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 ... 8 9 [10]

Author Topic: DSLstats v5.4 released  (Read 42479 times)

roseway

  • Administrator
  • Senior Kitizen
  • *
  • Posts: 43568
  • Penguins CAN fly
    • DSLstats
Re: DSLstats v5.4 released
« Reply #135 on: May 17, 2015, 09:57:01 AM »

Quote
I've just noticed that CRC error graph doesn't appear to match the telnet data - it seems to give a graph showing higher values......I wonder if there's some scaling issue?

Are you talking downstream only here? I'll try to check this out as soon as I can.
Logged
  Eric

jack21

  • Reg Member
  • ***
  • Posts: 144
Re: DSLstats v5.4 released
« Reply #136 on: May 17, 2015, 10:17:47 AM »

Yes Eric......just talking about Downstream on per-minute - the per-sample seems OK so far.
Logged

krypton

  • Reg Member
  • ***
  • Posts: 128
Re: DSLstats v5.4 released
« Reply #137 on: May 17, 2015, 12:09:21 PM »

I think this is normal if the sampling period is smaller then one minute. I guess yours is around 20 seconds. DSLstats seems to extrapolate the received values to one minute if the per minute option is selected.
Logged

jack21

  • Reg Member
  • ***
  • Posts: 144
Re: DSLstats v5.4 released
« Reply #138 on: May 17, 2015, 12:24:37 PM »

Sampling interval was/is 15 secs, so your estimate wasn't far out!
Sounds a plausible explanation; if that's so, then I can stick with the per-sample display.
DSLstats is a lovely tool though, and it has allowed me to ditch the sole, old Windows machine that I used purely to run Routerstats 24*7.....the Pi is much handier!
Logged

roseway

  • Administrator
  • Senior Kitizen
  • *
  • Posts: 43568
  • Penguins CAN fly
    • DSLstats
Re: DSLstats v5.4 released
« Reply #139 on: May 17, 2015, 12:45:16 PM »

I think this is normal if the sampling period is smaller then one minute. I guess yours is around 20 seconds. DSLstats seems to extrapolate the received values to one minute if the per minute option is selected.

Yes, you're right. The CRC and FEC graphs calculate a per-minute rate based on the change in the totals between successive samples, and the time between samples.
Logged
  Eric

JustAnother

  • Member
  • **
  • Posts: 37
Re: DSLstats v5.4 released
« Reply #140 on: July 27, 2015, 02:14:17 PM »

For the last 10 days, I've had adhoc 'duplicate upload file error's again, 24 of them.
Logged

tbailey2

  • Kitizen
  • ****
  • Posts: 1245
Re: DSLstats v5.4 released
« Reply #141 on: July 27, 2015, 03:40:54 PM »

For the last 10 days, I've had adhoc 'duplicate upload file error's again, 24 of them.

They only occur after three consecutive attempts at inserting a duplicate record into the database, that is one where the time in the modem file you upload, which is always a whole date/time minute with no seconds, already exists in the database. This can be caused by all sorts of things, usually running two instances of an upload program at once or having a clock that's suddenly lost or gained time. There is no easy way round this but 99.9% of the time there is no problem.

In your case, it's a bug because you are uploading almost exactly on the minute but varying by a few secs due to the time you take to process the data and rounding to the nearest sec at the server is causing the duplicates. If you stop DSLstats for 10 secs and restart it should stop this.

In theory you can have 1440 a day. I banned someone 4 months ago who had managed 5000 bum uploads consecutively - and they were still sending them but e-mail is invalid. Still have to be processed though, so blocked the IP.
Logged
Tony
My Books!
Plusnet 80/20 - DSLstats - HG612/TG582n - ECI

JustAnother

  • Member
  • **
  • Posts: 37
Re: DSLstats v5.4 released
« Reply #142 on: July 27, 2015, 04:05:11 PM »

I have quit and started dslstats up again - for reference I have uptimes of 0.5-1 month, incase something is getting out of sync.

Time keeping has been accurate on this Debian Testing machine for about a month now (prior to that, it was using a NIST server that never bothered to return the time...).

I can increase the sample period if needed (atm on the Configuration -> Graphs tab its set to 60 seconds).
Logged

roseway

  • Administrator
  • Senior Kitizen
  • *
  • Posts: 43568
  • Penguins CAN fly
    • DSLstats
Re: DSLstats v5.4 released
« Reply #143 on: July 27, 2015, 04:28:05 PM »

I should probably refine the DSLstats timing arrangement. It's designed to avoid uploading to MDWS during the critical period of the last few seconds of the minute (offhand I don't recall how many seconds that is), but I think I should give the timing a bit more safety margin to allow for differing times taken for sampling. I'll do this in the next release or pre-release.

In the meantime, if you avoid pressing the green button during the last 15 seconds of the minute you should be safe. So long as your PC clock is accurate, the long term stability of DSLstats' timing is very accurate.
Logged
  Eric

JustAnother

  • Member
  • **
  • Posts: 37
Re: DSLstats v5.4 released
« Reply #144 on: July 27, 2015, 06:15:32 PM »

Sorry, where is the green button for reference? The program is normally left alone to run.
Logged

roseway

  • Administrator
  • Senior Kitizen
  • *
  • Posts: 43568
  • Penguins CAN fly
    • DSLstats
Re: DSLstats v5.4 released
« Reply #145 on: July 27, 2015, 06:54:55 PM »

Sorry, where is the green button for reference? The program is normally left alone to run.

I was referring to the "Start recording" button at the top left. If you have DSLstats configured to start recording automatically when it starts up, perhaps for the moment it would be better to disable this.
Logged
  Eric

JustAnother

  • Member
  • **
  • Posts: 37
Re: DSLstats v5.4 released
« Reply #146 on: July 27, 2015, 07:08:21 PM »

OK, I have disabled that option and stopped the current instance recording for now.
Logged
Pages: 1 ... 8 9 [10]
 

anything