[rrd-users] like mrtg ?

[raptor]

hi,

could u give me the parameters that has to be passed to :

"rrdtool create" and "rrdtool update" so that i create the same db as for mrtg i.e :

rrdtool create test.rrd DS:COUNTER:test:300:1:100000 RRA:AVERAGE:0.5:300:?:288

288 <=  (24 hours * 60 min) / 5 min sample interval  <- is thi correct

i have to specify more RRA's for the other three images daily, weekly, monthly and yearly !!?
Then how I do "update" ? and If update is called every 5min I don't need to specify the time, isn't it ?! if yes, what is the easiest way ?

A second question, if I doesn't specify the MAX value in DS record, how can I later in image generation process cut very high values... what I mean ?  normaly I expect the values to go from 1 to 100 (i mean normally not nececary) but in fact the MAX value can be unknown or very high say 100000.. U know then that when the graph is generated it is almoust useless 'cause the interesting info is between 1 - 100, but u have to be able to view or graph sometimes the spikes too.. so u need them too..
The other complication comes from this that this 1-100 average after say 2 weeks may become 1-200.. so it is moving..
The idea is to detect automaticaly in some way this info, say walk all data and if 80% of the data is below value X (not known in advance but deducted from the data itself) then the range should be from 1 to X.(i.e. the 20% spikes can be cut for viewing purposes) did u got the idea.

One even more complex problem is if I ask it that way : if the difference between the low and high value is not more than Y, then draw normally otherwise go to cut-spikes mode i described before :")

In mrtg we normaly specify MaxBytes (i.e. MAX) in this line of thoughts how can i get this MAX value from the data itself so i still have the info that shows below the graph i.e. 60% of the max etc..

My main idea is that the graph should be self-tunable :")

sorry for the long post...

thanx in advance
raptor
PS. and yet another one not so related to rrd, but !! how do u draw "tendency!"-curve i.e. something like averaging over long periods of values but not exactly the same.... not sure if this is called "mean" value !!
I can explain it like a scenario what i want to detect, but probably this is not for this list..:"( but for "statistics one" :")

--
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