Cisco Bug: CSCut34478 - unicast route for the NVE peer loopback IP is missing on some ASIC inst
Jan 29, 2017
- Cisco Nexus 7000 Series Switches
Known Affected Releases
Symptom: This issue is specific to Vxlan - Flood and Learn. When the peer LC goes down and it is the only core facing LC on the peer switch, the peer route is not programmed properly on all the instances and hence the traffic fails. Conditions: This issue is specific to Vxlan - Flood and Learn. The peer route must be programmed on the VDC instances. This is achieved through the following two updates - a. URIB route update for programming the route on core vrf instances. b. NVE Manager peer add updates for programming the route on the VDC instances. This issue is observed when the peer LC goes down and it is the only core facing LC on the peer switch. In such a scenario, when OSPF timeouts, URIB triggers an unlearn for the peer route. However, the MAC Table timeout is longer than the OSPF timeout. Hence, the MAC table still has the Remote host MAC entry - thus the NVE peer is still seen in NVE Manager. When the peer LC is up again, OSPF sends an update for the peer route. However, NVE manager doesn't resend the peer route update since it already has the peer update. As a result, we end up with the peer route programmed only on the URIB notified VRF instances and not the other instances in the VDC.
Bug details contain sensitive information and therefore require a Cisco.com account to be viewed.
Bug Details Include
- Full Description (including symptoms, conditions and workarounds)
- Known Fixed Releases
- Related Community Discussions
- Number of Related Support Cases