MH-S314 (newer Z-wave Plus model) - 015F/3141/5102
The issue with this device was just a mismatch in the way the stock driver maps components to endpoints. The logic is that endpoint n maps to component switchn, and main maps to endpoint 1. So both main and switch1 map to endpoint 1, switch2 maps to endpoint 2, etc. The profile in use, switch-multicomponent-4 lacks a switch4 component, so events from endpoint 4 weren’t being mapped to a component. The fix to that was simply adding the missing component. Overall, the naming of profiles in the stock driver seem a bit inconsistent - while the switch-multicomponent-4 profile maxes out at switch3, the switch-5 profile goes up to switch5.
One oddity that I see in the screenshots for this device is that the hub isn’t currently in the Lifeline Association Group, which should’ve happened automatically. I would’ve expected that omission to make the whole switch seem non-responsive, but maybe something else is going on. The next version of the driver will handle adding the hub to Lifeline.
The fact that this switch has the SWITCH_MULTILEVEL command class, plus the naming of association groups 3, 6, 9 and 12, have me wondering if it’s actually a dimmer (or maybe it just supports sending dim start/stop commands through association). It also has the CENTRAL_SCENE command class, so it may produce some multi-tap or held button events. I’m also wondering whether there’s actually an Association Group 13 that just wasn’t showing - the pattern of 3 association groups per button would support it existing.
MH-S314 (older model) - 015F/3141/5102
The 5 association groups shown match the document @JDRoberts found, including the number of nodes allowed (5 each in groups 1-4 and 1 in group 5). Since the document says group 5 acts like the lifeline group, the Test 2 version of the driver is trying out the hub in that group. Hopefully that starts the Reports flowing, but if not then the next step will be to add the hub to groups 1-4 to see if that generates Set commands that can be interpreted as switch state.
MH-S312 (older model) - 015F/3121/1302
This one is a bit odd. Association groups 1-3 show up as I expected - presumable groups 1 and 2 are for Set commands from the 2 switches and group 3 acts as the lifeline. Groups 4 and 5 are unexpected though - I’m betting they reused the firmware from the 4 button model and didn’t bother to remove these groups. The Test 2 version of the driver is trying the hub in group 3, and if that fails we’ll either try 5 or 1 and 2 depending on results from the 4 button model.