>>>>> "tk" == Terje Krogdahl <tekr@nextra.com> writes:
> The CLI methods have already been mentioned, but you can get this
> via SNMP as well. Look at the CISCO-PROCESS-MIB. The trouble is,
> on all 7500s I've encountered a rather crucial table always shows
> zero:
> cpmCPUTotalPhysicalIndex.1 = 0
> cpmCPUTotalPhysicalIndex.2 = 0
[...]
> This means that you can get a lot of CPU stats pr. card or (or RSM),
> but there is no way to know which entry represents which card/RSM.
Correct. These ...PhysicalIndexes are supposed to point to rows in
the ENTITY-MIB, which hasn't been implemented for the 7500 yet, at
least not on any IOS release known to me.
I think the only thing you can count on is that the RSP always has
index 1.
> Has anyone managed to work around this?
Not really. What's worse is that on our routers, the CPU indexes
change every few days without apparent reason. E.g. a freshly booted
router may look like this:
: leinen@babar[leinen]; snmpwalk -Os test cpmCPUTotal5min
cpmCPUTotal5min.1 = Gauge32: 3
cpmCPUTotal5min.2 = Gauge32: 2
cpmCPUTotal5min.3 = Gauge32: 1
cpmCPUTotal5min.4 = Gauge32: 1
cpmCPUTotal5min.5 = Gauge32: 1
And a router which has been running for a few weeks looks like this -
note the high indexes (37,38,39,40):
: leinen@babar[leinen]; snmpwalk -Os ce1 cpmCPUTotal5min
cpmCPUTotal5min.1 = Gauge32: 4
cpmCPUTotal5min.37 = Gauge32: 24
cpmCPUTotal5min.38 = Gauge32: 28
cpmCPUTotal5min.39 = Gauge32: 20
cpmCPUTotal5min.40 = Gauge32: 10
This makes it pretty hard to track them with MTRG (which would be very
useful).
-- Simon.
This archive was generated by hypermail 2b29 : Sun Aug 04 2002 - 04:12:27 EDT