I have example scheme.
Laptop (IP 192.168.1.1) <-> switch (tagged vlan 6 on port laptop and trunk port to SM ) <-> SM 430 (VLAN Port Type: QinQ, Provider VID: 99) <-> AP 430 (Vlan Enabled, connected port FastEthernet0/2 ) <-> Cisco 3400 ME
Cisco 3400ME configuration
interface FastEthernet0/2
switchport access vlan 99
switchport mode dot1q-tunnel
interface FastEthernet0/3
switchport access vlan 6
switchport mode access
If I’m connected other laptop with Ip 192.168.1.2 in port FastEthernet0/3 but i not ping ip 192.168.1.1.
Why it does not work. Port FastEthernet0/2 must delete Operator Vlan 99 and send tagget packet with vlan 6.
but it does not work
Well, a few things immediately spring to mind. The first is that the Cisco config doesn’t work like you think it does. Any packets coming in FastEthernet0/2 will actually be tagged with VLAN 99 on the outside - they are not being stripped. You’ll need to remove the outer 802.1Q tag on the AP and set the 3400 as a regular trunk port with VLAN 6 allowed, bring the AP traffic into a Q-in-Q trunk port, then use some sort of loop back from an access/dot1q-tunnel port to a regular trunk port to peel off VLAN 6, or set the 3400 up so that the inner Q tag is specified. As it is a given frame will actually hit your switching fabric like VLAN99(VLAN99(VLAN6(original frame))) if the MTU is big enough.
Another thing that you might run in to is exactly how the Q-in-Q is done on the radios. I don’t know about the PMP430 system but on the regular Canopy stuff in 10.5 I believe that the software defaults to an EtherType that is not actually stacked 802.1Q but a new standard for provider/customer VLANs.