%VPC-6-PEER_KEEP_ALIVE messages in Nexus 7000 logging

[Bart Mars]

Hi,

We recently experienced that our VPC peer-keepalives are not working as we expected it to be.

Last Tuesday we saw that between 2 Nexus 7000 switches the VPC peer-keepalive was not working. On one switch we saw that it stated that the link was alive, while on the other switch we saw that the link was not alive.

After shutting down both interfaces, the link was restored.

We then searched in the log when this problem had occured, and we saw numerous entries with the following text:

2012 Nov  2 07:17:47.718391 CO-DERB-02-V01 %VPC-6-PEER_KEEP_ALIVE_RECV_INT_LATEST: In domain 2, VPC peer-keepalive received on interface Eth1/48

2012 Nov  2 07:17:47.718450 CO-DERB-02-V01 %VPC-2-PEER_KEEP_ALIVE_RECV_FAIL: In domain 2, VPC peer keep-alive receive has failed

2012 Nov  2 07:17:47.975636 CO-DERB-02-V01 %VPC-6-PEER_KEEP_ALIVE_RECV_INT_LATEST: In domain 2, VPC peer-keepalive received on interface Eth1/48

2012 Nov  2 07:17:47.975688 CO-DERB-02-V01 %VPC-6-PEER_KEEP_ALIVE_RECV_SUCCESS: In domain 2, vPC peer keep-alive receive is successful

We also saw these entries on a other Nexus 7000 pair in a other datacenter of ours.

We are using NX-OS 5.0.3, but are currently in a upgrade process. We could not find a similar bug in the Cisco BugToolkit, so we are wondering what could cause this. The N7K's are connected via one UTP wire that uses Ethernet1/48 (1 gig link) on both switches. Both switchports have similar configuration, but with a different IP address.

Here is the configuration for our VPC keep-alive:

interface Ethernet1/48

  vrf member vPC_keep

  ip address 172.20.252.5/30

  no shutdown

--------------------------------------------------------------------------------

Port          Name               Status    Vlan      Duplex  Speed   Type

--------------------------------------------------------------------------------

Eth1/48       vpc keep alive t.b connected routed    full    1000    10/100/1000

vpc domain 2

  peer-switch

  role priority 8192

  peer-keepalive destination 172.20.252.6 source 172.20.252.5 vrf vPC_keep interval 500 timeout 3

  delay restore 180

  peer-gateway

  reload restore

  track 10

Many thanks in advance!

[Reza Sharifi · Accepted solution]

Hi,

can you do a "sh vpc stat peer-kee"? to see how many times the state has changed?

You VPC may be flapping because of the interval 500 timeout 3.

Can you change these to default and test again?

HTH

[Bart Mars]

Hi,

Thanks for responding so quickly.

On one pair of N7K, I see as much as 900 state changes. I think you're right with the interval 500 timeout 3. I'll going to tweak this and get back to you.

[Reza Sharifi]

Also, keep this in mined if you are planning to upgrade the OS

Beginning with Cisco NX-OS Release 5.2(1), the reload restore command and method is deprecated. We recommend that you use the auto-recovery command and method.

http://www.cisco.com/en/US/docs/switches/datacenter/sw/5_x/nx-os/interfaces/configuration/guide/if_vPC.html

HTH