my graph thinks snmp is returning "micros"

[mtw00]

Hello-

I created data and graph templates to poll a few TDMoIP devices I have. Everything is working great except for one thing. My graphs are displaying numbers in millions, yet the OID I am polling is only increaing by maybe 15 or 20 every 5 minutes. How can I tell cacti that the values being returned aren't thousands or milliions?

[mtw00]

okay, starting to understand this a little bit more. Here's what I'm seeing in the rrd.log and what rrdtool is returning:

The last 3 data samples in rrd.log: (edited for readability)

03/08/2004 11:12 AM - CMD: durham_tdm_ipmux1_seq_errors_239.rrd --template seq_errors:jit_buf_under:jit_buf_over N:1:6145:10641
03/08/2004 11:17 AM - CMD: durham_tdm_ipmux1_seq_errors_239.rrd --template seq_errors:jit_buf_under:jit_buf_over N:1:6150:10649
03/08/2004 11:22 AM - CMD:durham_tdm_ipmux1_seq_errors_239.rrd --template seq_errors:jit_buf_under:jit_buf_over N:1:6151:10652


as we can see, seq_errors are the same, jit_buf_under increases by 5 then by 1, finally jit_buff_over increases by 8 then by 3. When I look at the data in the RRD file, I get something unexpected:

# rrdtool fetch -s 1078773000 /var/www/cacti/rra/durham_tdm_ipmux1_seq_errors_239.rrd AVERAGE
seq_errors jit_buf_under jit_buf_over

1078773000: 0.0000000000e+00 1.2772292828e-01 1.6012961724e-01
1078773300: 0.0000000000e+00 8.6034030008e-02 1.0310818268e-01
1078773600: 0.0000000000e+00 8.9985972684e-03 1.7073311185e-02

negaitive exponential values?

[Lux]

I believe that you are getting the correct information out of Cacti. You are doing an average of the number of errors over a 5 minute period. The average is calculated per second. So, if you have 5 errors over a 5 minute period, you have 5 errors in 300 seconds or 0.01666.... errors per second. Looking at your output, this is almost exactly what you have. (1.7073311185e-02 is 0.017073311185) Remember, negative exponents means that you move the decimal x number of spaces to the left rather than to the right.

Also, remember when looking at your graphs that Cacti uses:

m = mili (1/1000)
= normal base 10 number (1/1)
k = kilo (1000/1)
M = Mega (1000000/1)

Mike

[mtw00]

okay, that makes complete sense.

Is there a way for cacti to present me with the total amount of errors in the past 5 minutes instead of the average per second?

[moya]

Have you tryed changing the "Consolidation Function" from "Average" to "Last" (Graph Template)?

-Moya

[mtw00]

Have you tryed changing the "Consolidation Function" from "Average" to "Last" (Graph Template)?

-Moya

yes I have... it's my understanding that "last" is just the latest 5 minute interval, which still returns a value in units per second.

[bulek]

Use CDEF function called "Make Per 5 minutes" (or similar) in your graph items settings.

- Piotr