Nodes all die one by one

  • Platform information:
    • Hardware: ESXI 6.5 VM:2.40GHz/512 MB/10 GB
    • OS: Debian 9.4 64-bit
    • Java Runtime Environment: openjdk version “1.8.0_181” OpenJDK Runtime Environment (build 1.8.0_181-8u181-b13-1~deb9u1-b13)
    • openHAB version: 2.3.0


I’m new to the openhab and I just started converting from 433Mhz to z-wave.
So far I have 2 switches and 1 dimmer (fibaro 2). The controller is a Aeotec stick gen 5.
I’m running a Debian VM in ESXI and have done the following from the clean install of OH:

  • Disabled the legacy USB drivers in ESXI.
  • Added the openhab user to tty and dialout groups.
  • Fixed a symlink from /dev/ttyACM* to /dev/ttyUSBzwave
  • added: EXTRA_JAVA_OPTS=""
    in /etc/default/openhab2

And everything is running fine for some time. But then one or all nodes go dead. One at the time or all at once. I think that the more nodes I added the faster it breaks. If I restart OH it works again for some time. I’ve fallen all out of ideas and I can’t get up. Please help me :slight_smile:

Maybe I should add that I had a rough start. Added/removed, reset and abused the z-wave stick. So it skipped node 3 for some reason. I only have node 2,4 and 5.

RF is fiddly. There could be a metal trash can blocking signals, and moving it an inch could fix it! Zwave uses a mesh network. The more nodes you have the better they can all communicate. If you included the nodes using the zstick and not through OH, then your routes may be wonky until you heal the mesh. You can’t do that on the version of binding you are using, but it should be available soon, or you could use the development binding.

This is normal. The nodes will progress sequentially up to 232, but there will be gaps when devices are excluded. And IIRC, I think I’ve seen the zstick jump a node all on it’s own… probably from an incomplete inclusion.

I’ve tried moving the controller (its currently in a USB-hub with external power).
It didn’t help.
Maybe I should remove all nodes, do a hard reset of the controller and add the nodes through OH as a first step.
And then if that doesn’t work, try the dev binding?

Please follow the prerequisits and change your Java version to Oracle or Zulu.

Thank you for helping me.

I though I did follow the prerequisites. Its listed here :slight_smile:

But I will change to Oracle and test.

Edit: Now running java version “1.8.0_181”
Java™ SE Runtime Environment (build 1.8.0_181-b13)
And all systems are a go. Start the clock :slight_smile:

Yes, and it clearly says that there are limitations or issues with OpenJDK.
Nonetheless, glad you got it sorted.

And all nodes are listed as dead 5 hours later :frowning:
Forgot to enable debugging on the z-wave binding but will restart with debug and post a log,

I’ve attached the complete log (it was to much to post).
I’m no expert in reading this and cant tell whats normal but one row that is sticking out is
Got I/O exception Input/output error in writeArray during sending. exiting thread.
The controller is listed as online still…
Hope you guys have time to point me in the right direction :slight_smile:
huh.xml (203.3 KB)

I’m by no means an expert, but this behaviour is a dead give away that range/network is an issue. I’d advise you to limit the distance between the nodes to max 5 meters also to the controller, preferably even less.
I’m using the newer binding in development which does heal and after a night nodes do seem to get good routes, so that’s something you can try.

My layout is as follows:
The controller sits between node 4 and 5 with the following distance.

  • node 4: 3m/10ft through brick wall
  • node 5: 4m/13ft through drywall.
  • node 2 is maybe 6m/20ft from node 5. So approximately 8m/26ft from the controller through drywall.
    If there where a range issue would it just affect the nodes to far away and not the whole network?

I’m going to try the suggestions from 5iver and reinit all nodes from the software instead of the controller. And if that doesn’t work I’ll try the dev software.

If there’s no healing in the binding, chances are that the controller(openhab) is trying to reach the other nodes via node 2.
I’d advise you to try the newer binding which has been solid for me building up a new Z-wave network from scratch which does heal. imho healing is meant to solve problems caused by nodes moving after inclusion in the network, zwave doesn’t seem to like physical changes in the network.

Maybe so. Not so good run around with the controller in your hand adding everything then :blush:
z-wave 2.4.0 installed and I did a heal of the network. Lets hope this work.

Unless you mean the development binding, that version (the current snapshot) will not do a heal. In my first post I provided a link to the discussion about merging the development zwave binding, and in that thread there is a link to the original zwave development thread, with a link to download the jar.

Yes, I hope so :slight_smile:

I’m having som trouble adding my Z-Stick Gen5 controller. I have to read in to it a little bit more.

2018-09-07 21:04:55.779 [DEBUG] [g.zwave.internal.ZWaveConfigProvider] - No bridgeUID found in getConfigDescription thing:zwave:serial_zstick:512
1 Like

Short update:
It has been running all night with no signs of trouble.

Good work all of you involved in this and keep up the good work!