[rrd-users] RRD Graphs with bad averaging of missing data?

[Matt Elias]

I appreciate the patience, I'm no RRD pro and I appreciate the help.

Here the bit that creates the RRD. (Perl RRDs module)
RRDs::create("$rrd/bin$t-$n.temps.rrd",
"-s", "1",
"DS:Temp:GAUGE:10:U:U",
"RRA:AVERAGE:0.5:1:900",        # track 900 5sec samples for 1hr graph
"RRA:AVERAGE:0.5:2:900",        # 1800 5sec samples, 2hr graph
"RRA:AVERAGE:0.5:12:900",        # 10800 5sec samples, 12hr graph
"RRA:AVERAGE:0.5:24:900",        # 21600 5sec samples, 24hr graph
"RRA:AVERAGE:0.5:168:900",        # 1 week graph
"RRA:AVERAGE:0.5:336:900",        # 2 week graph
"RRA:AVERAGE:0.5:720:900",        # 1month graph
"RRA:AVERAGE:0.5:8760:900",    # 1yr graph
"RRA:AVERAGE:0.5:17520:900");    # 2yr graph

Looks like I'm using an XFF of 0.5? This I carried over from another 
project and it works well there but maybe that value should be 0 here? I 
can't remember why I used 0.5.

Here's my rrdtool info dump.
filename = "/var/www/Heatpump/bin120-1.temps.rrd"
rrd_version = "0003"
step = 1
last_update = 1397003090
header_size = 2248
ds[Temp].index = 0
ds[Temp].type = "GAUGE"
ds[Temp].minimal_heartbeat = 10
ds[Temp].min = NaN
ds[Temp].max = NaN
ds[Temp].last_ds = "358"
ds[Temp].value = 1.0645595400e+02
ds[Temp].unknown_sec = 0
rra[0].cf = "AVERAGE"
rra[0].rows = 900
rra[0].cur_row = 605
rra[0].pdp_per_row = 1
rra[0].xff = 5.0000000000e-01
rra[0].cdp_prep[0].value = NaN
rra[0].cdp_prep[0].unknown_datapoints = 0
rra[1].cf = "AVERAGE"
rra[1].rows = 900
rra[1].cur_row = 887
rra[1].pdp_per_row = 2
rra[1].xff = 5.0000000000e-01
rra[1].cdp_prep[0].value = 0.0000000000e+00
rra[1].cdp_prep[0].unknown_datapoints = 0
rra[2].cf = "AVERAGE"
rra[2].rows = 900
rra[2].cur_row = 55
rra[2].pdp_per_row = 12
rra[2].xff = 5.0000000000e-01
rra[2].cdp_prep[0].value = 7.1600000000e+02
rra[2].cdp_prep[0].unknown_datapoints = 0
rra[3].cf = "AVERAGE"
rra[3].rows = 900
rra[3].cur_row = 767
rra[3].pdp_per_row = 24
rra[3].xff = 5.0000000000e-01
rra[3].cdp_prep[0].value = 7.1600000000e+02
rra[3].cdp_prep[0].unknown_datapoints = 0
rra[4].cf = "AVERAGE"
rra[4].rows = 900
rra[4].cur_row = 880
rra[4].pdp_per_row = 168
rra[4].xff = 5.0000000000e-01
rra[4].cdp_prep[0].value = 2.3556098206e+04
rra[4].cdp_prep[0].unknown_datapoints = 32
rra[5].cf = "AVERAGE"
rra[5].rows = 900
rra[5].cur_row = 679
rra[5].pdp_per_row = 336
rra[5].xff = 5.0000000000e-01
rra[5].cdp_prep[0].value = 2.3556098206e+04
rra[5].cdp_prep[0].unknown_datapoints = 32
rra[6].cf = "AVERAGE"
rra[6].rows = 900
rra[6].cur_row = 876
rra[6].pdp_per_row = 720
rra[6].xff = 5.0000000000e-01
rra[6].cdp_prep[0].value = 1.6038098206e+04
rra[6].cdp_prep[0].unknown_datapoints = 5
rra[7].cf = "AVERAGE"
rra[7].rows = 900
rra[7].cur_row = 372
rra[7].pdp_per_row = 8760
rra[7].xff = 5.0000000000e-01
rra[7].cdp_prep[0].value = 8.7485996954e+05
rra[7].cdp_prep[0].unknown_datapoints = 32
rra[8].cf = "AVERAGE"
rra[8].rows = 900
rra[8].cur_row = 857
rra[8].pdp_per_row = 17520
rra[8].xff = 5.0000000000e-01
rra[8].cdp_prep[0].value = 6.3089923566e+06
rra[8].cdp_prep[0].unknown_datapoints = 497

Thanks again.

On 4/8/2014 6:17 PM, Steve Shipway wrote:
> With questions of this nature, it would be really helpful to be supplied
> with the structure of the RRD being used.  The output of 'rrdtool info'
> would be ideal.
>
> Also a sample of the incoming data and a dump of the relevant RRA around the
> glitch would help.
>
> Having said that, I concur with Alex that the cause is likely related to an
> overly large XFF setting, though it may possibly be down to your data
> retrieval script incorrectly storing  a zero when it should have stored an
> Unknown.
>
> Steve
>
> Steve Shipway
> s.shipway at auckland.ac.nz
>