Samsung wifi tablets running RDP session to a microsoft server in local LAN by mean of capsman system.
Capsman running in cap forwarding mode , chosen datapath is the same bridge LAN switch is connected to (thus MS server connected to same switch).
I'm experiencing RDP session hangs (need to re-login).
This doesn't happen if routerboard continously pings tablets.
Tablets never disconnect from wifi network
Tablets have their own static IP address on the same server subnet.
Also tried to use an old tp-link AP connected to the switch instead of capsman system, connections never hung.
With capsman, isn't like tablets are connected to server on the same L2 segment ?
(not tried yet) should "local forwarding" mode be of any benefit ?

Any suggestion please ?

Depending on overall network infrastructure ... but often it's advisable to use "local forwarding", it offers better performance at cost of lesser manageability of client data flows (which might be important or not).

Just wondering why they don't hang while under ping from routerboard.....
something kept alive ? disconnect timing ?
why it doesn't happen when linked to a common AP wired to the switch ?

Datapath via capsman involves a kind of tunnel between both devices. And tunneling involves all sorts of operation (never looked at it, but it might be even encrypted) which introduce delay jitter, possibly packet fragmentation, reordering and eventually packet loss. Effect on different apps might be different, from drop in throughput to dropped connections.

Again: if you fon't have a good reason to use datapath via capsman, it's better to use local forwarding.

I'll try local forwarding....
Question:
I'm using the same datapath for two different SSIDs in two different capsman configurations, can I use localforwarding YES and NO for the same datapath ?

I guess you can do it as you showed in the example. Hopefully setting in /caps-man configuration over-rides setting in /caps-man datapath ...