VSS on 4500/4500x feedback, stable?

[ds6123]

Hello Everyone,

For those of you who have implemented the recently released VSS on the 4500/4500x platforms, how is it running?  Stable?

I have a pair of 4500x's staged and things seem to work fine.  At least in the staging/lab area.

I'll be installing some 4500x switches into production in a few weeks and I have a few questions. 

• The configuration syntax looks pretty much the same as on the 6500's implementation of VSS.  Is the underlying code the same?  Or was it rewritten from scratch so I should expect huge issues in production?
• The image available on CCO ( cat4500e-universalk9.SPA.03.04.00.SG.151-2.SG.bin ) was released 5 Dec 2012.  Today is 24 Jan 2013.  So no critical bugs needed to be fixed in the last 6 weeks?

For now, I just need basic layer-2 services (vlans, stp, port-security, storm control, qos, etc.)

I was reading the release notes and the only thing that I noticed was

Packets that are routed on the same Layer 3 interface (or SVI) which they entered on are dropped if they are received on the VSS standby.

Workaround: None. CSCub63571

Uh oh.  That doesn't sound good if I need to do layer 3.  This might require some painful redesign of the existing network.

Any feedback would be appreciated.

[ds6123]

Thought I'd bump my own post to let everyone know I moved the 4500x VSS into production about 8 days ago and so far, so good!  Very simple layer-2 deployment, though.

[harmkanters]

Same issue here.

I hope CSCub63571 will be fixed asap

We have L3 interfaces with workaround to disable cef.

[ericsson]

This is information that I got from a Cisco sales engineer on this bug.

Please note that CSCub63571 / CSCuj67614 applies only to the following type of traffic:

-          Ingressing on a vnet trunk on the VSS standby switch

-          Routed to a network where the next-hop is back out the same subnet where the traffic ingressed for routing

So, this bug only refers to ports that are part of an EVN trunk.  EVN is kind of like a VRF-lite implementation.  This bug doesn't affect normal/simple L3 interfaces.

[Garry Cross]

Perhaps this experiance is related to this bug or not.

We a pair of 4500x staging at the moment.

We went to upgrade to 3.5(1)E.

After doing the upgrade none of the 3850 switches was reachable from 4500x.

3850x could ping and telnet to each other.

Fa1 was not reachable from outside. We have this interface from both switches plugged into our infrastructure.

After pulling our hair out we backed out to original  03.04.00.SG

Everything was back to normal except for 1 3850 still not reachable from 4500x.

We shut/no shut Po40 but no difference.

We reloaded the 3850 and it was again reachable from 4500x.

This is making me really nervous about putting these into production.