Z-Wave node fails on update_neighbors

Tags: #<Tag:0x00007fc20af5eed8>

I’m on an RPi 3 with a Nortek GoControl USB Z-wave / Zigbee stick. I’m on snapshot 2.5.0.201911241931 of the Z-Wave binding and 2.4 stable of OpenHAB.

I have 5 of the Centralite 3210L combo Z-Wave / Zigbee plugs, and all of them are working fine except this one. It was working for quite a while, things went a bit sideways when I updated from the 2.4 Z-Wave binding to the snapshot, but I learned a bunch and everything is stable again except this device, node 16.

I’ve done the following:

  • Reinstalled the binding
  • Reinstalled the GoControl serial device
  • Deleted the xml files
  • Cleared the temp and cache directories
  • Failed the node, healed the node
  • Factory reset the device
  • Restarted the binding and OpenHAB

It always seems to end up back here eventually at the “Polling deferred until initialisation complete” stage, seemingly driven by no response to a request for a neighbor update during a routine heal.

This is a fresh factory reset log from last night, things look fine until the evening network heal.

Z-wave log files here:

Anything I’m missing that I can try to recover this guy?

What OS?
What Java version?
How did you update the Z-Wave binding? There are dependencies that need to be added for a new binding on version 2.4.
Are these battery operated devices? Usually they fall asleep before discovery is complete.

Raspbian GNU/Linux Version 9
OpenJDK Client VM (build 25.232-b09, mixed mode)

To update, the binding I followed the manual steps found here:

The nodes are all mains-powered devices.

What is the device? I recently discovered some of my oldest zwave switches can cause problems on the zwave network

It’s a Centralite 3210L combo Z-Wave / Zigbee plug. The odd thing is, I have 4 others that work fine, and the Zigbee radio works properly on this one.

And since it comes up initially then seems to fail out somehow, it seems unlikely that it’s a HW failure.

Try deleting the thing from OH and re-discovering. That sometimes works for me.

I’ve done that many times - it always slips back into the same pattern, it stops talking when polled for a neighbor list during the nightly network heal. Many hours later the controller determines it’s dead, it comes back to life, then rinse and repeat.

This log shows it again last night node 16 @ 2:37am:

If your network configuration s not changing you can likely disable the nightly heal. I only turn it on for a while after rearranging devices.