If the VPLS service VMAN has LAG port and it has been reconfigured with new member ports after switch reboot, then the packets which are received on newly added ports will be lifted to CPU after deleting any of VPLS instances related to that LAG.
Steps for replicating the issue are next:
1. Initially, "PE" switch has port#1, 2 in the sharing group. Save and and reboot the "PE" switch.
2. After the reboot, verify whether VPLS instances are up.
3. Once VPLS sessions are up, add port#3, 4 in the sharing group #1.
4. Then delete one VPLS instance on "PE" switch.
5. After deleting VPLS instance service VMAN packets are being forwarded in slowpath.