You may not be able to do exactly what you want, but you
should be able to work out something that will indicate the
problem you are worried about. A lot depends on what
the agent is able to report.
If you are monitoring Cisco routers, they support a mib
variable for CPU 5-minute average. Just monitor that one.
It is in the private tree, cisco, local, system branch. The
reliability of this will depend on how often you poll it. Read
the description of the mib variable to help you decide. If you
can afford to poll it every 5 minutes, you get a very complete
picture of the rolling average cpu utilization.
If you are polling a device that gives only an instantaneous
cpu utilization number, you can try manipulating the rearm
value and the polling cycle, and maybe apply a Ruleset.
You can detect the beginning of a busy peroid with snmpCollect,
but only report it if the rearm does not come within 6 minutes.
by using a reset-on-match to correlate the threshold and
rearm traps.
The message part (varbind 3) of the standard threshold and
rearm traps includes a lot of information. I believe that information
is available in separate varbinds as well. Re-format those traps
with $* to see which ones they are so you can make rules based
on them. And consider defining separate sets of threshold and
rearm traps of your own. In the Netview enterprise, specific trap
ids 1000-1999 are reserved for your use. In snmpCollect you can
specify an odd numbered trap for threshold, and the rearm trap is
the next even number (eg 1001 and 1002). They will have the same
varbinds as the default thresh/rearm traps, but with these custom ones
you can more easily take specific actions based on the mib variable
involved.
Cordially,
Leslie A. Clark
IBM Global Services - Systems Mgmt & Networking
Detroit
Maxime
TRANNOY/France/Contr To: IBM NetView Discussion
<nv-l@tkg.com>
/IBM@IBMFR cc:
Sent by: Subject: [NV-L] Data
collection &
owner-nv-l@tkg.com Threshold
11/22/01 09:48 AM
Please respond to
IBM NetView
Discussion
Hi,
I would like to put a threshold on the CPU mib variable of one of my
switch.
I know that it's possible to see if the variable is over 90% and to send a
trap when this threshold is reach.
The problem is that i want only to be warn when this OID is higher than 90%
during 5 min.
Is it possible to do such a thing whith netview?
Kind regards/Cordialement.
Maxime TRANNOY
IGS - Network Delivery Support Center
Tel : +33 (0) 492 114 767
mailto:mtrannoy@fr.ibm.com
_________________________________________________________________________
NV-L List information and Archives: http://www.tkg.com/nv-l
|