Kitz ADSL Broadband Information
adsl spacer  
Support this site
Home Broadband ISPs Tech Routers Wiki Forum
 
     
   Compare ISP   Rate your ISP
   Glossary   Glossary
 
Please login or register.

Login with username, password and session length
Advanced search  

News:

Pages: 1 [2] 3 4

Author Topic: DSLstats pre-release version 6.3.1  (Read 12561 times)

roseway

  • Administrator
  • Senior Kitizen
  • *
  • Posts: 43432
  • Penguins CAN fly
    • DSLstats
Re: DSLstats pre-release version 6.3.1
« Reply #15 on: March 19, 2018, 10:56:08 AM »

I would like to flag a very minor cosmetic inconsistency.

In the configuration, all errors are set to be monitored. Under the top level "Errors" tab, there are secondary tabs for "CRC Errors", "FEC Errors", "SES", "ES per hour" and "Averages". The tick-box for "Logarithmic Y axis" is selected on the first three secondary tab screens.

For the "CRC Errors" and "SES" screens the Y axis starts at 0.1 but for the "FEC Errors" screen the Y axis starts at 1.

Would it be possible for the Y axis of all three screens to begin at 0.1 and, thus, be consistent, please?

Having reminded myself of how these graphs are dimensioned, I see that the minimum scale values are hard coded, not autoscaled. The reason for the difference of course is that FECs generally come in much larger numbers than the other types of error, and I wanted to reduce the number of gradations on the Y axis for better readability. (The connection speed graph can also be made logarithmic, and in this case the minimum value is 10).

As the maximum values on the various error graphs are autoscaled, and can vary widely, I'm not sure that making the minimum values all the same will help a lot. What do you think?
Logged
  Eric

jelv

  • Helpful
  • Kitizen
  • *
  • Posts: 2054
Re: DSLstats pre-release version 6.3.1
« Reply #16 on: March 19, 2018, 11:44:44 AM »

On a logarithmic scale a value of zero doesn't exist. The log of zero is minus infinity.

A value of 0.1 can exist in real life, because these values shown in the error graphs are per-minute values. For example, if you have one CRC in 10 minutes, that's 0.1 CRC/minute.

The problem is that it is not correct when in a sampling period there were no errors.

I don't know if you can do this, but could you set the minimum to 0.01 and then set the display to have only one decimal place which would round to 0.0?
« Last Edit: March 19, 2018, 11:50:54 AM by jelv »
Logged
Broadband and Line rental: Zen Unlimited Fibre 2, Mobile: Vodaphone
Router: Fritz!Box 7530

Westie

  • Kitizen
  • ****
  • Posts: 1596
Re: DSLstats pre-release version 6.3.1
« Reply #17 on: March 19, 2018, 11:54:13 AM »

On my line both CRC Errors and FEC Errors are low enough that I find a linear Y axis more readable, so I was unaware of the inconsistency. Having said that, my upstream FEC Errors disappear altogether if I choose a logarithmic Y axis for that graph.

I think G.INP is applied to the downstream more readily than the upstream, so maybe upstream FEC errors will always be an order of magnitude lower than downstream. (Is this correct?)

On balance then, I tend to agree with b*cat, although more from an accuracy of reporting viewpoint than a cosmetic one. However, my logic may be completely wrong - and I probably wouldn't notice if it was changed anyway!
Logged

roseway

  • Administrator
  • Senior Kitizen
  • *
  • Posts: 43432
  • Penguins CAN fly
    • DSLstats
Re: DSLstats pre-release version 6.3.1
« Reply #18 on: March 19, 2018, 12:48:35 PM »

Quote
I don't know if you can do this, but could you set the minimum to 0.01 and then set the display to have only one decimal place which would round to 0.0?

I could do that easily enough, although it would add an extra gradation on the Y axis (two gradations for the FEC graph), making the data a bit less readable. I really don't know what to do here.
Logged
  Eric

Westie

  • Kitizen
  • ****
  • Posts: 1596
Re: DSLstats pre-release version 6.3.1
« Reply #19 on: March 19, 2018, 04:13:19 PM »

In my opinion, the ideal logarithmic scale minimum is less than 1 (to avoid missing single errors) but equal to or greater than 0.1. Anything smaller than this is relatively insignificant, and would be better analysed by reducing the sample period (which is set by the user).

Explanation
A minimum graph scale of 1 does not identify single errors.

In the "Errors" graphs, an error value between 0 and 1 can only occur if the graphing option is set to "Per minute" and the sampling rate is set to greater than 60 seconds. Any other option would yield a value of either 0 or "greater or equal" to 1.

A value between 0 and 0.1 errors per minute can only occur if the graphing option is set to "Per minute" and the sampling rate is set to 10 minutes or more. I could be wrong, but I would have thought that anyone concerned about errors that low is likely to be sampling more frequently than every 10 minutes!

BTW, wouldn't changing the minimum on the logarithmic scale from 0.1 to 0.01 require two extra gradations on the graph?
Logged

roseway

  • Administrator
  • Senior Kitizen
  • *
  • Posts: 43432
  • Penguins CAN fly
    • DSLstats
Re: DSLstats pre-release version 6.3.1
« Reply #20 on: March 19, 2018, 06:27:15 PM »

I think you've put it in a nutshell there Westie. Thank you. Unless anyone has strong objections, I'm going to set the minimum value on all the error graphs to 0.1 when in logarithmic mode.

On your last point, I really meant factors of 10 when I referred to gradations.
Logged
  Eric

burakkucat

  • Respected
  • Senior Kitizen
  • *
  • Posts: 38300
  • Over the Rainbow Bridge
    • The ELRepo Project
Re: DSLstats pre-release version 6.3.1
« Reply #21 on: March 19, 2018, 07:41:28 PM »

In my opinion, the ideal logarithmic scale minimum is less than 1 (to avoid missing single errors) but equal to or greater than 0.1. Anything smaller than this is relatively insignificant, and would be better analysed by reducing the sample period (which is set by the user).

Explanation
A minimum graph scale of 1 does not identify single errors.

I was asked for my feedback, earlier, and have only just got around to reading the latest posts. Westie has typed the words which exactly explain my point of view . . . so much so that I just quote the first two paragraphs, above.

The sentence "A minimum graph scale of 1 does not identify single errors." is what was troubling me when I mentioned the inconsistency of the logarithmic Y axis, when the three (error) plots are considered.
Logged
:cat:  100% Linux and, previously, Unix. Co-founder of the ELRepo Project.

Please consider making a donation to support the running of this site.

jelv

  • Helpful
  • Kitizen
  • *
  • Posts: 2054
Re: DSLstats pre-release version 6.3.1
« Reply #22 on: March 19, 2018, 07:54:35 PM »

So I set my sampling period to 10 minutes and I use per sample. How do I tell the difference between a period where one error occurred and one where there were no errors?
Logged
Broadband and Line rental: Zen Unlimited Fibre 2, Mobile: Vodaphone
Router: Fritz!Box 7530

V_R

  • Reg Member
  • ***
  • Posts: 161
Re: DSLstats pre-release version 6.3.1
« Reply #23 on: March 19, 2018, 08:58:45 PM »

I would guess that the executable marker on the DSLstats executable has been lost in unzipping and transferring the files. In the RPi, open the file manager and point it to the location of the DSLstats files. Right-click on the dslstats executable and select Properties from the popup menu. In the Permissions section, set the Execute option to Anyone.

Sorted it! Thanks very much Eric. I'm still finding my way round this Raspberry Pi lark! :D
Logged
Zen Internet (80/20) on a Huawei 288 cab, via HG612 and an Asus RT-AC86U running Merlin FW.

Westie

  • Kitizen
  • ****
  • Posts: 1596
Re: DSLstats pre-release version 6.3.1
« Reply #24 on: March 19, 2018, 09:57:43 PM »

So I set my sampling period to 10 minutes and I use per sample. How do I tell the difference between a period where one error occurred and one where there were no errors?

You would not be able to tell the difference if the minimum scale value is 1, but you would if the minimum scale value is 0.1.
Logged

Westie

  • Kitizen
  • ****
  • Posts: 1596
Re: DSLstats pre-release version 6.3.1
« Reply #25 on: March 19, 2018, 10:01:40 PM »

Unless anyone has strong objections, I'm going to set the minimum value on all the error graphs to 0.1 when in logarithmic mode.

That gets my vote!  :thumbs:
Logged

NewtronStar

  • Kitizen
  • ****
  • Posts: 4898
Re: DSLstats pre-release version 6.3.1
« Reply #26 on: March 19, 2018, 10:03:44 PM »

No amount of logarithmic y scale is going to make my FEC's look any better than what the modem telenet data gives me.

one with it on and off

Logged

Westie

  • Kitizen
  • ****
  • Posts: 1596
Re: DSLstats pre-release version 6.3.1
« Reply #27 on: March 19, 2018, 10:30:14 PM »

No amount of logarithmic y scale is going to make my FEC's look any better than what the modem telenet data gives me.

I sympathise with you. :(

My first response to the linear scale (the lower picture) is "Wow! What happened at around 21:30?" My response to the logarithmic scale (upper picture) is "Wow! They look high ALL the time!"
On the other hand, at least it's FEC's not CRC's.  ;)
Logged

NewtronStar

  • Kitizen
  • ****
  • Posts: 4898
Re: DSLstats pre-release version 6.3.1
« Reply #28 on: March 19, 2018, 10:41:34 PM »

I sympathise with you. :(

It's not to bad during the day from 8am to 3pm it hovers around 3000 FEC per minute CRC spike errors used to shoot up to into the 48-68 have been able to get these down to 1-2 CRC's but still an errored second but no SES counts.
Logged

les-70

  • Kitizen
  • ****
  • Posts: 1254
Re: DSLstats pre-release version 6.3.1
« Reply #29 on: March 23, 2018, 11:38:40 AM »

  I am using the pi version of 6.3 and more recently 6.3.1.  Sorry to add some bother but I can't get the time per page to extend past 24 hours when a number of days is set.  The time per page stays at 24 hours and the hours setting moves to 1 hour after being set to 0 hours. It looks like the graphs may save after the number of days as they don't save every 24 hours when more than day is set, however I have mainly used 7 days and then messed about before 7 days so I have not tested what happens after 7 days.
Logged
Pages: 1 [2] 3 4