[smokeping-users] Graph scale problem

[G.W. Haywood]

Hi Tobi,

On Mon, 5 May 2008, Tobias Oetiker wrote:

> > Basically I just look for the range in the data and scale the graph
> > limits so they're a few tens of percent wider than that.  The scale
> > factors are easily adjusted.  I have no idea what I've broken. :)
>
> so are you looking at the smoke too, or just at the median ?

Just the median at the moment, it seems to be adequate.  The graph
limits in the example were set at maxmedian+50% and minmedian-20%.

> > > For the upcomming ajax version of smoekping I imagine to have graph
> > > properties configurable interactively so that you can make a graph
> > > look right by hand once you have identified an interesting section.
> >
> > Cool!  Would that propagate to the non-ajax graphs?
>
> the interface will be all-ayax. but it will be possible to get a
> url to a graph so that it can be used in other context with all the
> changes applied ...

In that case it seems likely my modifications may become redundant.
That's OK. :)  In the meantime I'll look further into the colouring
issues, since the graphs which cover longer periods seem to give me
far more pessimistic views than those which cover shorted periods.
You can see what I mean in these examples:

http://www.jubileegroup.co.uk/JOS/misc/link3-1.png
http://www.jubileegroup.co.uk/JOS/misc/link3-2.png

I estimate that between the two peaks (caused by the nightly backup)
in the 30 hour graph, less than ten percent of the time intervals show
loss of any packets.  In the 8 day graph it's nearer 40% for the same
time period.  That's unacceptable IMHO.

--

73,
Ged.