[rrd-users] Updating more than once per second using rrdcached

[Simon Hobson]

Graham Leggett wrote:

>I am currently monitoring streamed events that are written to RRD 
>files, and sometimes an update for an RRD will arrive more than once 
>per second. When attempting to use rrdcached to speed this up, I have 
>started getting the following messages appearing from time to time, 
>which I understand as data being dropped by rrdcached:
>
>illegal attempt to update using time 1294658384 when last update time 
>is 1294658384 (minimum one second step)
>
>This causes us dropped data.
>
>Would it be possible in this case, instead of erroring out, to rather 
>flush previous changes for this particular RRD file from rrdcached to
>disk, before accepting the new update?

It wouldn't help since the underlying data only supports whole second 
resolution. You'll have to do some scripting of your own to aggregate 
updates a bit before you feed them to RRD.

Alternatively, you'd have to hack the rrd code so that if you feed in 
more than one update with the same timestamp it will aggregate the 
values (ABSOLUTE data type only).
I suspect that such hacks will be far from trivial since you would 
effectively have to "undo" the previous update (hard since the older 
data won't be there) and re-do it with the new values.
-- 
Simon Hobson

Visit http://www.magpiesnestpublishing.co.uk/ for books by acclaimed
author Gladys Hobson. Novels - poetry - short stories - ideal as
Christmas stocking fillers. Some available as e-books.