FE2 Eth-Trunk FE2
Two NE40s were connected through Eth-Trunk, which consisted of two FE members. BDF is enabled for each FE link and there is transmission in between. OSPF ran between the two devices. When one FE link (between NE40_1 and transmission) was faulty, OSPF was interrupted.
Checks showed that FE1 interface of NE40-1 was physically Down and FE1 interface of NE40-2 was physically Up. In other words, BFD configured for the link did not work.
FE1 of NE40_2 could not feel the change in FE1 of NE40_1 when there is transmission in between.
On the NE40, if the outbound interface is a Trunk interface, a Trunk member interface is selected as the outbound interface for the protocol message according to the hash algorithm, which selects an available Trunk member interface according to the availability of the link layer of the latter. If there are transmission devices in between, the status of a physical interface of the peer device is changed to be Down. The local device cannot feel the change and the status of the link layer of the interface does not change. In this case, the outbound interface selected for the protocol message may be one that is faulty. It is probably that the protocol message sent by the local device still passes the interface that is faulty and therefore protocol messages cannot be exchanged between the two devices, leading to protocol interruption and then service interruption. It is true of the onsite situation.
NE40&80 V300R002C01B551 used onsite does not implement the function of cooperation between BFD session and interface status. When detecting that FE1 is faulty on NE40_1, BFD does not notify the upper-layer module to update Trunk member table, from which FE1 interface is not removed. In this case, FE1 is selected by the protocol and OSPF is interrupted.
The function of cooperation between BFD session and interface status has been implemented on the software version 8011V300R002C01B599.
Because a Trunk consisting of two FE interfaces is used between the two devices, packets should go to the other FE link and the protocol should not break in normal cases when one FE link is faulty. The reason may be that the protocol message was discarded. It was likely that the Hello message was sent to FE1 link that broke and therefore could not be received by the peer.