I’ve hit this on an x64 machine running HAOS as a VM under VirtualBox and Ubuntu 22.04. Hit the upgrade, never came back up.
exact same here, rock solid before this.
gg
EDIT: i shutdown the VM unsafely and retried and it booted. don’t have the last 24 hours of data since upgrading HAOS, and it looks like the backups failed even though i had enough space. never seen this before but looks like HAOS 13 has lots of issues when upgrading.
Update: After a reboot, my Rpi 3 went into reboot mode again when using HAOS 13.0. I have found out that it’s when Zigbee2MQTT starts that my Rpi 3 begins to reboot. I set Zigbee2MQTT not to start during boot. Now HAOS only restarts when I start Zigbee2MQTT. I managed to downgrade to 12.4. However, this did not fix my problem, and my Rpi 3 still restarts when Zigbee2MQTT is started.
I noticed that my Zigbee CC2531 dongle is on (the LED is green), but shortly after startup, it turns off. I then tried unplugging the USB stick for my Zigbee CC2531 and plugging it back in. Then the LED lights up green again. Immediately after, I start Zigbee2MQTT, and it starts up without problems, and I can use my Zigbee devices again.
I have never previously experienced that my Zigbee CC2531 doesn’t light up even when Zigbee2MQTT is off.
Does anyone know if there is some sort of power down for unused USB devices after a certain time, and is there a way to disable this?
You can’t use your memory usage reading for anything in this case. It is a picture of the usage in that moment, but during start all the parts of HA will allocate memory resources and that is where it will be critical.
When they have booted up some of those allocations will be unused and therefore freed again and when you read the memory usage, then it might be fine, but it could 2x or 3x higher during boot.
During boot the USB ports will be renegotiated
The first negotiation will be with a basic support for basic devices only, like keyboard, mouse, flash drive and network adapter among other things.
After the renegotiation more functionality is added through the driver and higher communication speeds might also be possible.
Another one turning here because RPi 4 & 13.1 upgrade problems.
The short story: 13.0 is fine, 13.1 reboot/hang-up problems.
The long story
System Setup
- Raspberry Pi 4 Model B Rev 1.5
- Only USB-SSD connected to it
- 14 task running (top)
- running time: 10 minutes
OS Ver. 13.0 memory usage (top): 800 MiB
OS Ver. 13.1 memory usage (top) : 795 MiB
13.1 change log
2x Raspberry Pi 4 Model B Rev 1.5 both loaded with same HA version and configuration
- 1GB: shows reboot/hang-up problems
- 2 GB: runs stable
Bootloader
- 1GB: upgraded from 2023/01/11 to 2024/04/15. No noticeable difference.
- 2GB: On 2023/01/11
COROLLARY based on my experience and what I read above
- Upgrading R-Pi 4 to HA OS 13.1 may show stability problems on some R-Pi 4 regardless RAM version.
- 13.1 consumes less memory resources than 13.0.
- The R-Pi bootloader version makes no difference.
SSD-USB Note
I have tried two different SATA2USB vendors with different chipset without noticing any difference concerning stability.
UPDATE
The short story: I reinstalled HA and restored the latest stable full backup. HA has been stable during the last 7 days.
The long story:
Corrupted System
I suspect my HA system/container became corrupted at some point and since then all upgrade efforts have been to no avail leading the system to become operationally unresponsive (automations, GUI and SSH) after running for some hours however, still responding to ping.
What confirmed my suspicions was that I got Error while loading page logs
(settings->system->logs) after one of the restore operations. This error message persisted even after restoring from full backup.
Flush & Reinstall
I decided flushing the SSD, installing the HA from scratch (Raspberry Pi Manager) and restoring the latest functional full backup I had dated June 2024 (OS Ver 13.0). After this, I upgraded the system, all add-ons and HACS. The system has been stable during the last 7 days. This is a remarkable improvement coming from suffering one system lost per day.
Nice To Know
HA running on Raspberry Pi 4 Model B Rev 1.5 1GB.
Did anyone tried 13.2 yet? To see if the problem persists
Tried it, same issue
seems RPI3 support reached EOL
RPi3 still works, but with each update adding more to HA the resources that are left for handling the actual home is less and less.
It is especially the RAM amount that is an issue, which will cause a crash.
The CPU is also underpowered, but you will just feel that as lag and slow responses, but rarely cause a crash.
It is just the evolution of things in IT and nobody can stop that.
Remember “A computer never needs more than 640Kb of RAM” - Bill Gates ???
@WallyR that’s fine. Just let people know upfront that’s the case.
And remove RPI3 from the list of supported devices.
What I found interesting is that some folks have reported problems with Pi4s
There is a note on the installation page for Raspis and it have been there at least for months now.
UPDATE #2
After 10 days of stable operation, the system halt issue started occurring again, with the intervals between each event getting shorter and shorter until the system could no longer boot.
I decided to load the latest functional full backup in my proxmox. Surprisingly, this is the result just after booting up:
1.7 GiB RAM in usage.
Current version:
I wouldn’t say that RAM is the problem since my HA is now running on an Intel platform. This is clear from reports of others facing similar stability issues, even with 2 or 4 GB of RAM.
Keep monitoring the RAM. You have a memory leak
I’ll keep an eye on it and post here possible “surprises”. Here is one more dump for reference:
and the corresponding proxmox:
(uptime: 4h:00m)
Don’t use the proxmox dashboard to monitor memory.
Use the system monitor integration within HA instead. Proxmox will mark memory as being in use even though it’s not
Raspberry Pi 3 Model B is ok to get started
Not exactly what’s happening
That is a question of definition.
HA can run on a Pi3B.
No it cannot. At least from HAOS 13.1 onwards it is impossible to do so. It just bootloops as many showed here.
And it is the recommended way of running HA