Strange nexus behavior

[glen.grant]

    I'll try to explain this best i can .   We have  2 6509's  running port channels down to 2 Nexus 7000' where the 7000's are running vpc .  All the routing is still being done on the 6509's .   The nexus 7000's only have routing running via separate ospf links back to the 6509's . The plan is to eventually have the routing on the nexus so we have all the layer 3 SVI's that are on the 6509's defined on the 7000's  with hsrp but all are admin down on the 7000's for now.  On the 7000's all the routing looks correct with all the ip routes being learned on the separate ospf links.  This is the strange thing from the 7000's if try to ping  the svi addresses both the primary and secondary addresses on the 6509's they  ping fine but if you try to ping any hsrp virtual address none of those ping .  The really strange thing we see in the routing table on the 7000's  is that hsrp virtual  is put into the ospf table as a /32  with a source of the vlan that it is on on the 7000's which is admin'd down so obviously that traffic is not going to go anywhere if it's trying to send it down a admin down SVI  . The actual subnet that the SVI is in is a /24 .   So what we tried is we removed the downed svi on both 7000's and then cleared that /32  virtual address and it works correctly  and the route for that address looks correct being the ospf uplinks to the 6509's . I'm not sure what to make of it . Is there some inherent difference in the way the Nexus handles hsrp even if it's admin'd down?   Or can anyone explain this behavior .  I will add this if you stick a device in that vlan on  either the 7000's or a nexus 5000 which is below the 7000's configured with an address and the virtual address as the gateway  it rides the vpc's as it should up from the 5000's to the 7000's and up  to the 6509s .  You can also ping the hsrp virtual from this device . The problem just seems to be with the nexus 7000 and ospf  and the hsrp virtual address being in the table as a /32 pointing to a vlan which is admi'd down on the 7000's .  Strange.   Any speculation welcome...

[glen.grant]

  Well there seems to be at least 2 or 3 bugs that are fairly close to this so my guess is it is a bug .  Seems like there is more hsrp bugs than there should be on the Nexus platform.   The only problem with this is a "clear ip route *"  does not fix what we are seeing so we really don't have a fix for this issue.

STP vPC: route still installed after no hsrp

CSCuh07613

Description

Symptom:
hsrp vip is down by "shutdown" or "no hsrp", but route still installed

Conditions:
- On a N7k running 6.2(2)/6.2(2a) when a new SVI is created and HSRP is configured.
- disable hsrp by "no hsrp " or shutdown interface.

Workaround:
clear ip route *

Further Problem Description:

Customer Visible

Was the description about this Bug Helpful?

(0)

Last Modified:

Jan 2,2014

Status:

Fixed

Severity:

3 Moderate

Product:

Cisco Nexus 7000 Series Switches

Support Cases:

5

Known Affected Releases:

(2)

6.2(1.111)S5

6.2(2a)