Topology: PC1-> NE40-1->FE board (Huawei SDH)――(Huawei SDH) FE board ->(NE40-2)->(PC2)
1. Ping PC2 from PC1. When the bytes of data packets reach 3K, few packets are dropped. When the bytes of data packets reach 8K, many packets are dropped.
2. From packets capture, when it is dropped, NE40 receives fragmented and reassembled packet with ICMP TYPE as 11 (Time-to-live exceeded), CODE as 1 (Fragment Reassembly time exceeded).
Through check the negotiation mode of FE interface of NE40 and interface of SDH transmission are not consistent. Change SDH transmission negotiation mode as self-negotiation and agree with NE40.
And then CRC fault does not increase and there is no dropped packet.
On NE40 compare with mirroring and header capture for FE por tof SDH transmission connecting NE40-1 and port of SDH transmission connecting NE40-2. Pin gpacket from any direction, and it is sent but the received port often does not receive all fragmented packets. For example, a packet with 8K is divided in five fragmentations. It is possible that one or three is/are dropped. Then there is no response.
From FE port of NE40-2, there are many CRC faults, but at FE port of NE40-1 there is no CRC fault.
NE40-1, NE40-2 and Huawei SDH link are problematic.