Just wanted to say thanks and that it works fine here using ubuntu 6.06
so debian users shouldn't have issues either.  Only thing I had to
install was grace using `apt-get install grace`

Couple notes for us ubuntu / debian users.  The peerstats files are kept
in /var/logs/ntpstats.  On here it keeps the last 10 days with the last
8 days gzip'd.  So when you modify the scripts you will have to use
gunzip -c on those files.  Also the default $work_dir doesn't exist and
the $ntp_template path needs to be changed.

I was able to get it hacked together in probably 15 minutes, most of
that noticing the above issues.  And I do mean it's hacked together
right now (hence why I'm not posting an example :) ).

The easiest to figure out includes and excludes is to just run it
without any include or exclude filters, it will then process all the
servers in your peerstats file (the 5 or so peers your client is talking
to, not the 1000s of people that are talking to your server).  You can
then see what it uses for the hostnames and filter accordingly.

Going to take a look at this to see if there's a way to overlay the
server's offset (output of `ntpq -c rv`) with this information since it
should generally match what your server is doing.  For example I have
two good servers that are following the curves of my server on 0ms line,
then another server follows the curves but with a +1ms offset.  This is
more of a concern for us st2+ servers since we aren't going to be as
stable as cesium st1 server :)

Thanks, John, for mentioning this.  I've seen your stats for a while and
never really paid much attention to the scripts till you mentioned it.

John Ackermann N8UR wrote:
> For the last couple of years I've been using a perl program together
> with the "grace" graphing program to generate nice plots based on the
> peerstats logs from ntpd.  Unlike most ntp monitoring tools, which seem
> to be aimed at interpreting local data, these tools monitor the offset
> of remote peers compared to the local system.
> 
> You can see the results at http://www.febo.com/time-freq/ntp/stats
> 
> I've finally gotten around to fixing some long-standing problems and
> doing a bit of code clean up, and ntp-stats-0.99 is now available for
> download from http://www.febo.com/time-freq/tools/index.html.
> 
> While it seems to work, I haven't exhaustively tested against all the
> possible configuration options (and the code needs linting and cruft
> removal).  So if you try it out, I'd appreciate hearing how it works for
> you, and especially if you've found (and hopefully fixed!) any bugs.
> 
> John
> _______________________________________________
> timekeepers mailing list
> [email protected]
> https://fortytwo.ch/mailman/cgi-bin/listinfo/timekeepers
> 

Attachment: smime.p7s
Description: S/MIME Cryptographic Signature

_______________________________________________
timekeepers mailing list
[email protected]
https://fortytwo.ch/mailman/cgi-bin/listinfo/timekeepers

Reply via email to