We're using Meraki M33 (I think) and we have them all on an untagged (access) port so they get an IP address, and then they're configured for additional tagged(trunked) ports on the other vlans that they serve.
Reading
https://documentation.meraki.com/MR/Client_Addressing_and_Bridging/VLAN_Tagging_on_MR_Access_Points this: "All APs must be configured with an IP address on the native VLAN" it seems that the "management" network needs to be untagged. I could be wrong, but that works for us.
In our setup, we have an access/untagged port for the Meraki's IP address - and we allow WiFi clients (network staff) to also use that vlan/IP range, i.e. it also has an associated SSID on the Meraki
We have an additional "plain users" vlan where the VLAN tag is set in another SSID - that VLAN is tagged/trunked on the switchport as well.
And we have a third "guest-wifi" vlan, set up like the "plain users" one, just a different vlan-id.
Now, what I haven't tried is:
- use an untagged port so the Meraki gets the IP address
- then go in and change the LAN on the meraki to tagged
- plug it into a tagged port
- see if things still work.
I'd shy away from that because if the Meraki ever resets to a default, it would likely reset to needing an untagged port again - and that gets to be a lot of work really quick.
i know I haven't really answered the question, but hopefully it helps a little. Maybe you should ask in the Meraki forums to confirm my theory about the Meraki needing at least one untagged(access) port connection.
Frank