Quad VSS failover

[JAMES WEST]

Hi,

We have a pair of 6500 with 2 x VS-S720-10G supervisors in each chassis running VSS & SSO mode -

Switch 1-

5   5 Supervisor Engine 720 10GE (Active)   VS-S720-10G     

6   5 Supervisor Engine 720 10GE (RPR-Warm) VS-S720-10G      

Switch 2 –

5   5 Supervisor Engine 720 10GE (RPR-Warm) VS-S720-10G      

6   5 Supervisor Engine 720 10GE (Hot)       VS-S720-10G     

The above scenario Switch Sup module 5 failed on Switch 1, Switch 2 module 6 then took over the active role, but all the VSL links where lost between the chassis and switch 1 reset all of the line cards. This caused the single homed servers etc to lose connectivity for around 7 minutes.

Channel Mode1        Channel mode 2

Sw1Te1/5/4      <->      Te1/5/4 Sw2

Sw1 Te1/6/4     <->      Te1/6/4 Sw2

Jul 15 12:54:07.429: %CONST_DIAG-SW2_SPSTBY-4-HM_TEST_WARNING: Sup switchover will occur after 10 consecutive failures

Jul 15 12:54:31.051: %LINK-3-UPDOWN: Interface GigabitEthernet2/1/13, changed state to down

Jul 15 12:54:31.075: %LINEPROTO-5-UPDOWN: Line protocol on Interface GigabitEthernet2/1/13, changed state to down

Jul 15 12:54:29.334: %VSLP-SW1_SP-3-VSLP_LMP_FAIL_REASON: Te1/5/4: Link down

Jul 15 12:54:29.334: %VSL-SW1_SP-5-VSL_CNTRL_LINK:  New VSL Control Link Te1/6/4

Jul 15 12:54:29.614: %VSLP-SW1_SP-3-VSLP_LMP_FAIL_REASON: Te1/6/4: Link down

Jul 15 12:54:29.614: %VSLP-SW1_SP-2-VSL_DOWN:   Last VSL interface Te1/6/4 went down

Jul 15 12:54:29.642: %VSLP-SW1_SP-2-VSL_DOWN:   All VSL links went down while switch is in ACTIVE role

Jul 15 12:54:29.758: %PFREDUN-SW1_SP-6-ACTIVE: Standby processor removed or reloaded, changing to Simplex mode

Jul 15 12:54:30.203: SW1_SP:  Switch 2 Physical Slot 5 - Module Type LINE_CARD  removed

Jul 15 12:54:31.599: SW1_SP:  Switch 2 Physical Slot 1 - Module Type LINE_CARD  removed

Jul 15 12:54:33.675: SW1_SP:  Switch 2 Physical Slot 2 - Module Type LINE_CARD  removed

Jul 15 12:54:34.799: SW1_SP:  Switch 2 Physical Slot 3 - Module Type LINE_CARD  removed

Jul 15 12:54:35.595: SW1_SP:  Switch 2 Physical Slot 4 - Module Type LINE_CARD  removed

Jul 15 12:54:36.347: SW1_SP:  Switch 2 Physical Slot 6 - Module Type LINE_CARD  removed

Jul 15 12:54:37.386: SW1_SP:  Switch 2 Physical Slot 7 - Module Type LINE_CARD  removed

Jul 15 12:54:38.122: SW1_SP:  Switch 2 Physical Slot 8 - Module Type LINE_CARD  removed

Can some one let me know if the VSL link on Sw1 Mod 5 fails, should this take down the whole switch? Or should we just lose the Module and then the Supervisors determine their Active/Hot/Stand-by roles?

Thanks,

James

[Reza Sharifi]

James,

Currently even with quad Sup per VSS pair when the primary Sup fails the switch will reboot.  In the future IOS releases, when for example module 5 fails, the switch will fail over to module 6 without rebooting, but that is not the case today.

HTH

Reza

[JAMES WEST]

Reza,

Thanks for the reply +5

I have been reading and re-reading the PDF's and just wanted to make sure that was the case.

Hope you have a good weekend.

Thanks,

James

[Eduardo Aliaga]

So what's the advantage of Quad VSS versus original VSS ?

[Jon Marshall]

Basically the only advantage is that with single sups if the active sup failed and the switch rebooted then it wouldn't come up because the sup has failed.

With 2 sups in the chassis if the active sup failed and the switch reboots then it can come up with the other sup.

Jon

[prasad.gsmc]

IS it working on SSO ?

[Reza Sharifi]

No, not yet.  It is RPR right now, but in the feature it will be SSO.

[Kevin-John Beasley]

Out of interest do you or does anyone else know when Quad Sup SSO is expected to be introduced into the IOS codebase? It seems to be a long time comming and now that Cisco is tied up with the new Sup2T and features they might want to implement on that I think it might be even longer before we see such a feature?

Thanks,

Kev