Update to HAOS 13.0 completely broke all network integrations

Configuration
1

Is it me only or there is something wrong with HAOS 13.0? After update restart of GA took ~10 minutes (instead of usual ~40 secs) and I could see that startup process was hanging one by one on all cloud related integrations (Audi MMI, iCloud3, Netatmo, DNS IP - these are few I noticed). I had to restore entire config from backup to get it running. On the other hand network configuration in HA itself looked OK and I could connect to web UI without problems…Strange… Now I’m affraid to try to update HAOS again.

2

I just updated without issue. Generic64 install.

OS updates generally take longer than a HA restart.

For me:

Restart: 22 sec
OS update: 2 minutes 24 sec

3

No issues here - RPI5.

4

No problem here: HA Proxmox VM. Restart was a little longer.

5

No issues here on a Dell 3050

6

Sometimes it is the router that craps out on a HA update.
Try to restart that and see.

7

OK, yet need to nail it down, but it seems to be issue with of these:

  • I am running HA as VM on ESXi and neither VM virtual HW or version of operating system was updated post migration of hypervisor from 6.0U2 to 8.0. I made these updates and seems to be better now (NIC driver support?), but not ideal.
  • I do have 2 vNICs connected to my VM, one for user/IoT network and one for SAN. As I’m running recorder on MariaDB on my NAS, having 10Gb SAN connection to DB makes things way faster. But SAN network does not have connection to internet and this might cause issues with integrations trying to use this connection (not sure why).

Bottom line; now, after second attempt and some configuration changes it is way better, but yet not ideal… Will ivestigate further… though at the same time I got hit with some Zwave outage (some nodes become unavailable), so this is the priority…

8

If there is no internet connection then you have not defined a gateway, right?

9

DNS?

It’s always DNS.

1 Like
10

Something seems wrong here. The install and restart seemed OK earlier today but restarting the proxmox host now, it took several minutes to get the VM started again…so long I thought it hung. Checking the HAOS journal, it shows this repeating for pages. I can’t say this wasn’t there before but something is definitely hampering the start up performance. Once started, these messages stop and everything seems fine.

image
image889×271 50.7 KB

EDIT: The messages are continuing after checking again.

11

There’s an Arch issue open for that (first google result for “bad gso error”):

https://bugzilla.kernel.org/show_bug.cgi?id=219129

You should report this here: Issues · home-assistant/operating-system · GitHub

Also see: HAOS 13.0 Supervisor Problem · Issue #3524 · home-assistant/operating-system · GitHub

12

Done.

1 Like
13

Supervisor doesn’t show any errors but after 3 tries to update I’m still at 12.3
RPI4 with SSD

1 Like
14

There are many issues with OS 13.0 if running as a VM on proxmox and I assume VMware - very odd networking issues that look to be fixed in a dev version of 13.1 - See here Home Assistant OS 13.0 Breaks Hikvision and all Streaming Generic Cameras · Issue #3528 · home-assistant/operating-system · GitHub and the other mentioned posts

I had many odd issues with MQTT from another system, I think it has something to-do with kernel updates, to resolve I had to roll back to 12.4 by booting into the 2nd OS slot with HA CLI command ha os boot-slot B

15

Yup, I also restored 12.4 from backup and now everything is back to normal. Indeed issues with 13.0 seems severe, as now it is even pulled back from available updates.

16

Only for VM installs.

Screenshot 2024-08-17 at 19-13-35

2 Likes
17

I think I am in the same boat did a update 16 hours ago
and now dashboards not updating any of my smart switches data on the display
glad a took a proxmox snapshot
the backend data’s all their just the front ends not seeing anything

  • Core2024.8.2
  • Supervisor2024.08.0
  • Operating System13.1
  • Frontend20240809.0
18

As 13.1 is already available I made update and it seems to work properly for my VM. Yet need to revert to double NIC setup, with fast access to NAS running MariaDB… perhaps microproject for today.

Did you made your update from 12.x? If so, you might be affected by database schema conversion, that might take hours (depending speed of your host and size of DB). During conversion, as I understand from other posts, there is no updates of history, nor it is possible to draw charts with historical data. Perhaps this is issue that you are facing?

EDIT: Just reconfigured my HA network to my preferences (adding 10Gb connection to NAS running MariaDB) and happy to report that everything works as expected. So for me clearly 13.0 was the problem that is now fixed.

19

mine works now rolled back my database a day

so i guess there was a 12 ~ 13x conversion issue