Technical Discussion
  >> Technical Issues


Register (or login) on our website and you will not see this ad.


  Print Thread
Standard User WilliamGrimsley
(experienced) Wed 04-Oct-17 10:21:59
Print Post

DSL Stats Query


[link to this post]
 
Hi guys,

I don't know if anyone else has noticed this, but on DSL Stats, especially on the US I've noticed that it tends to forget to plot CRC's reported by the telnet interface. For example, since last reboot there has been 2 US CRC's and 2 US ES.

On the telnet interface it says that, but on DSL Stats it says 1 CRC on its own stats page which means it only plots that on the relevant graphs. This means that t MDWS (MyDSLWebStats) then only shows 1 US CRC and ES, when there's
actually been 2 US CRC's equalling 2 US ES.

A very minor issue, but just wondering if anyone is aware of this occuring on their end?

Thanks!

William Grimsley.

Edited by WilliamGrimsley (Wed 04-Oct-17 10:24:26)

Standard User ian72
(eat-sleep-adslguide) Wed 04-Oct-17 10:36:23
Print Post

Re: DSL Stats Query


[re: WilliamGrimsley] [link to this post]
 
I don't use it but have you checked it hasn't reported 1 twice? I'm assuming it reports the delta from the last report rather than the total number so whilst telnet will tell you there have been a total of 2 I would expect DSL Stats to report 2 lots of 1 unless they both happened in the same "reporting period".
Standard User WilliamGrimsley
(experienced) Wed 04-Oct-17 10:43:55
Print Post

Re: DSL Stats Query


[re: ian72] [link to this post]
 
Yeah, it's only reported it once. I'm not sure whether they occured in the same minute, though it's possible.

https://i.imgur.com/ZLhGSoG.png

William Grimsley.


Register (or login) on our website and you will not see this ad.

Standard User WWWombat
(knowledge is power) Mon 09-Oct-17 01:24:53
Print Post

Re: DSL Stats Query


[re: WilliamGrimsley] [link to this post]
 
Software like this has to count the changes inbetween the snapshots it sees on each telnet session.

The first gotcha is that, whatever the first sample says, it should not be logged. They should just be used as a baseline to start identifying future changes ... which should then be logged.

A subsequent problem when identifying changes occurs when the counters cycle back to zero again ... perhaps because of a resync, but perhaps because a 32-bit counter has overflowed. At this time it is hard to identify just how much has changed.

Some software will then not include counters after they've cycled, such as after a resync or a reboot. Just resets the baseline.
  Print Thread

Jump to