I’ve been a long-time user of Home Assistant, primarily for Z-Wave integration, running it on various Raspberry Pi models over the years (starting with the Pi 2, then 3, and now 4). My setup also serves as a gateway for Apple Home, and I’ve been using the Aeotec Z-Stick Gen 5 (first version) for my Z-Wave network.
For many years, everything worked flawlessly. However, since the introduction of Z-Wave JS, I’ve been facing continuous issues:
Initially, the transition to Z-Wave JS brought about various problems, including installation difficulties, crashes, and dependency issues between Home Assistant updates and the integration itself.
Over the past year, I’ve been experiencing regular crashes, which could only be resolved by rebooting the system.
About two weeks ago, everything stopped working altogether. My Z-Stick was no longer detected, with the error message: “Zwave JS UI Add On Failed to initialize the driver error.”
I suspected the Z-Stick might be defective and replaced it, but the issues persisted, even with the new stick. After much troubleshooting, I managed to stabilize the system by starting from scratch with my Z-Wave network—re-pairing devices, etc. It was a very complicated process.
Now, after re-pairing several devices (though not all, as some require physically dismantling switches), the system was stable for a while. However, I’m noticing extreme latency issues again, and the driver error is randomly reappearing. This is with a new stick, the updated version, and a more powerful Raspberry Pi 4.
For the first time in years, I’m considering moving away from Home Assistant, at least for Z-Wave, which goes against why I initially chose Home Assistant—its extreme compatibility, centralization, and simplicity.
Has anyone else encountered similar issues? Are there alternative solutions or any advice on how to resolve these problems?
Any help or suggestions would be greatly appreciated!
My experience has been similar to yours. HA on a RPi3, 100+ zwave and 30+ zigbee devices, lots of integrations, scenes, automation, etc. Crashed frequently. Slow responses and high latency, etc. Never seemed to die the same way, so very difficult to debug. Switched sticks, ronfigured all devices, etc. Nothing worked.
The problem is now solved, and my setup has been quite stable for almost a year. Here’s what I did:
1 - replaced the pi with one that I had for another project. The new pi has 8GB of RAM. The original pi was a 1GB version.
2 - set up an automation that restarts HA every night at 3AM.
From a long career building hardware and software, I have my suspicions:
Something (HA, Pi OS, some integration or addon) doesn’t handle memory shortages gracefully. My original setup ran with less than 80% memory utilization, which led me to believe everything was OK. The new setup runs now with memory usage between 1 and 2 GB, so there’s always lots of free RAM. The original setup was likely spending a lot of time moving stuff between RAM and SD storage. It needed more RAM than the pi had.
Somewhere in all that software there’s also a “memory leak”. Even with the new system, I can see the memory usage creeping up. Restarting HA every day at a convenient time avoids eventual disaster.
Would it be possible for you to test a non-Pi environment, at least for a while, to rule that in or out as the source of your issues? Maybe an old laptop or desktop system where you could install HAOS either bare metal or on Proxmox or another VM. You could then restore a HA backup to that and move the Z-Wave stick over with (hopefully!) minimal effort…
I’m running ZwaveJS on HAOS on a Intel Celeron SFF PC with 8GB RAM and the Zwave network here is rock solid for years with a 5series Zwave.me USB stick (on a extention USB cable).
Your issue is definetly a hardware problem. Question is which hardware?
I had node with poor connection several years back. It would crash HA and my zwave network. Removed it and no issues since. Make sure you check device connectivity. Look for dropped packets and tx/rx issue for individual zwave devices.
RasPi is possible issue. What integrations do you have running? RasPi is great, until it is not. SD card issues, power supplies and sometimes your system needs just outgrow it without you realizing. I started with HA on RasPi B+ and had 12 total devices. Worked great I thought. I didnt realize how much it was limited until I moved to X86 hardware. The newer RasPi are better but still not sure how they work with 35+ integrations running, several services and 50+ devices.
I run both Pi and x86 server. The pi has 3 conmected zwave devices and really nothing else. Its just there to run zwavejs. 3yrs outside in 150degree fahrenheit metal case with dirt and humidity and never failed. Server hasnt failed since i found the bad node. Zwavejs is rock solid. I doubt software move will help or change your situation.
Are you doing daily heals or anything? If so, dont.
Actually, hardware was my first guess as well. When the issues started, I was running HA on an RPi3, and then I moved to an RPi4, aware of the known issues with the Z-Stick and USB3. I also changed the stick in the meantime and tested with a brand new HA instance on a Virtual Machine hosted elsewhere, but the same problems persisted. This was with the first version of the Z-Stick, though I can try again with the new one to see if the issue is the same. I’ve found that many others are experiencing similar issues with Aeotec USB Sticks in general, so maybe that’s the root cause…