[rrd-users] Fetching raw data
Luke Burden
lukeburden at gmail.com
Thu Mar 20 23:41:14 CET 2008
Hi all,
What I'm trying to achieve with RRDTool is quite simple, and I believe
possible with RRD - I just need you guys to run your eyes over my case and
point out where my misunderstanding of your code is.
I have a time series, with samples each 1 minute. I want to store this in
an RRD, and retrieve consolidated values to facilitate a zoomable graph
front-end. This works a treat, but what I want to do as well as fetching
consolidated data from the RRD is to be able to pull that original data out
at full resolution too.
I figured that an RRA:AVERAGE with 1 step and a heartbeat of 60 secs would
do the trick as long as there were only ever 1 sample updated per 60 second
period. But I can't get this to work! I thought it might be the AVERAGE
function doing sneaky things, but LAST and MAX give exactly the same fetch
results.
*I'm creating the graph using:*
rrdtool create subset.rrd -b 1199103570 -s 60 DS:users:GAUGE:60:0:1000000
RRA:MAX:0.5:1:525600 RRA:LAST:0.9:1:525600 RRA:AVERAGE:0.9:1:525600
*Then updating with the following data:*
1199103601 50283
1199103661 50282
1199103721 50320
1199103781 50365
1199103841 50359
1199103906 50307
1199103961 50332
1199104021 50298
1199104081 50329
1199104141 50307
1199104202 50267
*Then the following fetches:*
rrdtool fetch subset.rrd AVERAGE -r60 -s1199103590 -e1199104202
rrdtool fetch subset.rrd LAST -r60 -s1199103590 -e1199104202
rrdtool fetch subset.rrd MAX -r60 -s1199103590 -e1199104202
*Produce:*
1199103600: 5.0283000000e+04
1199103660: 5.0282016667e+04
1199103720: 5.0319366667e+04
1199103780: 5.0364250000e+04
1199103840: 5.0359100000e+04
1199103900: nan
1199103960: 5.0332000000e+04
1199104020: 5.0298566667e+04
1199104080: 5.0328483333e+04
1199104140: 5.0307366667e+04
1199104200: nan
1199104260: nan
Clearly there is something going on here which I don't understand. Anyone
care to explain how I can get at the raw numbers?
On another issue, are there any plans for RRDTool supporting calendar
monthly consolidations, or is this best done using daily consolidations and
some scripts?
Many thanks,
Luke
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.oetiker.ch/pipermail/rrd-users/attachments/20080320/a9b2e938/attachment.html
More information about the rrd-users
mailing list