Nexus 5000 Vpc delay

[dominic.caron]

Hi,

I have two 5000. Uplinks are Vpc to nexus 7000 and downlink are connected to Esx host. When I reboot one 5000, some Esx host lose connection to the NetApp SAN (Iscsi and CIFS) for 10-16 seconds. This is enought to generate errors and event corruption on disk. The 10-16 blackout occur when the switch as reloaded.

My theory is that VPC interface comme up 10-16 seconds after the edge interface(ESX). This create a blackhole effect for ESX host.

Any idea...validation...ways to avoid this...

[Oleksandr Nesterov]

Dominic

Do you use HSRP gateway on your n5k devices?

If yes - NetUp may use n5k box mac address for default gateway instead virtual address.

You can enable "vpc peer-gateway" on both switches - to allow them forward packets destined for MAC address of gateway switch:

http://www.cisco.com/en/US/docs/switches/datacenter/nexus5000/sw/command/reference/rel_5_0/vpc_commands.html#wp1219702

[dominic.caron]

No HSRP, just L2 switch.