Spent a day tinkering with the CM4 → CM5 upgrade on the HA Yellow w/ PoE, and it did not go flawlessly.
I went from the CM4108032 + to CM5108032. The HA Yellow data was moved to a 1TB NVMe on the CM4 installation.
Following the guide to use rpiboot and then the RPi Imager, the experience was not great:
after flashing the HA OS Yellow on the CM5 eMMC, when the device restarts, it goes into a bootloop.
Apart from the device LEDs telling you it is booting over and over, and the device being pingable for a few seconds, there were no other clues what is happening.
What was actually happening is that it was still booting off the NVMe disk. After removing the NVMe disk, the device actually booted and I could restore my full back-up.
As a sanity check, I put in the NVMe disk in and yes, it would still boot from the NVMe disk and have a boot loop. I had to wipe the NVMe disk.
The problem here is that there is no progress bar of any sorts, so it was not possible to know what was happening in the background, after letting it run for a long time, I decided to SSH into the device to check the logs, but I was none the wiser. (at this point the Web UI would still not load)
As mentioned above the MAC address actually changed; this meant it would no longer be assigned the static IP if you had mapped it in your router. After addressing that and restarting both the router and HA Yellow, the Web UI became accessible.
At this point I installed the wiped NVMe disk, and moved the data there; that was nice and smooth!
However as @roelof1 mentioned, ZHA is effectively dead. It looks like /dev/ttyAMA1
is unresponsive, as the Silicon Labs Multiprotocol keeps spamming
[01:36:12:551163] Info : Failed to connect, secondary seems unresponsive
[01:36:12:551176] Info : Connecting to Secondary...