Help! OS 16 killed my Home Assistant Yellow?

xstrex · July 10, 2025, 2:12pm

@agners I’ve already restored from backup, but if it helps here the lsblk & blkid, and no, nothing is installed on the eMMC, just the NVMe.

lsblk:

NAME         MAJ:MIN RM   SIZE RO TYPE MOUNTPOINTS
mmcblk0      179:0    0  58.2G  0 disk
mmcblk0boot0 179:32   0     4M  1 disk
mmcblk0boot1 179:64   0     4M  1 disk
zram0        254:0    0     0B  0 disk
zram1        254:1    0    32M  0 disk
zram2        254:2    0    16M  0 disk /tmp
nvme0n1      259:0    0 465.8G  0 disk
|-nvme0n1p1  259:1    0    32M  0 part /mnt/boot
|-nvme0n1p2  259:2    0    24M  0 part
|-nvme0n1p3  259:3    0   256M  0 part /
|-nvme0n1p4  259:4    0    24M  0 part
|-nvme0n1p5  259:5    0   256M  0 part
|-nvme0n1p6  259:6    0     8M  0 part
|-nvme0n1p7  259:7    0    96M  0 part /var/lib/systemd
|                                      /var/lib/bluetooth
|                                      /var/lib/NetworkManager
|                                      /etc/systemd/timesyncd.conf
|                                      /etc/hostname
|                                      /etc/NetworkManager/system-connections
|                                      /etc/hosts
|                                      /etc/modules-load.d
|                                      /etc/usb_modeswitch.d
|                                      /etc/modprobe.d
|                                      /etc/default
|                                      /root/.ssh
|                                      /etc/dropbear
|                                      /root/.docker
|                                      /etc/sysctl.d
|                                      /etc/udev/rules.d
|                                      /mnt/overlay
`-nvme0n1p8  259:8    0 465.1G  0 part /var/log/journal
                                       /var/lib/docker
                                       /mnt/data

blkid:

/dev/nvme0n1p7: LABEL="hassos-overlay" UUID="07580666-6f09-475f-a0b8-9109b202bfae" BLOCK_SIZE="1024" TYPE="ext4" PARTLABEL="hassos-overlay" PARTUUID="f1326040-5236-40eb-b683-aaa100a9afcf"
/dev/nvme0n1p5: UUID="e1611890-6502-45d1-a9e7-d17e7157c131" BLOCK_SIZE="4096" TYPE="erofs" PARTLABEL="hassos-system1" PARTUUID="a3ec664e-32ce-4665-95ea-7ae90ce9aa20"
/dev/nvme0n1p3: UUID="239d0be4-3043-488c-a906-89dd87665184" BLOCK_SIZE="4096" TYPE="erofs" PARTLABEL="hassos-system0" PARTUUID="8d3d53e3-6d49-4c38-8349-aff6859e82fd"
/dev/nvme0n1p1: SEC_TYPE="msdos" LABEL_FATBOOT="hassos-boot" LABEL="hassos-boot" UUID="A3F7-1BFD" BLOCK_SIZE="512" TYPE="vfat" PARTLABEL="hassos-boot" PARTUUID="b3dd0952-733c-4c88-8cba-cab9b8b4377f"
/dev/nvme0n1p8: LABEL="hassos-data" UUID="1487715e-c6e2-4267-a5e0-a3f26d47cd68" BLOCK_SIZE="4096" TYPE="ext4" PARTLABEL="hassos-data" PARTUUID="a52a4597-fa3a-4851-aefd-2fbe9f849079"
/dev/nvme0n1p4: BLOCK_SIZE="131072" TYPE="squashfs" PARTLABEL="hassos-kernel1" PARTUUID="fc02a4f0-5350-406f-93a2-56cbed636b5f"
/dev/nvme0n1p2: BLOCK_SIZE="131072" TYPE="squashfs" PARTLABEL="hassos-kernel0" PARTUUID="26700fc6-b0bc-4ccf-9837-ea1a4cba3e65"
/dev/zram1: LABEL="hassos-var" UUID="d09fef4b-da1f-4db1-b7bc-30847881c0fc" BLOCK_SIZE="4096" TYPE="ext4"
/dev/zram2: LABEL="hassos-tmp" UUID="9bd7dd00-407d-46f9-8ca8-138215c264ce" BLOCK_SIZE="4096" TYPE="ext4"
/dev/nvme0n1p6: PARTLABEL="hassos-bootstate" PARTUUID="33236519-7f32-4dff-8002-3390b62c309d"
/dev/mmcblk0: PTUUID="f0f4a740-08b9-4031-a335-e511ddbba367" PTTYPE="gpt"

edit: Probably should clean up some of those unused partitions, though they weren’t there before I upgraded to 16 and had the issue…

innate_ideas · July 10, 2025, 4:19pm

I think I have the same issue and was able to fix it. On my Raspberry PI5, after the update to 16 the system dropped into emergency CLI mode. It appeared to me that this might be related to file system checks running a long time and the system wasn’t ready.

I was able login to ha cli mode from emergency cli mode but could not connect via the normal ethernet interface connection, WIFI sporadically worked for a minute or two. This Pi has both the WIFI and ethernet interfaces active, the ethernet interface is the primary interface in this case.

The network info command showed the gateway IP was missing from the ethernet interface. Running the network update end0 --ipv4-gateway xx.xx.xx.xx failed complaining about IP address not set. After that I tried running just network update end0 and the gateway appeared in the network info command output and everything started working.

Hope this helps someone.

PS Turns out this just got me back online for an hour or so. After another reboot HA is unreachable again.

This, for me, is the first update that has had any problems at all.

PPS Setting the ethernet IP address to static in HA vs a dynamic static reservation IP makes it so the HA system is rebooting without issue.

cgdams · July 10, 2025, 5:24pm

Another one here with the same issue. RPi 5, 8GB. First Restart went into cli emergency mode. Reboot reactivated 15.2.

Tried reinstall, this time cli started, but HA was not available over the network. Checked the network connections via “network info”: Ethernet interface end0 does not have an IP Address. Unplugged and replugged the network cable from RPi 5 and retry “network info”. Now end0 has an IP address (and gateway and dns server), and i can ping HA over the network.

Unfortunately, HA still is not availabe for connect. I did a “ha core restart” to restart HomeAssistant Core. Two MInutes later, all is well. OS Version is 16.0, HA is connected and available.

What i do not understand: I did not have single HA OS Update since updating to 15.0 that went smoothly. 15.0, 15.1, 15.2 and now 16.0: Each and every time it’s a hustle to get HA working under the new OS Version.

Will that improve anytime soon, or do we have to live with this situation?

ace_archer · July 10, 2025, 6:13pm

Yes I have CM5, the variant is the one with 8GB RAM, 32GB eMMC, 2.4/5.0GHz Wi-Fi & Bluetooth 5.0 (CM5108032)

Feppe · July 10, 2025, 6:50pm

I had the same issue in my instance running in a Hyper-V machine. I lost all Ethernet connections - tried several things.
Deleting my network adapter for the instance, starting HA up, adding a new network adapter (while HA was running unconnected), I finally got it back running …

ObiWanCanOweMe · July 10, 2025, 8:28pm

Same issue in my Proxmox VM. I tried all the suggestions in this thread but ended up having to roll back to 15.2. Going to wait for the (hopefully coming soon!) patch

metapawn · July 10, 2025, 8:50pm

I’m running HA on a mini pc. After updating to OS 16 everything seemed fine, but then I noticed my Frigate integration failed to connect. Upon checking the logs of the Frigate add-on, I notice it kept getting some kind of error which kept failing due to some network issues.

I first tried restoring from a backup but still had the same issue. I eventually just reverted back to the 15.2 OS and everything is right back to normal.

jeffcrum · July 10, 2025, 9:14pm

This seems to be getting a wider audience. Just want to throw out a datapoint.

I am running HAOS on Proxmox VE 8.4.1 installed using the Proxmox VE Helper Scripts.

I had zero issues updating the OS from 15.2 to 16.

robvandegoor · July 10, 2025, 9:28pm

I had the same issue no Zigbee on my Yellow.
Via CLI I downgraded to 15.2. Problem solved.

The command I used: ha os update --version 15.2

After the update was done, Yellow restarts and everything works fine.

johnBoy · July 10, 2025, 9:30pm

I was just going to post that my Pi4 – 2GB had updated to 16.0 successfully and had been running for a couple of hours.
However, as I was going to take a snip of the system page, a HAASS.Agent notification popped up, as shown. Then trying to log in showed the following in the Microsoft Edge Browser.
So my HA has now gone down……

johnBoy · July 10, 2025, 10:02pm

Updating this, HA restarted itself after 5-10 mins.
But it is now going through the same unavailable cycle again,

mike15 · July 10, 2025, 10:23pm

Thanks, @agners but I’m struggling a bit to follow your instructions. Basically when I attempt to connect to the HA Yellow via USB from my Mac, I just get a blank screen. (FWIW, I’m using the built-in screen command in Mac OS, and the screen /dev/… command yields an error that $TERM is too long… sorry. So I ran export TERM=“vt100” first.) I’m simply seeing a solid green LED.

FWIW, my image is installed on the NVMe (my CM5 does not have embedded flash storage), so I could presumably pull it out of the HA yellow and mount it on my Mac using a USB adapter. But then (a) I’d have no idea what I’m looking for and (b) I’d have no idea how to fix whatever is wrong. (There is a LOT on that NVMe – including system backups --, so I would NOT want to just overwrite it with a new image )

Thanks!

urmas · July 11, 2025, 6:44am

Great, thank You!

klousf · July 11, 2025, 3:45pm

Hi, after updating to 16.0 from 15.2 I only have the CLI working with the Home assistant logo. The IPv4 address is the right one and the commands are working in the CLI BUT the gui is not working, I can not ping the IP address, the leds from the eno1 network port are blinking and working. I tried rebooting also, but nothing brings the GUI up and running. Any suggestions?

Thanks in advance!

330chauf · July 11, 2025, 5:34pm

Same for me, HA has been rock solid for years. But HAOS 16 bugged out.

Running RPi 5 8GB with 1gb SSD.
Hard restart fixes the problem temporarily but after a while connection is lost.

Hope it gets resolved soon.

EDIT:

Pulled the plug and restarted HAOS. Booted fine and without issues. Then started Terminal and downgraded to 15.2 which has been stable since it came out. Hope it gets sorted soon.

As far as HAOS 16 being pulled. I still get a message that there is an update available.

Thanks for the tips and advise here.

AseKarlsson · July 11, 2025, 7:44pm

@agners, any new information found of the issues on Yellow?

JEBradley · July 11, 2025, 10:19pm

I just installed HAOS on a new Yellow with a CM5 and 500gb nvme drive. The install seemed to go fine, and it boots.
It has the following config.

Core2025.7.1
Supervisor2025.07.1
Operating System16.0
Frontend20250702.1

I have a Yellow with CM4 and a 500gb nvme. (I decided to add the second Yellow, so I could avoid any potential problems upgrading the CM4 to a CM5).

Although it boots, it has not been without problems, though. I installed “ESPHome Device Builder” and “Advanced SSH & Web Terminal.” Matter was already listed as being installed. The problem that I have is that the SSH terminal crashes when I attempt to run it. I’ll keep updating as available and see if things change. I’m not ready to spend a lot of time trying to fix it, yet.

janpeter.lourens · July 12, 2025, 8:51am

Thanks! My problem doesn’t seem to be related to OS 16 but to my attempts to add SSD. I created a separate topic.

johnBoy · July 12, 2025, 12:37pm

A further update:

Switched Pi4 off overnight.
Switched on in the morning, It booted ok and isn’t continually re-starting any more.
But 80% of the Sonoff zigbee stuff didn’t show up / re-connect.
I just let it be and the sonoff bits gradually started connecting around 3 hours later.
It all seems ok now.
?

mike15 · July 12, 2025, 3:09pm

I’d be really grateful if anyone can help me! I feel like I’ve somehow missed a large chunk of this discussion and I’m still sitting here 4 days later with a dead Home Assistant and an unhappy house

I’ve been unable to connect to the HA Yellow as a terminal. Can anyone help me troubleshoot? As mentioned above, I’m using a Mac and the built-in “screen” command. I can physically attach the the HA Yellow, I can see the new USB device appear in the device list (“ls /dev/cu.*”), but as mentioned above, if I try to screen to it, I get an error to the effect that “$TERM is too long… sorry”. So I tried ‘export TERM=“vt100”’ first, and then screen seems to run but the screen is blank.
Even if I can get connected… what do I do next? Do I just run “ha os update --version 15.2” or is the recommendation to try some other troubleshooting first? (As mentioned above, I REALLY don’t want to risk wiping the NVMe that this is installed on – as it contains a lot of history and my backups (something I intend to fix as soon as this issue is resolved). I’m assuming this just reverts the OS without bothering all of the files/settings/history on the volume itself, right?
If I can’t manage to connect to a console, is there an approach to fix this by mounting the MVNe on a different computer?

I guess I’ve been taking Home Assistant’s stability for granted since it has been so solid for years, but this is really shaking my faith in it…

Thanks!