How are the 95th values generated for Cacti? Are they wrong?
James Bensley <[email protected]>
| Newsgroups | gmane.network.cacti.user |
|---|---|
| Message-ID | <CAAWx_pWGHirqZX_eAQ5hr88vdaqDjS2CMWCN0930UrPzkNGx_A@mail.gmail.com> |
Hi all, I am trying to run rrdtool manually to generate a 95th percentile value from an rrd file via the CLI, but I can't get the same results. See this first screen-short: http://i49.tinypic.com/6gip8m.png Here I am running a basic script called "rrd95th" and passing the name of an rrd file "xxxxx_traffic_in_6189.rrd" to it (the same one Cacti uses). I have also used `cat` to show the contents of the script (it just calls rrdtool really). It gives the 95th percentile values as 3.92Mbps in and 5.33Mbps out. In the screenshot below I am using the Cacti web interface to view a graph based on the same rrd file, for the same date and time period, but it has a different 95th percentile value. It says 95th percentile = 9.6Mbps (this isn't in or out, it doesn't say, but either way, it's higher than both my values put together [which isn't even what 95th is!]). http://i49.tinypic.com/2qkqxhx.png In that debug Cacti isn't showing me how the 95th value is calculated, how is it calculated here? Which is wrong, my script or Cacti? The graph in Cacti (if it makes and difference) uses the graph template "Interface - 64bit Traffic (bits/sec, 95th Percentile) PRECISE" and the data source uses template "Interface - Traffic PRECISE,64bit". Thank you, James. ------------------------------------------------------------------------------ Everyone hates slow websites. So do we. Make your web apps faster with AppDynamics Download AppDynamics Lite for free today: http://p.sf.net/sfu/appdyn_sfd2d_oct