Can not ping HSRP VLAN address on some VLANs

[q-le]

Hi All,

We have 7 Nexus cores ie Core1 and Core 2 and they have HSRP runing.

Say Core 1 add is 172.18.10.1 and Core2 add is 172.18.10.2 and HSRP is 172.18.10.3.

We have around 100 VLANs  say

VLAN 100 with add 172.18.100.1/24 and

VLAN 101 with add 172.18.101.1/24 etc.

and recently just notice that from:

Core 1 - I can ping a vlan 100 gateway say 172.18.100.1 and 172.18.100.3 but NOT 172.18.100.2 

and on

Core 2 - I can only ping VLAN 100 gateway 172.18.100.2 but not 172.18.100.1 and 3.

and on some VLAN say VLAN101 I can ping sucessfully all three Core IPs no problems at all.

Why this happen and what cause it ?

Did I not create the VLAN 100 properly?

My network is working fine but just could not understand this.

Any advice are much appreciated.

Thanks

Peter.

[cofee · Accepted solution]

I haven't looked at your configuration so can't say for sure, but if your configuration is correct and network is running the way it's supposed to then it's quite possible you may be hitting this bug.

Bug ID - CSCum49414

Symptom: HSRP Standby interface(s) cannot ping the vIP of the HSRP Group.

Conditions: This issue is observed on the Nexus 7000 running Fabricpath with Proxy L2 Mac Learning enabled.

Workaround: Disable Proxy L2 Mac Learning.

Further Problem Description: Only IP traffic from the standby to the vIP is affected, this issue will not have any impact on production traffic.

let me know if this helps you.

[q-le]

Hi Cofee,

Will try the workaround late night just in case something go wrong.

and need to understand more about L2 Mac Learning

Thanks

Peter

[Reza Sharifi]

Hi,

Check HSRP for vlan 100 (sh hsrp brief) to make sure that one device is active and the one is on stand-by (just like the working vlans).  If HSRP is active on both switches, check your config and make sure vlan 100 is added to the trunk port between 2 switches.

HTH

[q-le]

Hi Reza,

Thank you for your response.

I checked trunk and hsrp brief as suggested and found nothing wrong.

So look like it is caused by a bug as per response from  Cofee.

Thanks

Peter