Dead node rescan

I have some zwave bulbs in a few lamps. On occasion someone will accidentally turn the lamp off switch off, thus powering off the node. Soon as the poll interval hits it will show as a dead node in Home Assistant until I completely restart home assistant.

Is there any way to make Home Assistant/OpenZwave occasionally scan “dead” nodes to see if they have came back alive? Because right now, once a node is marked as dead, it stays that way.

that’s odd behavior. If the node has been added to the Z Wave network it will always be there dead or not until you specifically exclude it from the network. I recently unplugged a couple of outlets while rearranging some things at home and they went dead. Within seconds of plugging them back in they were picked up by the network and working again.

I would check your Z Wave configuration. My configuration.yaml looks like this:

zwave:
  usb_path: /dev/ttyACM0
  network_key: !secret zwave_network_key
  polling_interval: 30000
  debug: true
  autoheal: true

I also have a few devices listed in the zwave: section with some specific options like:

switch.living_room_lamp_switch:
  ignored: false
  polling_intensity: 1
  refresh_value: 1

however, the two outlets I unplugged recently do not have these options set.

Mine is set similarly. but my polling time is shorter, 5 seconds. And on the nodes in question I am using refresh_value of 3 (the bulbs take about 2.5 seconds to dim to off).

After a bit of experimentation the bulb appears to get marked dead after it misses a few polling intervals. And it will stay this way until HA is restarted. If I turn off polling, it appears to never get marked dead. But I really need polling on due to having some older Zwave devices that don’t report status on manual changes (GE zwave switches).

I will do some more experimentation tonight. Maybe once the bulb responds to a few polls it will get marked as alive again… I haven’t tested that yet. I would always just restart HA when I seen it was marked dead.

I believe it’s recommended to not set polling_interval: to anything lower than 30000, otherwise you may run into network issues and flooding devices with tons of requests. I have some GE switches as well that don’t report their status all the time (it’s caused by the add-on switch not forcing the master switch to update status). So yeah polling needs to be on, and the unfortunate side effect is the switches can take upwards of 1 minute to update their state in HA.

Ensure you only poll devices that don’t have any support for either Associations, or Hail. Five seconds is very aggressive, and risks causing issues.

For everything else, if it’s marked as dead then this article applies. In short, you can try manually waking it, or using the Z-Wave control panel to stop, then start, the Z-Wave network. You might find that running Heal or Test Network works.