[rrd-users] basic rrdtool usage - short intervals

[Charles Sprickman]

On Sat, 18 Sep 2010, Steve Shipway wrote:

> What he said.
>
> You might also want to increase the length of the RRA proportionally, as 
> well, and maybe decrease the DS heartbeat (as you have a shorter step 
> size now).  Also, with a 1-dp RRA, you dont need to have the MIN and 
> MAX, as they will be the same as the AVG for a group of a single point.
>
> Try this:
>
> rrdtool create arcsize.rrd --start N --step 15 
> DS:arcsize:GAUGE:30:0:5000 RRA:AVERAGE:0.5:1:43920

Wow.  Thanks (and thanks Pablo).  That whole part of the "rrdtool create" 
doc was totally unclear to me.  This kind of makes sense now, as long as I 
don't try to figure out the "xfilefactor".

> If you want the 1-hour CDP RRAs as well (for graphing purposes) then 
> just add the original 3 RRA definitions to the end of the statment:

At first I was a little stumped because my rrd had a ton of empty values 
before the collection started.  But I suppose that's what the "RR" part of 
rrd is, right?  I have to always pre-define how much data I'm going to 
store.  I just set a start/end time on the graph command and that seems to 
work well.

Anyhow, thanks again - this is an awesome tool.  I'm looking at %busy and 
ops/sec for 4 disks in a zfs raidz-1 setup as well as system load and the 
size of the zfs arc cache while I do some bulk loads into postgres and 
fiddle with various tuning parameters.  I settled on a 5 second sample. 
Having a visual representation of what's going on is great.

Charles

> rrdtool create arcsize.rrd --start N --step 15 
> DS:arcsize:GAUGE:30:0:5000 RRA:AVERAGE:0.5:1:43920 RRA:MIN:0.5:12:3660 
> RRA:MAX:0.5:12:3660 RRA:AVERAGE:0.5:12:3660 Steve
>
>
> Steve Shipway
> University of Auckland ITS
> UNIX Systems Design Lead
> s.shipway at auckland.ac.nz<mailto:s.shipway at auckland.ac.nz>
> Ph: +64 9 373 7599 ext 86487
>
> ________________________________
> From: rrd-users-bounces+s.shipway=auckland.ac.nz at lists.oetiker.ch [rrd-users-bounces+s.shipway=auckland.ac.nz at lists.oetiker.ch] on behalf of Pablo Sanchez [pablo at blueoakdb.com]
>
> The problem is your RRA's are consolidating 12 points:  12 x 15
> (second steps) = 180 seconds.
>
> Change your RRA's so they consolidate one step and you'll get what
> you're after.
>