signal Attenuation(dB): N/A 36.7 51.6 N/A 17.8 45.3 69.0
        SNR Margin(dB): N/A 14.9 14.8 N/A 4.9 5.1 5.3
         TX Power(dBm): N/A 5.6 0.9 N/A 12.5 7.7 15.0
From telnet:

Code: [Select]

xdslcmd info --pbParams
xdslcmd: ADSL driver and PHY status
Status: Showtime
Retrain Reason: 0
Max: Upstream rate = 11320 Kbps, Downstream rate = 63668 Kbps
Path: 0, Upstream rate = 5990 Kbps, Downstream rate = 54998 Kbps

Discovery Phase (Initial) Band Plan
US: (872,1203) (1972,2779)
DS: (32,867) (1208,1971) (2784,4051)
Medley Phase (Final) Band Plan
US: (872,1203) (1972,2779)
DS: (32,867) (1208,1971) (2784,4051)
       VDSL Port Details       Upstream        Downstream
Attainable Net Data Rate:      11320 kbps         63668 kbps
Actual Aggregate Tx Power:        6.9 dBm          13.7 dBm
============================================================================
  VDSL Band Status U0 U1 U2 U3 D1 D2 D3
  Line Attenuation(dB): N/A 37.0 54.3 N/A 17.8 45.3 69.0
Signal Attenuation(dB): N/A 36.7 51.6 N/A 17.8 45.3 69.0
        SNR Margin(dB): N/A 14.9 14.8 N/A 4.9 5.1 5.3
         TX Power(dBm): N/A 5.6 0.9 N/A 12.5 7.7    -15.0

[roseway]

Thanks for reporting that. It will be fixed.

[burakkucat]

b*cat was sleepy-eyed staring at the CRC & FEC Error graphs when suddenly he noticed that something "just didn't look right" with the y-axis of each graph.

Consider the CRC graph, attached below. Surely it is a physical impossibility for there to be fractional CRCs per minute? I'm certainly not seeing a constant 0.1 of a CRC per minute for the upstream frequencies! The (logarithmic) scale should start at zero, then progress to 1, 3, 10, 30, 100, etc.

Now consider the FEC graph, attached below. Again I'm certainly not seeing a constant 1 CRC per minute for the upstream frequencies. It is, of course, a value of zero.

[roseway]

That's what you get with a logarithmic scale. Those values are actually zero, but of course zero doesn't exist on a log scale. If you switch back to linear scaling you should see the actual values. I suppose I should really not plot those values at all when log scaling is in use.

[burakkucat]

Hmm, yes . . . I see it is a bit of a problem with a logarithmic scale.

Perhaps the (logarithmic scale) y-axes of both the CRC & FEC graphs should start 1, 3, 10, 30, 100, etc?

[roseway]

The difficulty with that is that, depending on the sampling period, it's quite possible to get fractional 'per minute' values. If, for example, sampling is at 2-minute intervals, and there's just one CRC in a sampling period, that equates to 0.5 CRC/minute. If the Y axis started at 1, and CRC values were very low, it would give the impression of a continuous 1 CRC/minute. This isn't the case with FECs, because the values are generally much higher (or zero if on fastpath).

What I'm more inclined to do is leave the scaling as it is, but whenever there are two successive zero values, nothing is drawn between them. This would avert the issue which you showed in your first pair of snapshots.

But I'm always open to any opposing views.

[burakkucat]

I can't see an obvious solution. (Other than not to use the logarithmic scale for the y-axis.) 

[roseway]

I've tried the approach which I proposed above, as in the attachments. The first one is with a log scale, the second linear.

[burakkucat]

The first example (logarithmic scale) looks good. And having read your reasoning for fractional CRCs per minute, I can accept the 0.1, 0.3, 1, 3, 10, etc, of the y-axis.

The second example (linear scale) now looks odd.    To me, it is indicating that there are never zero CRCs per minute recorded. 

[roseway]

OK, thanks, perhaps I'll apply the style only to the logarithmic scale.

[burakkucat]

If you are happy and nobody else disagrees, then I think that will be the best compromise.