Re: [php-weathermap] [0.96test1]Aggregate functions won't support - for a DS

[Erik Versaevel <e.versaevel@xxxxxxxxxxx>]

Howard Jones wrote:
> E. Versaevel wrote:
>> Node target (cpu usage in this case) lacks support for - (to ignore 1
>> value)
>>   
> D'oh. Didn't think about that. The aggregate stuff was hacked in pretty
> quickly, as you have found.
> 
> Thanks for the report (and the others, too!), Erik... I'll try and get
> these together into a test2 release sometime soon. I've added a few new
> features that I wanted while editing maps in the last few days, too.
> 
> I'm not sure what the best way to proceed is with the aggregation
> functions - I hadn't considered either --step or -w when I started. I
> suppose there should be an rrd_options variable too, since the
> alternative is to use rrdinfo to try and figure it out for each .rrd,
> which seems pretty ugly. It would also mean that I'd need to actually
> parse the date, to figure out what -w needs to be :-) Maybe that's not a
> bad thing... it would allow for special dates like "1st of this month
> midnight", which can't be done with rrdtool directly, as far as I know.
> 
> Howie

Hi Howard,

First of all, thanks for the tool in the first place :)
I'm a core network engineer at a dutch ISP and i'm using weathermap to map our access and core networks.

I think using a rrdtool_options variable would give you the most flexibility, also the rrdtool_period variable should support at style spec date/times
like rrdtool_period=start+1month rrd_start=now-1month

On the otherhand, for monthly reports (like maximum link/cpu usage for the previous month) generating a graph with end-date 1st of this month at
midnight would be useful :)

Is there a way to relay CLI parameters to the overlib mouse over graph's as i'm using those to draw the relevant data (link usage/cpu/mem/temp etc)
however, if i create a monthly report i stil get the data from the last day in my overlib's :)

Another problem i ran into with the overlibs is that the graphs (3 targets on 1 node) gets drawn off the bottom of the screen which results in a
scrollbar, but if you scroll down the overlib graphs vanishes and so does the scrollbar.


Erik Versaevel