Nous A1Z DDF causes high CPU

Installed several (10) Nous A1Z plugs over the last weeks. Homeassistant slowed down due to high CPU load. I noticed the Nous plugs flooded the deconz log with sensor readings. Editing the DDF solved the problem

Changed min interval and reportable change and save the new DDF in /data/.local/share/dresden-elektronik/deCONZ/devices which reduces the flood of sensor readings. Maybe this isn’t the best solution?

It can.
I use bigger value on my side.

Another user have do same here DDF: Add support for all current NEO smart plugs with metering (TS011F) by iisworks · Pull Request #6474 · dresden-elektronik/deconz-rest-plugin · GitHub

Will make a PR for that, added here DDF add support for some tuya device by Smanar · Pull Request #6244 · dresden-elektronik/deconz-rest-plugin · GitHub

I hoped I solved the problem but CPU load was stable for a day or so. Now it’s growing again…
Something is wrong with the DDF, The DDF show values for Voltage and Current. I don’t see the devices in deconz.

Yep there is an issue on last version that cause issue on CPU usage.

Deconz display node, so you have only 1 (it’s one device), even there is 4 or 5 entries in the API (1 by sensors)

Ok that would explain my (random) high CPU issue.

Regarding sensors. All my A1Z devices show only consumption and power sensors in homeassistant.
I have a few Nous Wifi plugs connected with the Tuya integration. They show power, current and voltage sensors. For these plugs consumption is not available in homeassistant (its available in the Tuya app)

Voltage is a attributes of one or those sensors. If i remember correctly power.

Thanks. It’s indeed an attribute, overlooked that.

Disconnected all A1Z plugs and rebooted several times. The result is a flat line in CPU load.
If I connect all 9 plugs and reboot the result is mostly an increasing CPU load which doubles overnight.
If I connect 1 or 2 plugs and reboot the result is sometimes increasing CPU loads.

CPU load of the deconz container is normally (without A1Zs) around 0,5%, the more A1Z I connect the more it fluctuates up to 10 % and the overall CPU load keeps growing and growing.

No idea how this is possible and tweaking with DDF setting doesn’t solve this, it only improves the success rate of a stable system after reboot.

It probaby has to do with the writes to the database. Manup mentioned this yesterday:

Spoiler alert, constant database writes may lead to high CPU usage.. most of them make no sense to refresh at all. Working on a fix

Today he mentioned:

 Update on the CPU/performance blues: My test setup with 140 DDF devices seems more relaxed now with a small database change.
I modified the DDF locally to provoke constant polling of the lights and therefore poking JS engine and database writes.
So far this change got the network from 20% CPU peak to 12%  after init phase.

So that means it prolly will be fixed later on.

Hopefully it gets fixed later. As a mitigation I installed ZHA and a no-name zigbee controller for the TS011F plugs next to deconz. CPU load is nice and quiet now. Deconz CPU is around 0.4% and stable without the TS011F’s

I hope this fix will also help for the Innr SP120 smart plugs. I’ve always seen the

7:35:03:415 ZCL attribute report 0xA4C138B79BDE4D87 for cluster: 0x0B04, ep: 0x01, frame control: 0x18, mfcode: 0x0000

style messages popping up every second per plug.

I don’t found DDF for the sp120, so its the same defaut reporting from the first time this device was added.