Interface statistics off a lot (Ubuntu 10.04 AMD64)

[TvL2386]

Hi!

I think it's a Ubuntu related problem. I have an ubuntu 10.04 AMD64 server and when I copy something large to another machine with roughly 100MB/s (1GBIT link), the graph created by cacti seems of by a factor 10. I have been copying for a while now...

I've running a test for a longer period of time to see it clearly

The command:


The test ran for 3233 seconds with an average speed of 33.2MB/s, which would result in 265.6MBIT/s.
It doesn't show in the graph:



EDIT: I forgot to mention that the server with the graphed interface is Ubuntu 10.04 AMD64. The Cacti host is a CentOS5.2 x86_64 machine.

Anybody know what could be the problem?

[TvL2386]

Done another test:

user@ubuntu_host:~$ dd if=/dev/zero count=1M bs=10240 | ssh user@otherhost cat > /dev/null
user@otherhost's password:
1048576+0 records in
1048576+0 records out
10737418240 bytes (11 GB) copied, 293.034 s, 36.6 MB/s

I did an snmpwalk of the IF_MIB during this time extracting the 'ifHCOutOctets.6 = Counter64'.

Results:

Code: Select all

td: 31 speed: 998 bytes/sec
td: 30 speed: 13187383 bytes/sec
td: 30 speed: 41962967 bytes/sec
td: 30 speed: 38440179 bytes/sec
td: 30 speed: 39788642 bytes/sec
td: 30 speed: 40642786 bytes/sec
td: 31 speed: 36056132 bytes/sec
td: 30 speed: 38879780 bytes/sec
td: 30 speed: 41439323 bytes/sec
td: 30 speed: 39078419 bytes/sec
td: 30 speed: 39354243 bytes/sec
td: 30 speed: 11636450 bytes/sec
td: 31 speed: 497 bytes/sec
td: 30 speed: 576 bytes/sec
td: 30 speed: 514 bytes/sec
td: 30 speed: 513 bytes/sec
td: 30 speed: 511 bytes/sec
td: 30 speed: 513 bytes/sec
td: 30 speed: 511 bytes/sec
td: 31 speed: 497 bytes/sec

NOTE: td is time difference between timestamps. I read that an octet is the same as a byte, so I printed bytes/sec instead of octets/sec

What cacti shows me:

Last peak is 70MBIT/s according to the graph. Which is wrong, it should be 288MBIT/s (36MB/s * 8)

Still at loss where this comes from, but I think I can say it's Cacti related

[gandalf]

Please post "rrdtool info" of the rrd file you're using.
Are you using 1 min polling or 5 min polling?
R.

[TvL2386]

According to the graph source the following rrd is used: /opt/cacti-0.8.7e/rra/213/9117.rrd. I'm using 5 minute polling. The total polling time is 17 seconds.

Code: Select all

rrdtool info /opt/cacti-0.8.7e/rra/213/9117.rrd
filename = "/opt/cacti-0.8.7e/rra/213/9117.rrd"
rrd_version = "0003"
step = 300
last_update = 1276925706
ds[traffic_in].type = "COUNTER"
ds[traffic_in].minimal_heartbeat = 600
ds[traffic_in].min = 0.0000000000e+00
ds[traffic_in].max = 1.0000000000e+07
ds[traffic_in].last_ds = "232675521"
ds[traffic_in].value = 2.2478000000e+02
ds[traffic_in].unknown_sec = 0
ds[traffic_out].type = "COUNTER"
ds[traffic_out].minimal_heartbeat = 600
ds[traffic_out].min = 0.0000000000e+00
ds[traffic_out].max = 1.0000000000e+07
ds[traffic_out].last_ds = "3946522749"
ds[traffic_out].value = 4.6400000000e+01
ds[traffic_out].unknown_sec = 0
rra[0].cf = "AVERAGE"
rra[0].rows = 500
rra[0].cur_row = 106
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[0].cdp_prep[1].value = NaN
rra[0].cdp_prep[1].unknown_datapoints = 0
rra[1].cf = "AVERAGE"
rra[1].rows = 600
rra[1].cur_row = 412
rra[1].pdp_per_row = 1
rra[1].xff = 5.0000000000e-01
rra[1].cdp_prep[0].value = NaN
rra[1].cdp_prep[0].unknown_datapoints = 0
rra[1].cdp_prep[1].value = NaN
rra[1].cdp_prep[1].unknown_datapoints = 0
rra[2].cf = "AVERAGE"
rra[2].rows = 700
rra[2].cur_row = 406
rra[2].pdp_per_row = 6
rra[2].xff = 5.0000000000e-01
rra[2].cdp_prep[0].value = 3.7498503753e+01
rra[2].cdp_prep[0].unknown_datapoints = 0
rra[2].cdp_prep[1].value = 7.7323090508e+00
rra[2].cdp_prep[1].unknown_datapoints = 0
rra[3].cf = "AVERAGE"
rra[3].rows = 775
rra[3].cur_row = 87
rra[3].pdp_per_row = 24
rra[3].xff = 5.0000000000e-01
rra[3].cdp_prep[0].value = 6.9172113065e+02
rra[3].cdp_prep[0].unknown_datapoints = 2
rra[3].cdp_prep[1].value = 1.3247018764e+02
rra[3].cdp_prep[1].unknown_datapoints = 2
rra[4].cf = "AVERAGE"
rra[4].rows = 797
rra[4].cur_row = 237
rra[4].pdp_per_row = 288
rra[4].xff = 5.0000000000e-01
rra[4].cdp_prep[0].value = 2.4738414400e+03
rra[4].cdp_prep[0].unknown_datapoints = 6
rra[4].cdp_prep[1].value = 4.7658189599e+02
rra[4].cdp_prep[1].unknown_datapoints = 6
rra[5].cf = "MAX"
rra[5].rows = 500
rra[5].cur_row = 398
rra[5].pdp_per_row = 1
rra[5].xff = 5.0000000000e-01
rra[5].cdp_prep[0].value = NaN
rra[5].cdp_prep[0].unknown_datapoints = 0
rra[5].cdp_prep[1].value = NaN
rra[5].cdp_prep[1].unknown_datapoints = 0
rra[6].cf = "MAX"
rra[6].rows = 600
rra[6].cur_row = 233
rra[6].pdp_per_row = 1
rra[6].xff = 5.0000000000e-01
rra[6].cdp_prep[0].value = NaN
rra[6].cdp_prep[0].unknown_datapoints = 0
rra[6].cdp_prep[1].value = NaN
rra[6].cdp_prep[1].unknown_datapoints = 0
rra[7].cf = "MAX"
rra[7].rows = 700
rra[7].cur_row = 551
rra[7].pdp_per_row = 6
rra[7].xff = 5.0000000000e-01
rra[7].cdp_prep[0].value = 3.9168918619e+01
rra[7].cdp_prep[0].unknown_datapoints = 0
rra[7].cdp_prep[1].value = 7.7323090508e+00
rra[7].cdp_prep[1].unknown_datapoints = 0
rra[8].cf = "MAX"
rra[8].rows = 775
rra[8].cur_row = 463
rra[8].pdp_per_row = 24
rra[8].xff = 5.0000000000e-01
rra[8].cdp_prep[0].value = 1.4235130494e+02
rra[8].cdp_prep[0].unknown_datapoints = 2
rra[8].cdp_prep[1].value = 8.5227937127e+00
rra[8].cdp_prep[1].unknown_datapoints = 2
rra[9].cf = "MAX"
rra[9].rows = 797
rra[9].cur_row = 676
rra[9].pdp_per_row = 288
rra[9].xff = 5.0000000000e-01
rra[9].cdp_prep[0].value = 1.4235130494e+02
rra[9].cdp_prep[0].unknown_datapoints = 6
rra[9].cdp_prep[1].value = 8.8499385178e+00
rra[9].cdp_prep[1].unknown_datapoints = 6

I don't know if it's relevant, but I have a few linux nodes and lots of Cisco devices. The new ubuntu 10.04 machine is the only one where graphs are off.

[gandalf]

According to the graph source the following rrd is used: /opt/cacti-0.8.7e/rra/213/9117.rrd. I'm using 5 minute polling. The total polling time is 17 seconds.

That's fine.

I don't know if it's relevant, but I have a few linux nodes and lots of Cisco devices. The new ubuntu 10.04 machine is the only one where graphs are off.

Not that I'm aware of.
R.

[gandalf]

10737418240 bytes (11 GB) copied, 293.034 s, 36.6 MB/s

YOu copy process lastet around 300 sec. Before and afterwards, there was no traffic. So it may have occured, that one half of that transfer hit a first polling interval and the rest may have hit a second polling interval.
Please retry and make sure, that the copy process covers a whole polling interval. This can be done by e.g. covering 10 minutes instead of 5.
R.

[TvL2386]

Hi Gandalf,

Indeed, true for my last test. Though I started at approximately polling time.
Please see my first post for a 3000+ seconds test.