Homey Pro (2019) brought to its knees?!

Well, that took u long Roy! And no aprils fool here.

A long long time ago, …

My best guess is that there are like 5 or 6 of them.

3 Likes

You speak of flows you like that need to be disabled to reduce cpu, but if used correctly flows should only take little CPU.

Are you by any chance using flows like “every xxx minutes” (or worse, seconds)? There are almost always better alternatives. Also, I believe the HUE app that uses the bridge uses a lot of polling, that may be part of the problem.

in other posts I see people furiously reducing the number of flows used, but often sacrificing performance in doing so. Reducing the number of flows shouldn’t be a goal in itself, keeping them simple and specific is way more relevant to keep Homey running smoothly.

The amount of devices and apps should not bring Homey to it’s knees, unless you have misbehaving ones.

3 Likes

Dear Edwin,

Thank you for your answer. At this moment, I’ve got the following in use:
-Aeotec (with repeater and 2 duty switches)
-Afval herinneringen
-Aqara @ Ziaomi (20 door sensors and 2 temp sensors)
-Better logic
-Bosch siemens (1 dryer)
-Count down (with a lot of timers)
-Denon AVR (1 AVR)
-Fibaro (3 double switches and 4 motion sensors)
-Harmony Hub (1 hub)
-Heimdall
-Homeykit (nearly all devices connected to homekit and google home)
-iCal callender (2 ical)
-INNR (17 SP120)
-Neato botvac (3 robot cleaners)
-Neo Coolcam (5 water sensors and 3 sirens)
-Openweathermap
-Osram (2 wall plugs outdoor)
-Outdoor air quality
-Philips hue (4 bridges connected. around 85 bulbs, 15 switches and 20 motion sensors in total)
-Philips Tv (1)
-Picnic
-Plugwise smile p1
-Popp (7 wall plugs)
-Power by the hour
-Public holidays
-Sonos (4 speakers)
-Sure pet care (1)
-Tado (11 zones and 1 AC controller)
-Virtuele apparaten
-Xiaomi mi flora (5)
-Ziggo (1)
-Zonnepanelen (1)

I’ve got around 225 flows. I think that 95% is ’ smart’ . Triggered by motion, open door, changing temp, etc. Just around 5 are triggered every 10m. (this is because homey is not always initiating all flows. It is an backup. Here a picture of my ’ normal’. load.

If you have any suggestions, would be really helpful!

I am more on like 30% average and I have a lot of the same stuff: aquara, innr, harmony, siemens, neato, philips, tado, … but way more Fibaro now, way less innr (which are misbehaving at startup a lot for me, and two wall plus just broke down). I also have no Philips hubs connnected, al zigbee is direct, most are innrs and sensors. I have no Denon app (yet, I do own a receiver), no ical, afval, ziggo or picnic.

I do not use Homeykit but have tried the new Athom experiment, just don’t really use it a lot. I use Chronograph instead of the countdown timers but that shouldn’t differ a lot. Are there always running a lot of timers, or only when needed? It shouldn’t matter much as far as I can tell though.

I’d be suspecting some apps to be polling frequently: Hue bridges (due to other forum posts), Picnic, Ziggo, ICal, Neato app. I disabled the last one because it seems to block the real Neato app from access (I use ifttt to trigger the bot when needed). To me that is also a sign it may be going to their server a lot, and I own only one bot.

Can you see from the insights graph per app if e.g. any of those apps is taking a lot of CPU? I know the graphs are inconvenient to say the least, but it’s pretty much all we have. If the graphs do’t make sense you could also check if the load significantly drops if you disable one of those apps?

And can you give some example of what payload is triggered every xxx minutes, and what it is fixing? Might be a clue in there if some flows don’t get triggered.

I have a Homey early 2016 and when running on firmware 5.0.4 the load was most of the time way over 100% and many times close or over 200%
After updating to Homey firmware 6.0.0 rc4 the load is back to normal and most of the time way UNDER 100% again.
I changed nothing in hardware devices, apps of flows, so the new firmware did the trick.

1 Like

Mine is down to 20% now, so there is improvement somewhere

Same feeling here, V6 seems to be a lot less hard on the Cores

Edit;

It seems to be more than just a feeling; this is the load over the last 31 days. You can see the drop in cpu load directly after the update;

2 Likes

I just to have a way to high load (with not that many devices and flows). Then i reconnected all zwave devices that were connected secure, now unsecure. After that, the load problems disappeared.