Solved: VSS on 6500

mohammedrafiq · ‎12-31-2009

Hi,

We have set up two cisco 6509 VSS switches with sup 720 in each. both Tengig ports on each sup 720 are conectected each other and one port on each 16-port 10gig card is connected eachother to achice VSS as topology daigram attached.VSS is working fine, but if one sup fail on 6509-2 its loses the whole that switch, also when adding one port on each 16port 10g card into VSS ,we have to issue "no h-w moudle oversub" command on globle per slot base, this then means that remaing 15 ten gig ports on these cards will not use for normal traffic. ( if we try to configure its say ports in over sub mode )

Please advise,

1- How we can achive full redundency if one sup fails.

2- How can we use remaining 15 ten gig port for normal traffic.

Regards,

Reza Sharifi · ‎01-01-2010

mohammedrafiq wrote:

Thanks Jerry,

That's answer my question 2, but I am still not sure about question1.Lets say my sup fails in switch 2, is there any way, that all the other cards in different slots on switch2 will still operational but handled by sup in switch 1? in VSS pair as shown in my topology .

Regards.

Hello Mohammed,

When the Sup fails the entire switch is dead and NONE of the ports or cards on the dead switch can be managed by the other Sup, that is why in a server farm environment you should always connect your important servers via 2 NICs to 2 different VSS switches, this way you have NIC redundancy and also switch redundancy.. Think about the Sup being the brain, when there is no brain there is no body.

HTH

Reza

View solution in original post

Jerry Ye · ‎01-01-2010

Correct, and that is the expected behavior for VSS.

And you are correct on devices/services behind the VSS, they should be dual homed (NIC teaming, MEC, etc.).

Regards,

jerry

View solution in original post

Jerry Ye · ‎01-01-2010

The currently release only support single SUP per chassis, however, dual SUP is on the roadmap. To answer your question

1. The idea of VSS is really you want to have devices dual homed (PO, etc.) to both chassis to achieve a full redundancy

2. performance mode (no hw-module slot x oversubcription port-group y) of the 6716 module will give you 4 ports, please see the link below for detail explanation. You will need 4 modules if you want to get 16 x 10GE port per chassis

http://www.cisco.com/en/US/prod/collateral/switches/ps5718/ps708/qa_cisco_catalyst_6500_series_16port_10gigabit_ethernet_module.html

Regards,

jerry

mohammedrafiq · ‎01-01-2010

Thanks Jerry,

That's answer my question 2, but I am still not sure about question1.Lets say my sup fails in switch 2, is there any way, that all the other cards in different slots on switch2 will still operational but handled by sup in switch 1? in VSS pair as shown in my topology .

Regards.

Jerry Ye · ‎01-01-2010

When a SUP failed, all the linecards in that chassis will be down, and there is no way behind it. From the attached link (PDF), page 36

https://www.cisco.com/en/US/prod/collateral/switches/ps5718/ps9336/white_paper_c11_429338.pdf

"Upon detecting the failure of the active supervisor, the hot-standby supervisor engine performs an SSO switchover and assumes the role of the active supervisor. An online insertion and removal (OIR) removed event is simulated for all modules in the previous active chassis to remove those cards from the running chassis inventory."

Hence, devices to the VSS should be connected to both chassis to maximize the redundancy.

Regards,

jerry

Reza Sharifi · ‎01-01-2010

mohammedrafiq wrote:

Thanks Jerry,

That's answer my question 2, but I am still not sure about question1.Lets say my sup fails in switch 2, is there any way, that all the other cards in different slots on switch2 will still operational but handled by sup in switch 1? in VSS pair as shown in my topology .

Regards.

Hello Mohammed,

When the Sup fails the entire switch is dead and NONE of the ports or cards on the dead switch can be managed by the other Sup, that is why in a server farm environment you should always connect your important servers via 2 NICs to 2 different VSS switches, this way you have NIC redundancy and also switch redundancy.. Think about the Sup being the brain, when there is no brain there is no body.

HTH

Reza

mohammedrafiq · ‎01-01-2010

Thanks,

But I am talking in VSS pair, according to Cisco Doc , which clearly says that in VSS, if active sup fails, standby Sup takes over .

I am just trying to verify because I have configured VSS and when I pulled the Sup from one switch all the chasise became unaccessable.

Regards,

Jerry Ye · ‎01-01-2010

Do you mean when you pull a SUP out of a chassis and the both chassis in the VSS pair are down completely?

When a SUP is down/pull out, the associate fabric (chassis) will be down completely. If you not that seeing this behavior, but seeing both chassis in your VSS are down complete, something might be configured wrong.

Regards,

jerry

mohammedrafiq · ‎01-01-2010

Hi,

Sorry if I was not very clear,When I pulled the sup from switch 2, only all the line cards in sw2 became unaccessable, but Switch1 fuction normal.

Now,

by reading all the response, I am write to say as below,Please correct me if I am wrong.

in VSS pair,

-If sup fail in one switch, all the line cards in that switch will become unaccessable, so the best way is to connect your server to both VSS chasis by using Daul Nic in Server.

Regards.

Jerry Ye · ‎01-01-2010

Correct, and that is the expected behavior for VSS.

And you are correct on devices/services behind the VSS, they should be dual homed (NIC teaming, MEC, etc.).

Regards,

jerry