[172752] in North American Network Operators' Group
Re: Question on Cisco EEM Policies
daemon@ATHENA.MIT.EDU (Daniel van der Steeg)
Mon Jul 7 07:33:36 2014
X-Original-To: nanog@nanog.org
In-Reply-To: <53BA7457.1060908@direcpath.com>
Date: Mon, 7 Jul 2014 13:33:26 +0200
From: Daniel van der Steeg <d.p.m.h.vandersteeg@student.utwente.nl>
To: nanog@nanog.org
Errors-To: nanog-bounces@nanog.org
Ah, these objects are very useful, thanks. I have noticed the TCL policy is
run in a process named EEM TCL Proc, which I can then monitor along with
EEM Server and EEM Helper Thread. Indeed it seems to return 0 every time,
although this is not unexpected as the runtime (usually) is less then 5
seconds. Any idea if there are more processes I should monitor?
Regards,
Daniel
On Mon, Jul 7, 2014 at 12:20 PM, Robert Drake <rdrake@direcpath.com> wrote:
>
> On 7/6/2014 5:07 PM, Daniel van der Steeg wrote:
>
>> Hello all,
>>
>> I have implemented two EEM Policies using TCL on a Cisco Catalyst 6500,
>> both of them running every X seconds. Now I am trying to find a way to
>> monitor the CPU and memory usage of these policies, to determine their
>> footprint. Does anyone have a good idea how I can do this?
>>
>
> It looks like cpmProcExtUtil5SecRev is what you need. This should be
> available but it might depend on your IOS. CISCO-PROCESS-MIB shows all the
> different incarnations of it. You can also use cpmProcExtMemAllocatedRev
> and cpmProcExtMemFreedRev to track memory usage.
>
> Use cpmProcessName to find the process you want to monitor (in this case
> grepping for PID but you can look for name):
>
> [rdrake@machine ~]$ snmpwalk -v2c -c community routername
> 1.3.6.1.4.1.9.9.109.1.2.1.1.2 | grep 318
> SNMPv2-SMI::enterprises.9.9.109.1.2.1.1.2.1.318 = STRING: "ISIS Upd PUR"
>
> The 1.318 is the important bit.
>
> [rdrake@machine ~]$ snmpwalk -v2c -c community routername
> 1.3.6.1.4.1.9.9.109.1.2.3.1.5 | grep 318
> SNMPv2-SMI::enterprises.9.9.109.1.2.3.1.5.1.318 = Gauge32: 0
>
> One problem being that this is a percentage with a minimum resolution of
> 1% (integer based) so even though this is the busiest process on the box I
> tested on, I always got zero percent. It should be good for thresholding
> if you want to make sure your process doesn't spike the CPU though. Also,
> the PID might change every reboot so long term monitoring might be
> problimatic unless you can associate the process name with the other thing.
>
> Reference:
> http://tools.cisco.com/Support/SNMP/do/BrowseMIB.do?
> local=en&mibName=CISCO-PROCESS-MIB
>
> Look at this for oids:
> ftp://ftp.cisco.com/pub/mibs/oid/CISCO-PROCESS-MIB.oid
>
>
>
>> Thanks,
>> Daniel
>>
>>
> Hats,
> Robert
>
