Z-Wave Help: Dead Nodes

Hi everyone,

I am new to HA but pretty much love it. Have made my way around so far. Only Z-Wave gives me headaches in several respects. My main issue is that I have added and subsequently reset some devices without intending to. After including them correctly, I am now stuck with these dead nodes that (obviously) will never get in touch with HA again. Everything else works fine, but Z-Wave never fully comes up and is probably busy trying to connect to these dead nodes. I would really love to avoid setting up all devices from scratch.

Can anyone point me to a easy to follow step by step procedure how to get rid of these nodes again?

My configuration:
Raspberry Pi 3 with Razberry module
Hassio 0.62.1

This is what the nodes look like:

Any help would be highly appreciated!

I have a similar setup. I have used customize to hid the in-used z-wave items.

You may be able to add everything that is on the z-wave card to a group and then hide the card, on my setup I donā€™t do anything on that card, all control is done else where.
2018-02-01_14-06-05

1 Like

Yes, thanks. That way at least I would not have to see and bother about them. But I would really prefer to clean them up instead. I suspect they slow my installation, especially start-up big time. And as I said, Z-Wave is never really ā€˜readyā€™. Not sure if that has implications but still.

you should be able to go to the zwave control panel in the configuration area and remove dead nodes.

it only works if they are defined as dead in the attributes tho. if they donā€™t show as ā€œid_dead: trueā€ or something like that then they wonā€™t be removed.

if all else fails you may have to unpair everything from your controller, reset it to factory and re-add your nodes to get it all cleaned up again.

The key attribute is is_failed: true - only if that is shown will Remove Failed Node work.

Yes thanks, understood. My question is: what to do if I have a failed (even non-existant) node that is is_failed: false

From your screenshot it doesnā€™t appear that those nodes are dead. It looks as though you may need to wake your nodes which is usually done by pressing a physical button on each node. Also making sure your controller is close to your node will ensure that the interview process finishes without failure. What youā€™re seeing is by design to conserve battery life of each node so will only wake to complete this process every so often.

When I was getting started with ZWave on the Raspi I added those nodes but they were not added completely. At the time I was not aware I need to wake them up update several times in order to include them correctly. Thus I (hardware) reset them and added them again. They are now added correctly and working fine (with new node ids) most of the time.

So if I want to remove them, I would have to remove the working instances first, then try remove the dead nodes, then add them again and fix all references in automations etc. If I need to go that route I might as well start from scratch and reset the whole Zwave network.
But I would like to avoid that if possible at all as then lights and other things would not work until I am done with the whole process. This might casue some family friction. :slight_smile:

It seems there is no path to clean up such nodes left in limbo. Really startles me.

If you have factory reset the nodes then they will remain in HA until you manually remove them from your z-wave config file. Which you can find in the same directory as your configuration.yaml file named something like zwcfg_XXXXXX.yaml. First click on each old node in the HA frontend to get itā€™s node ID and then edit your z-wave config file to remove the Node instances that match those IDs. Then save and restart HA. You should then no longer see them.

Plus no matter how many times you remove and add a node back in, it will always retain the same name so it wonā€™t affect your automations. Itā€™s just the node id that changes.

Thanks, that sounds very interesting. Will try and report. :slight_smile:

You have to wait for it to be marked as failed.

Note that zwcfg_*.xml is a cache - if the controller still thinks the devices are included, itā€™ll re-populate the cache because as far as far as it knows they should be there.

I can confirm that deleting those nodes from the .xml does not help. It gets repopulated after the next reboot.

The nodes have been in the same state for several weeks now. I do not have the hope they will ever get marked as dead.

at this point you might have to resort to factory resetting your z stick in order fr them to go away.

If you do that remember to un-pair your existing devices before you do or you will have to factory reset those as well before you can re-include them again.

1 Like

Yipp. Thanks everyone. When I have a free weekend. :slight_smile:

I had a ā€œbad nodeā€ that didnā€™t exist, but wasnā€™t marked as dead either, that I couldnā€™t get rid of. I believe it was stuck at the Probe state. Some searching on google led to this:
https://www.domoticz.com/forum/viewtopic.php?t=17497#p134637

I applied the procedure to Home Assistant (HassIO for me) and voila, the node was gone!

For me, it went like this, your mileage may vary:

  • Go to Z-Wave Network Management

  • Click STOP NETWORK (first time I have ever done thatā€¦)

  • Open zwcfg_xxxx.xml (done that lots of times to rename nodes)

  • Find the ā€œbad nodeā€ (search for --> Node id=ā€œXXā€ <-- where XX is node number of the ā€œbad nodeā€)

  • Delete the following from the zwcfg_xxxx.xml file (this part was a bit scary)
    Find where it says ā€œCOMMAND_CLASS_WAKE_UPā€ that is the section of code that needs to be deleted. Delete this section starting with ā€œ<CommandClassā€ all the way up to, and including, ā€œā€. You will also see and in this section. 4 lines of code are deleted.

  • Save the zwcfg_xxxx.xml file

  • Go back to Home Assistant and click START NETWORK under Z-Wave Network Management. At this point I had 2 of everything listed in the Nodes section of Z-Wave NODE Management. I was pretty sure this was going to go badly, but it ended up working. :grinning:

  • Under Z-Wave Node Management, find the ā€œbad nodeā€ in the drop down box, click it, then click REMOVE FAILED NODE. Since all nodes were duplicated, I did this twice.

  • Restart Home Assistant

  • Go back to Z-Wave Node Management, each node is only listed once now (phew!) and the ā€œbad nodeā€ is now in the Unknown State instead of the Probe state it was in before. I clicked the ā€œbad nodeā€ and once again clicked REMOVE FAILED NODE.

  • Restart Home Assistant again

  • ā€œBad Nodeā€ is gone! Yay.

I didnā€™t come up with this. It was taken from the link above. Just figured others may benefit from it.

9 Likes

@RonJ103 - I know this is an old thread but thank you so much! Iā€™ve been banging my head against the wall trying to get rid of 3 ā€œphantomā€ nodes that were failure to add correctly. Even with the battery power device powered-down (smoke detector) I waited weeks and it would never show as failed. This worked on the first shot! I really, REALLY appreciate it!

1 Like

Ron,

many thanks indeed! for this walk-through. As reported above I had a few bad nodes. Now I donā€™t. :slight_smile: Great stuff!

I had to make a few slight alterations or rather additions to your method though using the original source and try-and-error. My first try ended in me re-including and/or refreshing a few of my nodes as I believe I went through the process to quckly)
I therefore post your amended walk through for future reference:

Go to Z-Wave Network Management

Click STOP NETWORK (first time I have ever done thatā€¦)

Open zwcfg_xxxx.xml (done that lots of times to rename nodes)

Find the ā€œbad nodeā€ (search for --> Node id=ā€œXXā€ <-- where XX is node number of the ā€œbad nodeā€)

Delete the following from the zwcfg_xxxx.xml file (this part was a bit scary)
Find where it says ā€œCOMMAND_CLASS_WAKE_UPā€ that is the section of code that needs to be deleted. Delete this section starting with ā€œ<CommandClassā€ all the way up to, and including, ā€œā€. You will also see and in this section. 4 lines of code are deleted.

Save the zwcfg_xxxx.xml file

Go back to Home Assistant and click START NETWORK under Z-Wave Network Management. At this point I had 2 of everything listed in the Nodes section of Z-Wave NODE Management. I was pretty sure this was going to go badly, but it ended up working. :grinning:

Under Z-Wave Node Management, find the ā€œbad nodeā€ in the drop down box, click it, then click REMOVE FAILED NODE. Since all nodes were duplicated, I did this twice.

ā€“> WAIT a few minutes! (I used around 10 mins)

Restart Home Assistant / I had to use REBOOT!

(I) Go back to Z-Wave Node Management, each node is only listed once now (phew!) and the ā€œbad nodeā€ is now in the Unknown State instead of the Probe state it was in before. I clicked the ā€œbad nodeā€ and once again clicked REMOVE FAILED NODE.
ā€“> WAIT a few minutes! (I used around 10 mins)

Restart Home Assistant again / I had to use REBOOT!

ā€œBad Nodeā€ is gone! Yay.

ā€“> If there are any bad nodes left, repeat from (I) - I had to do it three times to get rid of ALL of my bad nodes

Yay!!!
Hope this helps anyone cleaning up their ZWave network!

3 Likes

[quote] Delete the following from the zwcfg_xxxx.xml file (this part was a bit scary)
Find where it says ā€œCOMMAND_CLASS_WAKE_UPā€ that is the section of code that needs to be deleted.[/quote]

I was just trying to delete a dead zwave device but that command class does not show up for the affected device. So any more ideas on how to delete a failed device?

Thanks

You might try removing/refreshing and healing the node through the ZWave configuration in Home Assistant.

Been there, done that. No success.