2023.5.0 broke 12 different aqara Zigbee devices - still broken on 2023.5.3

I can’t figure out why it affected anything. This integration has been rock-solid-stable for I don’t even know how long now. Why they felt the need to mess with it at all is beyond me.

1 Like

5.3 appears to be worse for me than 5.2 was.

I re-paired missing stuff yesterday morning. Now this morning, different devices have fallen off. That never happened before, it was always the same devices affected for some reason (always aqara, but always the same handful of them).

sigh

Were you able to get your devices back online with 5.3, or is it still broken?

I reverted to 4.5 and everything has been rock solid, but I would like to upgrade at some point again…

After multiple re-pairings, they appear to be stable again. I’m not looking forward to any subsequent upgrades, ever again, since it seems like this is the new normal…

It might be a good idea if you have to re-pair to move to zigbee2mqtt, which decouples zigbee from HA updates.

Yeah, you mentioned that above in this thread. I certainly have not ruled it out, just too many irons in the proverbial fire at this point to add another to-do onto my list…

I was looking at this possibility, but it doesn’t look like the HUSBZB-1 is supported by z2m?

I have HUSBZB-1 and aqara devices that did not work pre 2023.5.3, but work again on 2023.5.3. Not sure if others still having problems have HUSBZB-1 or something else because the fix in 2023.5.3 seems to target HUSBZB-1 and aqara.

Just chiming in with my data point… ZHA/SkyConnect and HA 2023.5.3

My Aqara devices (motion and door sensors) don’t show offline, but some of them just don’t work. Sometimes re-adding helps, sometimes it doesn’t. Sometimes they start to work randomly by themselves.

VERY FRUSTRATING… never had any type of upgrade issue this serious.

Still not working correctly on 5.4

Oddly enough, this time it’s nearly ALL my water leak sensors that were knocked offline, with one or two other random devices here and there.

This is beyond frustrating, and even MORE frustrating is that the devs don’t appear to care.

Sorry to hear that… I finally updated to 5.4 a few days ago and didn’t have any issues this time.

Out of curiosity, what is the firmware version of your controller?

I’m sticking to 2023.4.6 until I can be reasonably sure I can control my lights after another upgrade.

Uncertain. If you know how I can check that, I’ll gladly let you know.

I have a large zigbee network, and just upgraded from 2023.4.5 to 2023.5.4 and it worked just fine with zigbee2mqtt, and the latest HAOS (I run it as a vmware guest) with z2m running on a remote raspberry pi.

I have a very complicated set of integrations, and I was worried about this upgrade, but things seem to have gone well.

If you go to the integrations and download the diagnostics for ZHA, you’ll see something like this in the json file.

"ezsp": {
            "manufacturer": "HubZ ZigBee",
            "board": "HUSBZB-1",
            "version": "5.4.1.0 build 962",
            "stack_version": 4,
            "can_write_custom_eui64": true
          }

Fairly sure the version is the firmware on the controller.

I’m fairly confident this only affects users using a HUSBZB-1 controller which is not supported by z2m.

Something changed with the built in zha integration that makes it unreliable now.

Mine is exactly as you show there:

"ezsp": {
            "manufacturer": "HubZ ZigBee",
            "board": "HUSBZB-1",
            "version": "5.4.1.0 build 962",
            "stack_version": 4,
            "can_write_custom_eui64": true
          }

Interesting. I would be curious ro see what others are. Especially those that are working again.

I’m on 6.7.8.0 and ZHA has been stable for me on 2023.5.4, though I only have two Aqara devices (TVOC air monitors).

      "ezsp": {
        "manufacturer": "HubZ ZigBee",
        "board": "HUSBZB-1",
        "version": "6.7.8.0 build 373",
        "stack_version": 8,
        "can_write_custom_eui64": true
      }

You might want to consider upgrading the firmware as per one of the links that was posted earlier.

Two devices is an exceptionally small sample size to say that the code is error-free. I have dozens of Aqara devices, and the issue does not affect all of them. Further, the issue did not exist prior to 5.0, and there were ZERO listings about breaking changes.

Devs need to find a solution for this.