Have someone met with the same problem? (see description below...)
We run JUNOS 15.1R6.7 on our EX3300 configured as L3 gate for local vlans and particularly for vlan 102 Ex.:
my-user@my-ex3300> show configuration interfaces vlan.102
description hosts;
family inet {
address 10.136.19.1/24;
}
There is a physical host in vlan 102
my-user@my-ex3300> show arp no-resolve | match 00:25:90:06:22:3c
00:25:90:06:22:3c 10.136.19.200 vlan.102 none
For several weeks there was no problems with it.
One day EX3300 started sending L3 forwarded packets destined to the host's IP with zero DST MAC. So, the host started discarding all the packets from other vlans and remote nets.
At the same time L2 switched IP packets from neighbouring hosts in vlan 102 was coming with correct DST MAC to the host.
So, we examined mac/arp tables and even FDB! ...
PFEM0(vty)# show shim bridge fdb vlan vlan-index 4
FDB table for VlanIdx: 4, VlanTag: 102, HwToken: 102
Legend: SP - Storm Prevention, V - Valid, s - Skip,
R - Refresh, S - Static, Vl - VlanId, Vx - Vidx
...
0 0 102 00:25:90:06:22:3C 0/5 F|1|0|1|0 0|0 0 0 0 0 0|0 0|0 0x2CE4/0x0
1 0 102 00:25:90:06:22:3C 0/5 F|1|0|1|0 0|0 0 0 1 0 0|0 0|0 0x2CE4/0x0
PFEM0(vty)# show nhdb id 1412 detail
ID Type Interface Next Hop Addr Protocol Encap MTU Flags PFE internal Flags
----- -------- ------------- --------------- ---------- ------------ ---- ------------------ ------------------
1412 Unicast vlan.102 10.136.19.200 IPv4 Ethernet 0 0x0000000000000000 0x0000000000000000
BFD Session Id: 0
Interface: 71
Flags: 2 nh_idx: 0
CMD: Route Arp Idx1: 31
CMD: Route Arp Idx2: 0
MTU Idx: 2 Num Tags: 0
Upd Cnt: 1 Tun Strt: False
Chain_nh 1416:
Hw install: 1
Mac: 00 25 90 06 22 3c
So, ethernet switching table + arp table + FDB records were OK.
We tried to plug the host to different physical switch port and that didn't help.
In the end we changed the host MAC address from 00:25:90:06:22:3C to 00:25:90:06:22:3D and the EX3300 started L3 forwarding packets to the host with correct DST MAC 00:25:90:06:22:3D !
We can't figure out if it is a software bug or some kind of FDB collision in the switch ASIC...
I hope, this workaround will help someone!