- A virtual community of network engineers
 Home  BookStore  StudyNotes  Links  Archives  StudyRooms  HelpWanted  Discounts  Login
RE: 3550 CPU OVERLOAD 80% [7:76304] posted 09/27/2003
[Chronological Index] [Thread Index] [Top] [Date Prev][Date Next] [Thread Prev][Thread Next]

AFAIK, SNMP engine handles mib and counters lookups. ip snmp handles
packing and unpacking requests and replies from/to ip packets. See show
Pool manager handles buffer pools allocation. See show buffers.
I have no idea about VUR_MGR. It seems to be some hardware specific
process. Probably ASIC or some internal interface management.

On 3550 switches generation of ICMP errors, ARP requests/replies,
routing protocols and others are handled by the CPU. Everything else
related to forwarding is handled by the ASICs. The rate at which the
switch can handle central-cpu tasks is limited by several factors (cpu
speed, internal cpu to fabric interface speed...) so you can't expect
for example the switch to produce ICMP errors at wire speed. You have to
take measures to protect your switch from being overwhelmed by these
central cpu tasks.

int X
 no ip unreachables 
This will disable generation of ICMP unreachable errors related to this
interface (access prohibited (acl), host unreachable, protocol and port
unreachable for packets addressed to the switch). It is especially
meaningfull to place this command on interfaces that you know will drop
a lot of traffic on (e.g. Null0).

Take a look at
for a good example of how to secure a Cisco router.

Boyan Krosnov, CCIE#8701
just another techie speaking for himself

> -----Original Message-----
> From: nobody@xxxxxxxxxxxxxx [mailto:nobody@xxxxxxxxxxxxxx] 
> Sent: Saturday, September 27, 2003 4:11 AM
> To: cisco@xxxxxxxxxxxxxx
> Subject: 3550 CPU OVERLOAD 80% [7:76304]
> hello all,
> started to get this msgs a few days ago. there have been no 
> changes to the
> config. I am not too worried about them since they go away 
> when the CPU is
> freed up and returns to 10-20% but I would like to take the 
> time to learn
> something about the switch
> A sh proc cpu revealed IP Input (obviously) as the main 
> culprit but a few
> other processes were using CPU time as well. ( More than I 
> think would be
> normal)
> SNMP Engine 4- 7 %
> IP SNMP 3-5 %
> Pool Manager 2-3 %
> VUR_MGR bg proc 3-4 %
> I am not sure waht these are and have been hunting around CCO 
> without much
> luck. Can someone give me a steer in the right direction as 
> to what these
> are and if the levels are egregiously high level. 
> Particularly, I haven't
> had any luck on VUR_MGR and the difference between SNMP 
> Engine and IP SNMP.
> Thanks!
> Jim
> **Please support GroupStudy by purchasing from the GroupStudy Store:
> FAQ, list archives, and subscription info: 

Message Posted at:
**Please support GroupStudy by purchasing from the GroupStudy Store:
FAQ, list archives, and subscription info: