Hi, I was wondering if there is anyone our there who has an expierence in graphing the Load statistics by using the raw counters supplied by net-snmp (CPU and IO).
We utilize RRDtool to graph several things, from temperature trough IP traffic and router CPU loads, so I am not entirely novice to it, recently I have tried to visualize the CPU load on our machines by using RRDtool and querying the ssCpuRawSystem (.1.3.6.1.4.1.2021.11.52.0) and ssCpuRawUser (.1.3.6.1.4.1.2021.11.50.0) on net-snmp enabled hosts. Those values happen to be raw Counters, representing CPU time(?) used by each of the load areas. We query the data every 15 seconds using our own perl script and save it as COUNTER with AVG + MAX dimentions into the RRD Database. We have 2 problems gaphing the data though, we use following satemements to read out the data: "DEF:smax=system.rrd:ssCpuRawSystem:MAX", "DEF:savg=system.rrd:ssCpuRawSystem:AVERAGE", "DEF:umax=system.rrd:ssCpuRawUser:MAX", "DEF:uavg=system.rrd:ssCpuRawUser:AVERAGE", "CDEF:tavg=uavg,savg,+", "CDEF:tmax=umax,smax,+", The first problem is that the values in umax/uavg, and accordingly tmax/tavg will often overshot the 100% mark if the system queried was under heavy load, even for a short time. The second problem is that the value of absolute tmax changes as we adjust the graphs timeframe, for example: we have a day worth of data, if we plot a day, the top tmax will be show as 130%, and if we plot a month with the same data (Which logically results in only a small part of the graph being filled) the top tmax will be shown as 149%. Does anyone has an idea what are we doing wrong ? Thanks and best regards Kirill -- Unsubscribe mailto:[EMAIL PROTECTED] Help mailto:[EMAIL PROTECTED] Archive http://www.ee.ethz.ch/~slist/rrd-users WebAdmin http://www.ee.ethz.ch/~slist/lsg2.cgi
