ZWave integration - HA crashed

Hello !

I run HA OS on a RPi 3 with a NVME SSD with a USB adapter. No more SD card. Only 2 adapter : Zwave dongle and RfxCom on USB.
No problem during 6 months.

And few years before, I was using a SD card, and no problem with my ZWave integration.

On october 2024, I updated HA OS with one of the release of october (monthly update at home, no exception). I take a very long time, more than 2 hours… I don’t remember if I reboot, but HA UI suddenly appears. One error with old yaml template format. I fixed it, the error disappear, but HA UI is unstable : I reboot my rpi if I don’t see UI, sometimes it failed after 5 minutes, or after 2 hours.

I can’t use HA on web brower or on my smartphone app.

I try to reinstall everything using my last backup file. No enhancement.

I see a new update last week (HA core I think ?) and HA UI was up. I tried to update. After 5 hours without HA UI, I reboot my rpi. It’s even worse than before.

I desactivate all automations, all integration, and HA works like a charm. I even do a HA core update few days ago very quickly. I reactivate only one integration : ZWave integration.

HA crashed again.

And it’s very hard to reboot. HA is unavailable during many hours, event i I reboot it hardly, and sometimes it reappears.

I found a solution to restart HA : I unplug my Zwave Gen5 USB device.

I don’t know where to search : After HA is online, I plug my Zwave USB device again, and HA crashed.

Edit :
I click on Enable debug logging and… no crash at this time.

I tried to rebuild server (“Configure” button). No crash.

I desactivate logs after 3 hours, no crash.

And now, after 8 hours, HA crashed again…

Zwave was crashing my HA server due to poor communication on a single node.

Verify you have no communication errors such as many failed TX/RX

An upgrade may cause this since zwave routes may change and a previously stable route may become unstable

Thanks for your idea !

2 more crashes yesterday evening. I had to hard reboot this morning.
HA was online so I enable Zwave debug logging, I recalculated routes, I deleted an old inactive node, and now I cross my fingers…

This can make things worse

You must look at individual nodes and check for ooor communication. Only for nodes with high res errors would I rebuild route

Wing and a prayer is not a good troubleshooting method :wink:

I enable “Debug logging” on ZWave integration, my HA crash and automatically restart this night, but log is now disable and I have no log…

How can I analyze default if logs are flushed at every restart ?