i am having big problems with my hassio installed on a raspberry pi 3.
After upgrading to the newest version of hassio my home assistant crashed a few moments later with error codes like this :
SQUASHFS error: Unable to read inode 0x27f00ad
SQUASHFS error: Unable to read metadata cache entry [5fc8724]
SQUASHFS error: Unable to read directory block [5fc8724:9fa]
I did a few restarts, but after ~15 minutes it crashes again. I deactivated the deCONZ addon and unplugged the conbee stick. After that the hassio seemed to work normally again.
So i changed the conbee stick -> still getting these freezes.
in the meantime i have tried everything possible and reinstalled hassio, tried a new rapsberry, tried 3 differend SD cards and even changed the power supplie. Still my hassio crashes after installing deCONZ. is there an incompatibility with the newest version of hassio and deCONZ?
I have the same problem.
Today I have install Home assistent, but then HA start with auto update so the HA is updated to 2021.1.15. After the update, you see on the screen of the Pi3 a list of error codes scrolling over the screen non stop. even after restart the Pi it is not possible to log in by web interface on the HA.
also it’s not possible to downgrade the version by command core update --version 2021.1.1.
After update you get message successvol, but when you want restart the core, you get a message of unexpected / unknown error, but you can’t find information in log under supersisor because the error message scroll in rapid tempo all over the screen.
Please what is the solution to prevent this problem or is it possible to downgrade the OS on another way.
I have a Raspberry Pi3B with 16GB sdcard mem for the test of HA
I seem to have the exact same issue and it’s driving me crazy.
After many hours of tests, tweaks, restores, reinstalls and what have you I got it to a point where it kinda works untill I try to actually do something in HA. And it can be anything really, sometimes it crashes on a database purge, sometimes on installing/uninstalling an add-on, once it even crashed on deploying a flow in node-red. A common denominator always being deconz devices unavailable. Then they will report and actually work for a couple of minutes, then it will all drop again and so forth. What helps is a hard reset for the Pi (I’m running Pi 4) and only actually unplugging cause at this point it will not even reboot from supervisor.
After it reboots it will usually run great till I try to change something again.
Also same thing seemed to happen on a fresh and empty install of deconz with default config. It also crashed when I restored my deconz backup to it. But without deconz all seemed fine except that other than my zigbee mesh I only have a couple of WiFi devices to test on so that might not be as thorough as I would have liked.
Upgrading to deconz 6.6.4 today did not help either.
Alright, so I’ve had a couple of days now without an incident. I’m not exactly sure which part actually helped or if it was the whole process but here’s what I did:
Fresh install and switch to 64bit (I’m using rpi4 4GB, previously on 32bit)
Change SD from Sandisk Ultra to Sandisk Extreme Pro (my money is on this step as everything is just so much snappier now - previously a fresh install + restore would take like 2 hours, this time it was all done within 10 minutes. Also everything in Supervisor is just so much faster)
Restore excluding Deconz, then fresh install of Deconz addon and just restoring config from the gateway backup file.
Reapply the Deconz integration
All seems to run smoothly now, ran another update today, very quick, without issues.
I’m having similar issues, and will try your steps 3+4 - I am already running 64bit rpi4, and replaced a dying SD card last week with a Sandisk High Endurance.
One more thing came to my mind as the latest update broke my limitlessled setup. In this case I had to delete my light_profiles.csv for it to work again as I was using one default profile for all of my lights. This made me think that it might relate to deconz problems as well. And sure enough when I went and checked my node-red automations they did control the lights fine but shown an API error on the service nodes that turned on the lights.
So in general it might be worth to first check how everything behaves without the profiles file as it’s a quick and easy test and might just solve the issue.
After having a snapshot fail to restore, I went the other way and updated everything!
I unplugged my Conbee II which allowed the RPi4 to boot and gave me the webUI. I found out later that it stopped working after 10 minutes or so. So I rebooted, then immediately updated the core from the terminal, then the OS. Then deconz (through the UI - I don’t know how to do it from the CLI). Then I plugged the stick back in (while running) to see if it would detect it, and it did. 12 hours later I restarted the core (for other changes), and it’s still booted up ok.
So the versions I updated to are:
Core-2021.2.2
Supervisor-2021.02.6
Home Assistant OS 5.11
deconz 6.6.4
I see just now that there is a 6.6.5 available, but I’m going to wait a bit and see if I have any more issues with my current config.