For example, the following configuration makes xe-0/0/1.10 a member of S-VLAN v10, enables Q-in-Q tunneling, maps packets from C-VLANs 100 through 200 to S-VLAN 10, and enables xe-0/0/1 to accept untagged packets. If a packet originates in C-VLAN 100 and needs to be sent across the S-VLAN, a tag with VLAN ID 10 is added to the packet. When a packet is forwarded (internally) from the S-VLAN interface to interface xe-0/0/1, the tag with VLAN ID 10 is removed.
set vlans v10 interface xe-0/0/1.10 set interfaces xe-0/0/1 flexible-vlan-tagging set interfaces xe-0/0/1 encapsulation extended-vlan-bridge set interfaces xe-0/0/1 unit 10 vlan-id-list 100-200 set interfaces xe-0/0/1 native-vlan-id 150 set interfaces xe-0/0/1 unit 10 input-vlan-map push set interfaces xe-0/0/1 unit 10 output-vlan-map pop
In the above example, it is using a bridge-domain to tunnel CEs. You can also use VPLS or L2 circuit. But on CE facing side, the VLAN configuration and VLAN input/output mapping operation is the same.
My understanding is, native-vlan is supported on all types of layer 2 interfaces. In Q-in-Q case, it can determine if the packet is coming in with both S-VLAN/C-VLAN or just C-VLAN by checking EtherType(TPID)