Symptom
Static NHRP mappings stop functioning after upgrading to IOS-XE 17.3.3 on their ASR Hub Routers. Static NHRP mappings appear to be going down intermittently.
Static NHRP mappings go from working to not working. Statistics gathered from the ip nhrp details and debug logs
008711: Jan 4 10:01:00.016: NHRP: Could not find instance node for vrf: global(0x0)
008712: Jan 4 10:01:00.016: NHRP-ERROR: Incompatible Destination NBMA (UNKNOWN) for 'Tunnel20'
Conditions
This issue happens only after the software upgrade to 17.3.3 on ASR router.
Frequency of the issue seems to be 4 days on average.
Workaround
As per the below bug after removing and reapplying the configuration, the tunnels came back up.
BUG:
https://bst.cloudapps.cisco.com/bugsearch/bug/CSCvy33338?rfs=iqvred
Further Problem Description
From the spoke CLI output, the HUB’s tunnel IP entry was in correct state and we captured the message level stats to understand which messages were exchanged during the window.Few minutes later the HUB’s entry state was changed on the spoke, and we see some resolution request and purge messages being exchanged, which we believe is the trigger for this issue. Also from the trace events we see cache entry was deleted during the same time.
From the HUB logs, we see that the HUB’s tunnel entry is advertised to this spoke and other spokes as well.
Also we see that purge request was sent from HUB to this spoke to delete the hub’s tunnel address.
Feb 9 13:55:31.072: NHRP: Send Purge Request via Tunnel1 vrf: global(0x0), packet size: 83
Feb 9 13:55:31.072: src: 10.199.3.2, dst: 10.199.3.18
Feb 9 13:55:31.075: NHRP: Receive Purge Reply via Tunnel1 vrf: global(0x0), packet size: 83
