Reducing Risk During Upgrades/Updates

Installation
1

Hi, I’m running HA ( Home Assistant Core) on a dedicated Raspberry Pi 4.

Twice now it’s blown it’s brains out when I’ve applied an update. First time it was hopelessly corrupted and wouldn’t boot. Nor could I work out how to find the backups.

Since then I’ve set up the service that stores backups on Google Drive.

This morning I had another failed update, this time it boots and runs, but the Solaredge Modbus integration won’t work.

I did a new install and restore which got the system back and running. But this still leaves me with two updates pending, the Solaredge integration and Home Assistant Core.

Is the a lower risk way of doing this? Or a better way of reverting after another fail, reinstall and restore takes quite a long time.

2

I’m assuming you’re referring to the Solaredge Modbus custom integration. If so, did you read the release notes for the latest release?

image
image987×302 21.6 KB

It’s also mentioned in one of the closed issues on github here.

All you have to do is follow these simple steps:

  • Update HA Core to 2026.7 & wait until HA restarts. At this point Solaredge will stop working.
  • Update Solaredge to 2.3.0 & restart HA. It should now be working again
  • (In case of emergency) Take note of the Solaredge logs. Restore your backups. Report the issue on the Solaredge Modbus github.
3

Install as a virtual machine, e.g. on Proxmox. Back up the vm before doing the updates. Then it’s completely reversible.

4

Thanks, that’s the sequence I used. Updating Solaredge failed with a HACS authentication error requiring me to change my Github password and some kind of verification key.

Following that the SE integration appeared, with the latest version and no update available. But if also said it wasn’t loaded, and obviously wasn’t working. Nor was there any apparent way of removing it so I could try reinstalling.

I get that every now and then software has a wobble so I’m not really bothered about an autopsy on that failure. But I’d like to see if I can minimise the risk in future, and speed up recovery if it goes pear shaped again.

I don’t know if the Pi 4 has enough oomph to run a hypervisor, but I agree running in a VM makes recovery as easy as “pie”.

Tactically I’m going to leave it, running exactly the same steps and banking on a different outcome doesn’t sound appealing. I’ll have another go when either comes out with their next release.