[rrd-users] Re: Data granularity

[Tolga Yalcinkaya]

Serge,

Thanks for your response and detailed explanation.  I really appreciate 
that.

I have been experimenting with different settings.  Below is an example:

	rrdtool create resp.rrd --step 5 \
	DS:resp:GAUGE:10:0:U \
	RRA:AVERAGE:0.999:1:1000

If I do single data point update using:

	rrdtool update resp.rrd 1106249083:45

and *not* post any update for a while ( > 10 seconds ), the RRD fetch 
command shows that the AVERAGE CF has the following value:
	1106249083: nan

Why do I have a NaN?  Shouldn't I have "4.5000000000e+01" instead?  Is 
there a way to retain the value entered for that step?

Thanks,

Tolga.


Serge Maandag wrote:
>>I have a monitor scripts that gets invoked every time a user 
>>interacts with a Web service.  There are several Web services 
>>that we are collecting data from.  Each one goes to a 
>>different RRD.  Some services are very busy (several hits per 
>>second) and some are not (a few hits per 
>>week).   There is no way for me to identify which is which...
> 
> 
> I cannot see how this works. Rrdtool collects rates, while you use
> It as a counting mechanism. But as long as it works for you..
> 
> 
>>I would like to use 5sec STEP so that there is enough detail 
>>for busy services.  A few questions to RRD gurus:
>>
>>(1) What is best value for HEARTBEAT?  Is 10 seconds too low? 
> 
> 
> That depends on what you want to see in your database.
> If you set it to 10:
> - updates that are longer apart than 10 seconds, will cause NaNs
>   to appear in your database
> - updates that are between 5 and 10 seconds apart, will cause
>   2 consecutive rows to be influenced, it will be averaged out.
> 
> If you set heartnbeat to N x step, N rows will be affected by
> An update, provided your high res RRA has a row time that is equal
> To the step size.
> 
> 
>>(2) What is the best value for XFF?
> 
> 
> That also totally depends. 
> 
> Say you have these RRA's:
> 
> RRA:AVERAGE:0.5:1:1000
> RRA:AVERAGE:0.5:10:300
> 
> Every step tim, one value will be put in the first RRA.
> Every 10 step times, 10 values from the first RRA will be averaged 
> And the result will be put in the second RRA. Unless more than 50%
> Of the 10 values are NaNs. If that is the case NaN will be put in the
> Second RRA.
> 
> If the second RRA would read:
> 
> RRA:AVERAGE:0.9:10:300
> 
> Then the average of the 10 values would still be put in the second
> RRA if 9 out of 10 values would be NaN.
> 
> So it all depends on what you want to see.
> 
> In your case I'd probably just dump the hits with a timestamp in a 
> log file and setup a cron job that would do every 5 minutes:
> - read back all lines and count the hits that were within the last 
>   5 minutes
> - update the RRD database with the sum of the counted hits as a gauge 
>   value.
> 
> Serge.
> 
> -------------
> Op de inhoud van dit e-mailbericht en de daaraan gehechte bijlagen is de inhoud van de volgende disclaimer van toepassing: http://www.zeelandnet.nl/disclaimer.php
> 
> --
> Unsubscribe mailto:rrd-users-request at list.ee.ethz.ch?subject=unsubscribe
> Help        mailto:rrd-users-request at list.ee.ethz.ch?subject=help
> Archive     http://www.ee.ethz.ch/~slist/rrd-users
> WebAdmin    http://www.ee.ethz.ch/~slist/lsg2.cgi
> 
> 
> 

-- 
Tolga Yalcinkaya

_________________________________________________________________
WSABI:  A fiery condiment for Apache Axis!   http://www.wsabi.org

--
Unsubscribe mailto:rrd-users-request at list.ee.ethz.ch?subject=unsubscribe
Help        mailto:rrd-users-request at list.ee.ethz.ch?subject=help
Archive     http://www.ee.ethz.ch/~slist/rrd-users
WebAdmin    http://www.ee.ethz.ch/~slist/lsg2.cgi