What’s strange is I just tried out the RC’s and none of them have the problems I have experienced on the final release.
the PI has now no functioning bluetooth wihtin HA. However an add on i use, (theengsgateway for bluetooth mqtt transport) is working fine.
I’m running HAOS as a VM inside Proxmox, running on a Dell micro pc with 8GB RAM and an i5-8500T CPU. I noticed, while still running OS9.5, that updating to the latest core 2023.4.5 caused my CPU to spike to 100% and temps to rapidly climb to 60-65 degrees. Before the Core update, I ran a recorder purge and repack which I’ve started doing, as recent Core updates seem to take an age to complete, and running this before each update seems to help a little.
During this state, the frontend became unresponsive. I reverted to my backup from before the Core update and all was well again. I figured it was maybe my recorder database sorting itself out and doing migrations (aroung 700MB size), so I gave the Core update another try and left it running for an hour. The CPU usage eventually came down gradually to around 20% and temps to 55 degrees - still way higher than they should be. Each VM or host reboot would see high usage again for several minutes before settling to these levels again.
Proxmox processes only showed the main CPU usage coming from the KVM process, and no warnings or errors in the HA logs.
I tried disabling integrations and addons, until they were all disabled, and nothing seemed to affect the CPU usage, even after a vm restart.
Then HAOS v10 was released, so I thought I’d see of updating to that had any impact on the issue.
After several reboots, and just letting it run, the CPU has finally come back down to normal levels - idling around 2 or 3%, with the odd spike that lasts a minute or two, however idle temps are still higher than when running the previously Core version, and sit around 48 degrees, possibly due to the seemingly random processor usage spikes, and so it never gets a chance to get any lower. It used to idle around 38 degrees prior to the Core update.
On OS v10 my Bluetooth seems to all be running fine, using an external USB dongle passed through to the VM, running BTHome for several BLE devices. None of my integrations or addons appear to be broken, like some of the above reports, I’m just left with a higher CPU temp, but feel this is the latest core rather than OS.
i initially also had BT running, but it dies after some time. How long is your running on OS10 now?
Guys you are the best ! I’m running HA on a HP T630 Thin Client and after the update the CPU temperatures went from 50-60ish to 30ish 
Well done !
2 Likes
I am running HAOS on a VM in proxmox, with 4G Ram allocated.
Keeping the same version of HA (2023.4.1) the OS upgrade from 9.5 to 10.0 resulted in a ram usage spike from a steady 1G to around 3.5G. Reverting to 9.5 takes it back to 1G.
This is not my normal HA machine, it only has a few integrations and one addon (SSH & Web Terminal).
Here is a graph of the RAM usage when I reverted
@agners This is an experimental install - if you want further experiments, let me know what I can do to help.
Hi,
I had issues when updating from 9.5 to 10.0, my service didn’t restart. I ended up restarting the physical device a few times (was moving it so I could connect an external monitor).
No clue why it started working after about 5 maybe 6 restarts and just letting it sit each time until I knew it was done.
Was running 9.4 on a Pi 4 with external USB SSD. Made a full image from the SSD and a full backup before even considering updating.
Update went smooth. After restart almost nothing worked. Z-Wave and Zigbee dead, EMS-ESP via MQTT dead. What helped was patience - after a while Zigbee came back on it’s own. Z-Wave still dead. So I shut down the Pi, switched off Power (important!), waited a few seconds and switch on again.
After that all was working again - except EMS-ESP via the MQTT integration. Further investigation showed that HA renamed (and doubled) almost all EMS-ESP and Z-Wave entities with ‘_2’, ‘_3’ or ‘_4’. It was a mess. So I removed the integrations, restarted HA and added them again. Had to edit some of the automations to adjust entity names. All was working fine then. So I spent a whole day with this update and fixing all the various ‘problems’ here and there.
The reason I jumped ship from ioBroker to HA was the broken ioBroker updates who almost always broke my system (especially Z-Wave).
But anyway, now my system runs fine and I can continue with my visualizations.
Great update team, went from 9.5 to 10.0 on a bare metal NAS, no issues, and no increase in resources. Thank you!
HA OS 10 update disappeared. Triggering manual update search did not bring it back. Has the rollout been paused?
My RasPi4 system also no longer shows the update as available, but my Intel laptop build still shows 10.0 available.
I’m also not able to see the update. It had previously appeared but it disappeared before I could update. When I run ha refresh-updates or ha available-updates I’m getting a 403 error.
Hello everyone, Proxmox user here.
I had problems in the past with high memory usage, solved updating to kernel v6.x. After updting to OS10, the problem is back. Heres a screenshot - vertical red line means after/before OS10.
Any ideas? Thanks!
It has been removed from stable for RPi4/CM4: Remove HAOS 10 from stable channel for Raspberry Pi 4 by agners · Pull Request #288 · home-assistant/version · GitHub
1 Like
I can no longer run netdata; previously (9.5) I could apk add netdata and then run netdata -d but now it hangs; any help?
What hangs? The apk command or the netdata command?
Can’t be sure of the cause as it does not look like any logs past 2 am survived, but between 4 and 5 am this morning HAOS catastrophically crashed with containers including supervisor dying. Core was still running until I noticed after waking up, but when I tried to download raw logs, it also crashed. SSH was already dead by the time I tried to connect, but the network stack itself was still running.
Using the Yellow’s power button did not do anything, so I finally forced a reboot by replugging the power supply. I copied the text log and the journal folder, but so far I have not been able to find anything obviously related to the crash.
This is a Yellow with a full CM4 (i.e. not Lite), but running on NVMe only. Since there are a lot of issues with HAOS 10 (at least BT was manifest for me, as well as increased CPU usage), I have since downgraded to HAOS 9.5.
Running v10 on Intel NUC w/Built-in bluetooth is causing problems. 9.5 works fine.
2 Likes
My cpu use is also way up. It’s completely insane. I’ve seen a 35%+ increase in CPU usage. I am not running Motion Eye(which some have purported to be an issue.) Even without Frigate(which does run the CPU hard) I’m still idling at something like 30% usage on a 5th gen NUC(old, yes I know but still plenty powerful.)
It is very clearly tied to the OS update(10) as can be seen in the history shot below:
I can’t rollback with a backup with totally sucks… can someone help me out here?
Oh, also tasmota devices are unavailable, which I saw a github issue has already been logged for. Whew, this update was a doozy. sigh
2 Likes