VSS - what happens when a supervisor fails?

[tvelasco]

We've been mocking up a test lab to test VSS on two 6500's.  Each 6500 has one sup720 and a 6708-10ge blade and we've established the two 10ge links between the two chassis; the first from the each chassis' sup and the second from each 6708.

My question is, what happens when the supervisor fails on one of the chassis?

I was under the impression that NSF and SSO would kick in and the entire VSS would be managed by a single sup.  If the active sup fails, the stby becomes the active and vice versa.  However, when we pulled out the active sup720 linecard (chassis1), the stby sup720 (chassis2) failed over as planned, but access to all of the linecards in chassis 1 were lost.

Is this how it's supposed to react?

[Jerry Ye]

Yes. When the SUP failed/removed, the fabric/bus on that chassis is down also.

Regards,

jerry

[tvelasco]

That's odd.  I thought the whole point of VSS was to be able to manage two chassis via a single sup if necessary.  Otherwise, what's the real benefit?

quoting from the doc found here:  http://www.cisco.com/en/US/prod/collateral/switches/ps5718/ps9336/prod_qas0900aecd806ed74b.html

What is a VSS1440?

A. VSS1440 refers to the VSS formed by two Cisco Catalyst 6500 Series Switches with the Virtual Switching Supervisor 720-10GE. In a VSS, the data plane and switch fabric with capacity of 720 Gbps of supervisor engine in each chassis are active at the same time on both chassis, combining for an active 1400-Gbps switching capacity per VSS. Only one of the virtual switch members has the active control plane. Both chassis are kept in sync with the interchassis Stateful Switchover (SSO) mechanism along with Nonstop Forwarding (NSF) to provide nonstop communication even in the event of failure of one of the member supervisor engines or chassis.

no?

[Reza Sharifi]

tvelasco wrote:
That's odd.  I thought the whole point of VSS was to be able to manage two chassis via a single sup if necessary.  Otherwise, what's the real benefit?
quoting from the doc found here:  http://www.cisco.com/en/US/prod/collateral/switches/ps5718/ps9336/prod_qas0900aecd806ed74b.html
What is a VSS1440?
A. VSS1440 refers to the VSS formed by two Cisco Catalyst 6500 Series Switches with the Virtual Switching Supervisor 720-10GE. In a VSS, the data plane and switch fabric with capacity of 720 Gbps of supervisor engine in each chassis are active at the same time on both chassis, combining for an active 1400-Gbps switching capacity per VSS. Only one of the virtual switch members has the active control plane. Both chassis are kept in sync with the interchassis Stateful Switchover (SSO) mechanism along with Nonstop Forwarding (NSF) to provide nonstop communication even in the event of failure of one of the member supervisor engines or chassis.
no?

No, when one of the sups fails the chassis for that sup is also dead.  The other sup and chassis will take over the responsibility without any service interruption, but you can NOT mange both chassis with one sup.  Think of VSS as when you stack 3750s together, when one of the 3750 fails that specific switch is dead and if you have a server single homed to it then the server is out of business too.

HTH

Reza

[Jerry Ye]

Addition to Reza's comment, to ensure "nonstop communication even in t1he event of failure of one of the member supervisor engines or chassis" you should create redundancy for every link, like using etherchannel to dual home to both chassis.

Regards,

jerry

[swakelen]

This is answering a question I was going to post, but unfortunately not the answer I was hoping for.

As has already been mentioned, I thought the point of having a VSL etherchannel between the two chassis, with one connection between 6708s and another between the Sups, in the event of one sup failing, the VSL would remain up via the 6708 connection and hence all functionality of the 'failed sup' chassis and its modules be available. Is this, therefore, not the case?

[Jerry Ye]

No. When the SUP failed/removed, the fabric/bus on that chassis is down  also. Hence, any line cards on that fabric/bus will be down.

That is part of the reason when you have VSS, you should dual home any connection to both chassis like MEC - multichassis etherchannel, etc.

Regards,

jerry

[networkalbras]

Hi

I have lived through a similar situation, however real and very unpleasant.

I have in my environment two 4506E with a SUP8 in each. When one SUP8 failed and the other did not assumed, continued as standby. I think the reasons were the VSL link and dual active detect were UPs because Other modules were available, but without management of a SUP. The Forward, did not occur even with redundant link (s) on both chassis.

Any ideas for improvement or problem analysis?

[Jon Marshall]

swakelen wrote:
This is answering a question I was going to post, but unfortunately not the answer I was hoping for.
As has already been mentioned, I thought the point of having a VSL etherchannel between the two chassis, with one connection between 6708s and another between the Sups, in the event of one sup failing, the VSL would remain up via the 6708 connection and hence all functionality of the 'failed sup' chassis and its modules be available. Is this, therefore, not the case?

Not sure what i can add to Jerry's and Reza's posts. Short answer is no this is not the case. If the supervisor in a chassis fails then all linecards in that chassis are removed from the running inventory and are in effect dead. You have to have a supervisor in the chassis to drive the linecards.

I understand what you are saying ie. if you have a VSL link and all the control protocols are managed by the active supervisor then why can't it drive the standby linecards. The reason is that VSL communication between chassis's relies on there being a functioning supervisor in the standby chassis. VSL packets are proxied by the standby supervisor to the standby linecards so there is no direct path from the active supervisor to the standby linecards.

This is why Cisco recommend dual honing everything to a VSS pair of switches. Note also that Cisco are due sometime this year to announce support for dual supervisors in the same chassis which will eliminate the issue of a single supervisor failure.

Jon

[swakelen]

Thanks Jon and Jeye. It makes sense that there needs to be a Sup to control the linecards in the chassis. Just thought Cisco were being clever with VSS and had somehow improved on that.

[Joseph W. Doherty]

Disclaimer

The  Author of this posting offers the information contained within this  posting without consideration and with the reader's understanding that  there's no implied or expressed suitability or fitness for any purpose.  Information provided is for informational purposes only and should not  be construed as rendering professional advice of any kind. Usage of this  posting's information is solely at reader's own risk.

Liability Disclaimer

In  no event shall Author be liable for any damages whatsoever (including,  without limitation, damages for loss of use, data or profit) arising out  of the use or inability to use the posting's information even if Author  has been advised of the possibility of such damage.

Posting

The only VSS configuration that keeps the chassis up with a failed sup is quad sup VSS.

[thefranmanatt]

Thanks!