Well somehow it’s working now after excluding it and re-including it multiple times. Finally, I restarted OpenHAB and from a different room of the house (not right next to the controller), I kept on pushing the action button.
Based on the manual, to wake the device, I’m supposed to press and hold the button for 2 seconds. However doing this, even though the LED indicator on the device responds correctly (turns on red to indicate it is awake), nothing appears in the binding debug logs. The only way I could get some response on the binding side was to press the button once, which according to the manual sends an NIF.
The behaviour I was seeing before sounded similar to this: https://github.com/openhab/org.openhab.binding.zwave/issues/127
The device would “wake”, then OpenHAB would try to communicate with it, but it seems like the device isn’t listening and never responds. When I saw in the logs that the binding was starting to complain about retries, I would press the action button again, and that would at least prevent it from going to the offline state. Eventually after continually pressing the button (not according to how the manual seems to suggest the wakes should work), the binding managed to get all the information from the device and it seems to be connected (at least, reporting the motion alarm now).
I’m not sure why this seems so flaky for me - does anyone have any tips or suggestions about getting more consistent results with the inclusion process? With the first TriSensor I included, I didn’t seem to have any of these problems at all and it just worked without a hitch.
I did send a support request to Aeotec and will see if they have any suggestions (I’m wondering now if it’s a problem with the device or a wireless communication issue).