round robin databases work well for this sort of data. They
progressively drop data off the end, yet still manage to keep enough
stats to continue to do nice graphs :-)

an example is mrtg, which has been taken over by a more generic but
similarly setup package called RRDTool
http://people.ee.ethz.ch/~oetiker/webtools/rrdtool/


On Fri, 21 Jan 2005 13:13:28 +1300 (NZDT)
Steve Holdoway <[EMAIL PROTECTED]> wrote:

> I was going to suggest the same as Nick, but this post put paid to that.
> 
> However, starting up using either init.d scripts *or* crontab scripts is
> good. I know it sounds a bit strange, but if you put it in cron, then it
> can act as a means of automagically restarting the process if it fails. I
> wish I could take credit for that one, but I stole it off the seti project
> (:
> 
> But, you're missing something in all of this... where's the database to
> store it all in? That way, you can generate up-to-date graphs on the fly
> when requested!
> 
> I do similar with the METAR weather data that I get from noaa ( I think!
> ). I've just got to write the data display part (^:
> 
> Cheers,
> 
> Steve
> 
> On Fri, January 21, 2005 11:56 am, Andrew Errington said:
> >> I am thinking that these programs probably do nt need to run
> >> continuously anyway.
> >>
> >> i assume they are doing something like:
> >>
> >> begin
> >> poll sensor
> >> write data
> >> sleep x minutes
> >> again
> >>
> >> would it be better to rewrite the program to just do:
> >>
> >> begin
> >> poll sensor
> >> write data
> >> end
> >>
> >> then let cron do the scheduling every x minutes. that way as long as
> >> cron is running, your data collection program should run. if the data
> >> collection program outputs a value indicating whether or not it has
> >> finished successfuly, you can get cron to send you an email on failure.
> >>
> >> just another option...
> >
> > Nice idea.  The temperature sensor could work like that, but it has its
> > own
> > scheduling built-in (just provide the interval at the command line), and I
> > have set it up for every 5 minutes.  The wind sensor is outputting data
> > every three seconds.  My Perl script accumulates this data over 10 minutes
> > and outputs the average and max values, so it really needs to be running
> > all the time.
> >
> > Thanks,
> >
> > Andy
> >
> 
> 
> -- 
> Artificial Intelligence is no match for natural stupidity.

-- 
Nick Rout <[EMAIL PROTECTED]>

Reply via email to