vPC to HP client, very slow recovery

[richard.priest]

Hi,

I've vPC setup on a pair of Nexus 9k (93108) and a pair of 9348. the 9348 have LACP links to the 93108 within vpc.

Before putting them into production I wanted to test the speed of failover of a link, so have used an old-ish HP server (2008R2) and teamed x2 of the NICs, then plugged those into both 93108 with their own vPC member ports.

vPC and the LACP come up fine at both ends and connectivity is normal.

I then set off a continuous ping to the test server and shut one of the vPC member ports, it takes pretty much 19 dropped pings before connectivity is restored.

It screams STP to me, but the links are LACP, and I have the port channel and the ports set with portfast

Any thoughts?

[nazimkha]

Is the IP you are pinging HSRP VIP which may be setup as the default gw for HP server ?

[richard.priest]

No, I have a vlan with SVI's on each switch within the VPC in the same subnet as the device

[nazimkha]

Generally we have seen these deployments and there is no lag in case when one of leg goes down

Ideal / real world application of vPC would be

switch 1-------------------vPC----------------switch 2
vlan 10               HSRP / VRRP IP : 10.10.10.1             vlan 10
10.10.10.2                                                              10.10.10.3

Server pinging to 10.10.10.1