TOPOLOGY:

• Unable to ping the Multi-homed client from Core-switch 1
• We were able to ping the client from Core-switch 2
• Issue started after ICCP flapped

• On Core Switch 2 , the MAC address of the Multi-homed client is being learnt as DL
• Also on core switch 1 , the MAC address of the client is being learnt as DL
• However, the ARP was being learnt on Core -switch 2 and not on core-switch 1 .

Possible Trigger:

• ICCP went down
• Once ICCP came up, there was an issue with ARP synchronization across MC-lag peers.

General MAC learning working scenario: for multi-homed clients:

• Traffic from Client will either reach core-1 or core-2
• MAC will be learnt locally on one core ,it would flood the MAC on all interfaces including ICL
• On one core mac learning will be local DL while other core will have remote DR
• Mac-address will sync to peer over corresponding MC-AE through ICCP (not ICL)

Resolution:

• We can try to clear the concerned MAC address on Core 1 ,post which the MAC-address should be re-learned as DR on core 1

COMMAND: clear ethernet-switching table <mac-address>

• If that does not help in resolving the issue, we can flap the concerned IRB interface / MC-ae interface.