For those interested, here is a _really_ rough perl script that does what is described below. It uses a dbm hash to store the data between invocations. I have done ZERO optimizations on it, so there are undoubtedly many improvements that could be made, but it seems to work well enough.
I have been using this with MRTG for several weeks now without a problem. It does depend on a working snmpget binary as I was too lazy to use the snmp module from CPAN :) I am using UCD-SNMP.
--PLB
The numbers for cpu busy percent (1.3.6.1.4.1.789.1.2.1.3.0) aren't working so good. It keeps returning "48". Does anyone know why this would be? A 5 minute sysstat (sysstat 300) is showing more variation.
That value isn't an instantaneous snapshot of the busy percentage, it's a busy percentage over the total running time of the filer.
To compute a busy percentage over a specific amount of time, you need to sample the cpuBusyTime (.1.3.6.1.4.1.789.1.2.1.2.0) and cpuIdleTime (.1.3.6.1.4.1.789.1.2.1.4.0) values, and use the changes in those values (over whatever time period you want) to compute the busy percentage, i.e.:
( change_in_cpuBusyTime ) / ( change_in_cpuBusyTime + change_in_cpuIdleTime
)
The numbers for net usage (1.3.6.1.2.1.2.2.1.10.2 and 1.3.6.1.2.1.2.2.1.16.2) keep returing zeros. Should I be using a different set of numbers for an F220?
I'm not sure why they're returning zeros. (They're returning non-zero for the machines, with more recent versions of ONTAP, that I've tried here.)
As an alternative to those numbers, you should be able to get the total KByte
s
received and sent (since the last boot) from miscNetRcvdKB (.1.3.6.1.4.1.789.1.2.2.2.0) and miscNetSentKB (.1.3.6.1.4.1.789.1.2.2.3.0).
...Tim Thompson...Network Appliance, Inc...tjt@netapp.com...