thomas at braindumped.com
Wed Apr 26 16:57:47 MEST 2006
i read alex' really usefull docs from www.vandenbogaerdt.nl/rrdtool but
still don't get the point about normalisation.
i thought i could prevent normalisation by updating at times that are
multiples of stepsize.
so if i `rrdtool create test.rrd --start 1146061700 --step 300
and start updating at 1146061700 and then update at 1146062000 and at
1146062300 and so on, i can dump the exact values i provided to the
rrdupdate command, since they arent normalized.
i used the exact time during updates (not N:), to be shure to fill the
i thought i got it, but the next example made me nervous.
i have an rrd with the following parameters:
step = 300
ds[users].type = "GAUGE"
ds[users].minimal_heartbeat = 600
ds[users].min = 0.0000000000e+00
ds[users].max = 5.0000000000e+02
ds[users].last_ds = "UNKN"
ds[users].value = 4.8000000000e+02
ds[users].unknown_sec = 0
rra.cf = "AVERAGE"
rra.rows = 600
rra.pdp_per_row = 1
rra.xff = 5.0000000000e-01
rra.cdp_prep.value = NaN
rra.cdp_prep.unknown_datapoints = 0
i provided only integers to rrdupdate (4, 5 and 6 [logged in users]), but
i see a 4.8 value.
i have pretty much data in this rra that became normalized since the
update-times didn't match the rra-slots.
i changed the update-process, so updates where provided with exact times,
but i still get values that are normalized (the 4.8)
is that because of the pervious 'normalized data' that is allready in
the rra, or am i wrong in thinking that i could get the exact values out
of rrdtool, that i provided during update?
any enlightenment appreciated
Unsubscribe mailto:rrd-users-request at list.ee.ethz.ch?subject=unsubscribe
Help mailto:rrd-users-request at list.ee.ethz.ch?subject=help
More information about the rrd-users