• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Cisco 6500 is doing process switching

S

spidey07

No Lifer
So here's a tough nut to crack. Sup2-PFC2-MSFC2, SFM, mixture of 65xx blades and 63xx blades. Truncated mode. A large and constant 1 gbs load is added to it on a single port of a server farm vlan. Destination of this server load is 1000s of client machines.

When this load is added performance of entire switch is brought to it's knees. Backplane utilization is under 30% during load. Switch fabric utilization is never more than 10% on any channel. Take this load off, and everything is fine.

This is a core routing switch supporting some 50 SVIs.

What could be the problem? GO!
 
You already gave out the hint of process switch, so something's causing the traffic to punt to the CPU.
It could be a number of things:
1. no ip cef
2. wccp redirect is done in hardware, but the negotiation is not.
Sometime it drives up CPU
3. server sending multicast traffic, but PIM isn't set up
 
Cooky said:
You already gave out the hint of process switch, so something's causing the traffic to punt to the CPU.
It could be a number of things:
1. no ip cef
2. wccp redirect is done in hardware, but the negotiation is not.
Sometime it drives up CPU
3. server sending multicast traffic, but PIM isn't set up
Click to expand...

Yes I did.

1. CEF looks good, table and adjancies are good.
2. No wccp
3. mcast rp and everything else is good

another hint, you're never going to get it until you look at sup2-pfc2-msfc2 limitations.

*cough* HSRP
 
Cooky said:
One link I found was this:
https://supportforums.cisco.com/thread/11796
What I don't understand though, is why it would drive up your CPU, unless you're doing sub-second hello/dead and they're flapping.
Click to expand...

DAMN!

You're good. More than 16 HSRP groups configured means every group past that gets process switched as the FIB entry isn't put into hardware. I tried to give as many hints as I could without giving it away.

Damn fine catch.
 
actually cisco engineers used to recommend only using a single hsrp group on the old sup2's. Are people still using them? I would have figured they were eol by now.
 
sactwnguy said:
actually cisco engineers used to recommend only using a single hsrp group on the old sup2's. Are people still using them? I would have figured they were eol by now.
Click to expand...

The non-E 6500's are EOS now...there will be a lot of upgrades coming from this.

I am surprised how much really old technology is out there in even highly profitable companies.
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top