I do have some experience on Unix and Linux systems from several years ago, but Docker is a new to me. I obviously erroneously thought I was having a root shell in my terminal window…
Could you please give me some advice on two things:
After moving the data disk to NVMe, what is the correct way to verify that the ~9 mounts (kernel and system partitions) still exist on the eMMC disk?
If I decide to keep the data disk on the eMMC, is there a way to mount the NVMe disk so that it can be used e.g. for storing backups? What I have tried was to open a terminal window and use a simple mount command, but I received a “permission denied”. This made me somewhat confused since “whoami” still returned “root” in the terminal window. So I think that I may be missing all the necessary commands and arguments that are needed in the Home Assistant environment…
1, That’s what my post tried to give you - ls /dev/mmc* /dev/nvme*
2. Not easily. HASSOS is designed to be more of an appliance so you would need to boot with a generic Linux image to gain access to the tools needed to remove, repartition, and format the unused areas. I’d not recommend this as the space gained is small, the risk of error is high, and storing backups on the same hardware doesn’t give any resilience. The number of writes to the boot OS disk is very small (good for a soldered-on eMMC) and for the user database area very high (good for replaceable NVME SSD).
Ok, I’ll give up with my plans to use the NVMe as a backup device.
And yes, I did see the /dev/mmcblk* device names, but I had no visibility to the hassos-boot, hassos-kernel0 etc. mounts / partition contents.
After moving the data disk all mounted partitions were on NVMe device (and even before the move the hassos-* mounts where hidden somewhere). Maybe that’s because of the Docker stuff, so I guess I should just quit thinking Home Assistant Yellow as some kind of Unix/Linux system…
Yeah - that’s pretty much what I thought when Solaris zones appeared years ago, and the word hypervisor replaced chroot! You might have uid zero, be called root, but you’re still stuck in a virtualised jail with limited powers.
There’s a shell, SSH/SCP, Bash, Python and MQTT - that’ll do for my dev tools!
Once upon a time I’d be trying to consolidate all my workloads onto one server (NAS, SMTP, WWW, automation, prod/dev/pre-prod) but sometimes it seems easier to just get another RPi4 appliance than mess about with Docker / Podman / VMware and almost working containers. Turtles all the way down…
Last night my Yellow decided to stop working. Could not get it to do anything, so reinstalled yellow code base and at least got it back up on the network. Trying to restore, discovered it ran out of space, then I recalled the “move data disk”, which is when I get the “no suitable storage found” greeting.
Got here and noticed the comment about the blue LED. When powering up, the light flashes, does not stay illuminated. Should it stay lit? Has my original SSD bit the big one?
Get a cheap USC-C dock and try the M.2 SSD in another machine.
My preference would be to use a RPi4 for testing to give the full range of fsck and fstrim, but other OS would also work (e.g. Win10 + Samsung SSD = Samsung Magician tools).
Worst case, wipe the SSD and try Move data disk again (I use dd if=/dev/zero ... or shred -n1 -z ... to zero the whole device). Even worse case, SpinRite works on SSDs.
Just an FYI - I recently found myself migrating from one Samsung NVME drive to another Samsung NVME drive. I found that Samsung Magician was not able to ‘see’ a Samsung NVME drive in my USB external enclosure (SSK SHE-C325). Samsung Magician was only able to ‘see’ the drive when I connected it directly to one of my motherboard’s m.2 NVME slots. Also, the Samsung Data Migration tool, which can see a drive in a USB enclosure, can only copy a OS Boot Drive to another Samsung drive. I was replacing my main boot NVME drive with a larger, faster model. Then I wanted to use that old drive to replace a second Data drive. I ended up having to use Macrium Reflect to accomplish that task.
Meanwhile, my HA Yellow is still humming along nicely. It is booting directly off of a 1TB Western Digital NVME drive.
But do you have an SSD in the M.2 slot (and used the “Move the data disk” before)? I re-read your message and could not be sure of it (especially because then you mention the “original” SSD)
If you didn’t install an SSD drive yourself, then you’re using only the built-in storage (quite limited, I can imagine you running out of space). It’s normal to see that message.
Does it work to install HA into the SSD from start?
The manual says:
" * Advanced: Forcing installation onto the NVMe SSD when using CM4 with eMMC:
Press the blue button while the yellow LED is on constantly (during the 5s window, see next step)."
But can not get it working… Need to hold the button? short press? anything specials besides this?
That is how I installed HAOS on my NVME SSD drive, before installing the NVME drive in the HA Yellow. You will want to make sure your CM4 does not try to boot off of the eMMC (if equipped). If your CM4 does have eMMC, then I believe as long as the eMMC is not bootable, then it should try the NVME drive next.
There is a way, using jumpers on the HA Yellow, to allow the eMMC module to show up within your PC as a USB based external flash drive. This requires installing software on your Windows, Linux, or MAC computer. Doing so, I would think you’d be able to wipe the eMMC drive clean.
A similar procedure allows one to change the boot order on the CM4 module to prefer the NVME drive over the onboard eMMC. This is what I did to my CM4 which I received months before my HA Yellow shipped. I used the Raspberry PI CM4 I/O board to do this procedure. The same process can be done using the HA Yellow, again with the jumpers in the correct locations and using software from Raspberry Pi. It has been over a year since I did this, and I do not recall all of the steps I followed. I found some online guides on how to change the boot order of a CM4. I am pretty sure I had to use a Linux machine to compile and make the proper changes. I used a RPI4 to accomplish this, as I know MS Windows could not be used for this purpose.
Update: Looks like I posted some information earlier in this thread regarding the above…
Damn, don’t think i can do that
Wonder why the “press the blue button” does not work out of the box. It is the first “yellow LED is on constantly” right? I have one in the beginning right after the setup is loaded and one time at the very end…
but never 5 second, more like 2 or max 3.
I am trying to install HA on an ssd inside the yellow but have a cm4 with emmc. I got as for as to boot select boor with ssd (on initial boot wait until yellow led lights up for 5 seconds and press the blue button) however once i do that after a couple of minutes (and seemingly after successful install because it has the red led on after and i have to unplug it and plug it back in) after rebooting the yellow goes offline and the green and red led stay on. (I tried to check logs over uart but i am not getting anything making me think its stuck in the bootloader. Anyone have any idea? Do i have to format the ssd? (Its a new ssd)
Fixed: flashed a usb with the yellow installer software on it plugged it in and ever since then its been booting of the ssd no problem. No clue as to why.
New here I also have a HA yellow for some time now, have a Samsung V-Nand SSD 860 Evo SATA M.2 500GB.
When I try to move it comes up with No suitable storage found is this drive compatible or do I need to buy a different one?
(especially because then you mention the “original” SSD)
