Ad blocker detected: Our website is made possible by displaying online advertisements to our visitors. Please consider supporting us by disabling your ad blocker on our website.
I have installed Fridh - ucd/net - System Stats data queries on two different servers. Now CPU usage graph generated by this data query is having strange issue. On one server it is generating CPU usage in Percentage (which i want) and on another server it is generating CPU usage in milli seconds (m). I want both graphs to be generated in percentage of CPU usage. How can i change this graph ? I am attaching graphs from both the servers.
Yeah but i am reporting screen shot of these graphs to higher managment and it looks odd when one graph is having percentage and one is having milli second. I want that to be converted in percentage. Does any one has idea how can i achive this??
I am having some different type of issue on the CPU Utilization graph on other server. This is a newly built server and i haven't installed any application on it but still its CPU utilization in cacti showing around 70%. When i cross checked the same on the server it isn't using that much. I have used same template which i corrected as per above recommendation. I don't know what is the problem with this server? I deleted this server from cacti and reconfigured again but still it is having same problem. I have kept snmp setting on server side same for every server. I am attaching screen cap of the graph.
Though I'm not sure why your Logical CPUs are shown as 0,
catchvjay wrote:
This is a newly built server and i haven't installed any application on it but still its CPU utilization in cacti showing around 70%.
-snip-
I don't know what is the problem with this server?
Well on server side there is no problem as far as CPU utilization concerned. It is showing 99% idle. Can this be snmp related issue?? Does any one has idea about this??
0.8.7e on old server, 0.8.7g on new server.
Different levels of snmpd
Old server never had clients with more than 16 cores. The new server is working for 8 cores at least.
I suspect something odd in the templates for large numbers of cores. If I find anything I'll post back.
OK, I have resolved this. As suspected it's related to the number of cores. In the associated data template the maximum value for each of the data sources was set to 2000, which was perhaps a reasonable sanity check allowing for 20 cpus.
When there are over 20 cores, and any one of the sources would exceed 2000, it's value would be recorded as unknown/nan. For an idle server this would always happen to "idle". When one of the data sources was invalid, the calculation for number of cpus would fail, and the cpu numbers would not get divided by the number of cores.
So, you can either increase the data template max for each source to say 7000, or set it to zero for no maximum, and remember that if any rrd files have been created already they will have the max set already so they should be removed or fixed with rrdtool.