System overload at 3% cpu usage / packet loss

[mgessner]

Hello everyone!

Here a short description of the setup:

2x CAT6509, sup 2t, vss config

two internet Uplinks- one 34mbit and a 2 mbit as backup

(probably doing HSRP - we don't have access to the config)

Access switches are connected redundant to both 6509.

port & port-channel config are in channel-mode on

some switches are (still) single path connected due to missing fibers

Using the Router as Default Gateway everything is working like a charm.

Changing the DG to use the core-switch causes the network get slower and slower. (Packet losses?)

The CPU ideals around 3-5%

Any ideas are welcomed

Thanks

cheers

Mike

[Georg Pauwen]

Hello,

what exactly do you mean by this:

--> Changing the DG to use the core-switch causes the network get slower and slower.

How is the router connected to the core switch ?

[mgessner]

Hello Georg,

the internet-routers are connected to single ports.

The 34MBit ist connected to 1/7/48

The Backup to 2/7/48

switchport

speed 100

duplex full

spanning-tree portfast edge

Changing the client default gateway from .254 (Internet) to .251 Core-Switch

Thanks 

Mike

[Georg Pauwen]

Sorry for the misunderstanding, what I meant to say is: how do you change the DG on the clients to be the core switch and not the router ? Is .251 on the core switch a layer 3 interface (as a matter of fact, is the entire core switch layer 3 enabled) ?

[mgessner]

Hello Georg,

we change the default gateway on the client.

With only a few clients using the core switch as default gateway we are fine.

This morning we changed the gateway via DHCP for "most" clients - and about 45 - 60 min. later no one was able to work...

Yes, the core has ip routing enabled.

Yes, the .251 belongs to a vlan interface of the core

Thanks,

Mike

[Georg Pauwen]

Hello,

ok, understood. Can you post the full configs of both the core and the router ?

[mgessner · Accepted solution]

Sorry for the delay!

problem solved: someone activated ip redirects!

just in case someone is running into it too!

thanks & Cheers

Mike

[Richard Burts]

Thanks for the update explaining the problem. It is a very interesting (and unusual) way to create a problem.

HTH

Rick