I have a scenario where the ospf is flapping and it is configured as point-to-point network i.e the ospf communication between neighboer is multicast. We have found that when the neigh tries to exchange database (of size 1500 routes approx.)the ospf relationship flaps.
We did some furthet testing and found out that the only multicast packets with size 1497 and above do not go through. But also found out that for all the multicast packets under size 1497 have FCS header attched with wrong checksum.
If we change ospf network type to non-broadcast i.e unicast the ospf flapping stops.
I would like to know why i am seeing those packets with FCS in the sniffer plugged to transit device , ideally packets with wrong checksum should have been dropped.