My Homey Pro isn’t relyable. I have 40 + apps and almost 200 flows. Now my Homey Pro is getting slow en unreachable sometimes.
I was searching and got an idea, maybe if I can replace the SOM ( system on a module). The Homey Pro has an C1000DL and can I simply replace the SOM with a https://www.solid-run.com/wiki/doku.php?id=products:imx6:microsom:quad . The SOM is connected to the board with a Hirose DF40 connector.
The thing is the software, how to get the homey’s software on the SOM.
Anyone got an idea? Or only replace the processor and the memory chip?
Or perhaps try and find out why your current Homey is getting slow? 40+ apps and 200 flows isn’t an awful lot (in fact, Athom advertises the Homey Pro as being able to run “over 40 apps”).
I had contact with Athom about this and they suggested to deinstall some apps. I told them I can’t I need the apps. And that was the support.
Enable the “Power User” experiment so you get CPU/RAM graphs in Insights, perhaps there’s a specific app that’s using a lot of CPU or memory. Also try and see if there are particular flows that may be causing the problems.
I have it enabled and here is a screenshot:
Here is a screenshot of Grafana:
But how can I see if a flow is causing problems?
“gemiddelde belasting” is high, around a 1.4 average, which means that there are always processes (= apps) waiting for the CPU. Try disabling apps one at a time to see if you can find a culprit, or look through the different apps’ CPU usage graphs to see if you can find an app that sticks out. I would start with apps that use Bluetooth, like the Beacon app.
Finding problematic flows works the same: try disabling flows (perhaps ones that are run often, or have a lot of actions) to see if the CPU usage drops.
here is the cpu load of every app. But no one sticks out
I don’t know if its save to put the log of homey here? I do have Realtime Homey logger V2 enabled.
Foscam and Prometheus use more CPU than other apps, although not shockingly much (but worthwhile to disable them and see if that improves performance). I don’t know Realtime Homey Logger so not sure if it’s safe or not.
If you have a lot of Z-Wave devices, search the forum to include those unsecure, I believe that a lot of people noticed a markable performance improvement after they did so.
@robertklep Can I send you a message so that not everyone sees my log file?
@Jeroen_Kokhuis Try adjusting the interval of prometheus as described here: Prometheus.io exporter
Makes an night and day difference for me
I’ve reinclude every zwave device insecure and adjust Prometheus interval and now my average_load is:
Still peaks, but the average is a lot lower.
it is more flattened
@robertklep and @matrover thanks for your help
Great findings here! Definitively something I need to do myself. How can you show the app cpu load in Insight?
Just took a few minutes. Now the apps shows in Insight. Let me see…
So obviously something is making my Homey get a really high load every 45 minutes 24/7. Any tips on how to figure out this one? I guess it might be a flow as none of my apps are using more than 0.00008 (vThermo app) of the cpu. Any way to show the flow in a detailed scheduled view?
I tried the Prometheus app once, because I wanted to use Grafana. It really slowed down my Homey. Couldn’t find a solution so I deleted Prometheus. Maybe you can disable the Promotheus app for a while to see if it makes any difference for you?
I dont have Promotheus installed. Just for fun, I stopped ALL my apps and still I get the peak every 45 minutes. That includes a reboot of Homey when all apps were disabled before rebooting. Still peaking. So that is interesting.
I sent a support case of it and created a diagnostic report to see what Athom can tell me.