[rrd-users] Slow collection runtimes occurring regularly

[Rick Jones]

On Mon, 2011-05-02 at 19:04 +0100, Simon Hobson wrote:
> Joshua Keroes wrote:
> 
> >Is our theory correct, that the RRD consolidations may be causing
> >these long runtimes
> 
> It would be logical that each time a "bucket" is completed, then 
> additional processing will happen in order to do the necessary 
> normalisation and consolidation.
> 
> >, and if that's the case, is there a way to evenly
> >stagger the consolidations over time so we can better distribute RRD
> >update load?
> 
> You cannot offset the consolidation times as everything is referenced 
> to unix epoch (Midnight 1st Jan 1970). However, if it is just system 
> load you are concerned about, can you delay feeding in some update 
> values a bit ? As long as you use an explicit timestamp in the update 
> statement, then it doesn't matter what actual clock time the update 
> is submitted to rrd - as long as you maintain the right sequence of 
> updates.

At the risk of typing into Joshua's keyboard I suspect he wasn't asking
to shift the times to which the RRAs are aligned, but spread the work
done for those RRAs - the hypothesis being that for an RRA which
aggregates N samples it not do all the work after N samples but try to
do 1/Nth the work upon the presentation of each sample.

If indeed it is waiting to have N samples and then does the work en mass
then all delaying updates would do is shift the spikes rather than
smooth them out right?

rick jones