Re: Clearing extreme points in graphs?
Ed Ravin <eravin <at> panix.com>
2003-01-28 16:49:29 GMT
On Tue, Jan 28, 2003 at 11:11:40AM -0500, Jeff Stewart wrote:
> Hopefully this is doable. I have about 6 months worth of data for my
> graphs. Typical RTT is alway 10 - 20 ms. During this past weekends
> sql/slapper fiasco, I've got this huge spike in the graphs, up around
> 2000+ms, that lasts for a couple of hours. As expected. [... Is]
> there any way to clear out these extreme entries? Short of wiping out
> the .rrd files and starting fresh?
The Cricket folks have excellent spike-removal technology (now, if we
could only prevent the damn spikes from happening in the first place).
Please see the Cricket contrib page at:
http://www.certaintysolutions.com/tech-advice/cricket-contrib/
You'll be interested in these two programs:
killspike2 (by Matt Zimmerman) - removes entries in a DS that
exceed a specified value.
rrdcut (by yours truly) - removes all entries in the RRD file
between a specified date and time.
I've used both of these scripts for Cricket RRD files, I'm not
familiar with the construction of a Smokeping RRD, but these
scripts have no Cricket-specific stuff in them so they might
work for you. Suggest you set up a test copy of an RRD and
experiment with it.
-- Ed
(Continue reading)