BladeSystem - General
1753600 Members
6215 Online
108796 Solutions
New Discussion юеВ

Virtual Connect 10 GB LACP trunks failover issue ?

 
Fatih Celebi_1
Occasional Contributor

Virtual Connect 10 GB LACP trunks failover issue ?

Assume you have configured virtual connect with 1 uplink trunk not using shared uplink from 10 gb x2 connector in bay 1 and bay 2 to 2 different core switches ? (in this case 2 Procurve 5406zl with 10gb modules with cx4 connectors)
(There is also ethernet modules in bay 5 and 6 which are stacked with the x1 connector to respective bay 1 and 2 )

If we degrade the trunk by disconnecting the cords or powering off the switch its working correct, the active connection fails over to the other as expected. However, we want to simulate a soft error for example in the switch by an software error or not responding switch. We tried by disabling the actual active trunk port or changed the vlan for the active trunk port then Virtual Connect still sees the link and does not failover to the standby trunk port.

Can this be solved so the active fails over to the standby ?

2 REPLIES 2
HEM_2
Honored Contributor

Re: Virtual Connect 10 GB LACP trunks failover issue ?

Fatih:

Currently, I don't see a way of Virtual Connect failing over to a standby link given the soft error (or configuration error) scenario you are describing.

For example, because link is maintained when an upstream VLAN change happens, VC won't be aware of the change.

When you say "disabling the actual active trunk port" what exactly do you mean? Do you mean disabling the port on the external switch? If so, this should cause link to be lost on the VC and then a failover.

I suspect that VC's features associated with resiliency will increase over time.
Fatih Celebi_1
Occasional Contributor

Re: Virtual Connect 10 GB LACP trunks failover issue ?

Thanks for you answer HEM.

Yes, when I say disable the active trunk port I mean the port on external 5406zl switch, my mistake...and no VC does not detect that. VC still hangs on to the trunk that its still think is working because it has a physical link on it I think. I don't think it's a big issue but its good to know that it is not working with soft errors.

By the way, we detected another strange thing with the 10gb cx4 cables. When we use 10m cable connected to 5406zl switch its working. But when we use 5 and 15 m cables the link goes up and down every 3-4-5 seconds and VC does failover and failback constantly.
And we have tried separate 5 and 15 m cables with same result. If we use the 5 and 15 m cables for stacking on VC between e-net modules or to connect a trunk between two 5406zl its no problem. Seems to be a support case, I think my colleauge has opened a support case or he will anyday...