Hello everyone,
I saw that there are several topics addressing a problem similar to mine, but didn’t find a real solution.
Basically i’m on a HP23 with the latest stable firmware available today installed (13.1.5).
My setup has approximately 100 zigbee devices (light relay, socket relay, thermostats, door sensors, motion sensors) all tuya devices, running with Tuya Zigbee app and some Reolink cameras, added with ONVIF app.
My homey runs about 50 advanced flows, most of them automating turning on and off lights with motion or on a time basis.
Then I created a couple of flows for realizing a sort of security system with Heimdall app.
As you can see from the picture, the problem is that I’m experiencing a slow but constant increase in memory usage by the homey process.
All apps seem to have a reasonable memory usage, buth the homey process arrived at 1,13 GB and it keeps going.
As a consequence, I notice that flows sometimes fail or are becoming slower and slower to execute.
The graph starts at february 2026, because I was forced to do a factory reset and restore from a cloud backup.
What can be causing this? And how can I free up some memory? I already tried restart through flow and PTP, with no much success.
Thank you in advance to any of you that can help.
That’s a lot 1,13 GB for the Homey process. There’s probably some kind of memory leak, you should contact Athom:
What can be a cause of memory leak?
You are saying that upon restart, the memory utilization is still at 1.13GB ? I don’t think so… still, I suggest implementing regular reboots, it might be depending on number of apps you might have, I don’t think that Athom will really focus on this - already tried to report to them multiple times, the answer was kind of “it’s normal”.
My top value is 1.6GB with ~220 hours of uptime (please note I have 8GB RAM/CM4 modded)
Athom support just replied. They suggest to temporarily disable ONVIF app, but looking at memory usage of that app, it doesn’t look like it to be root of the problem.
SEnt a diagnostic report to see if they can help more.
Just FYI
This is the “Homey” process RAM usage on a Pro 2026 (4 GB RAM), which also runs the Onvif app from the start:
I solely (re)boot it for creating offline backups, and it auto-reboots after firmware updates.
On a 2 GB RAM Pro 2023 the RAM usage shouldn’t exceed 600 MB in general (but some peaks are normal) imho.
I confirm that my homey , after a PTP of 10 minutes, at startup, has homey process increase in memory usage. This is the status this morning. System, of course, is slow and unresponsive.
Just for completeness of information, I have a couple of devices (motion sensors) included with Tuya Cloud app and many virtual devices (all simple ON/OFF switches), which are basically used in flows to include/exclude sensors from heimdall alarm modes and some others to make Alexa interact with some devices in homey.
The very bad thing is that, when i try to restart homey, it happens to bootloop with red pulsating ring light, probably due to very low amont of available RAM.
So the Homey process already uses over 1GB RAM right after a PtP?
I hope Athom can find anything in the system diags.
In case you don’t want to sit and wait:
What might help, and can do no harm, is to create a cloud backup, perform a factory reset, and then restore that cloud backup.
(This won’t work with offline backups (per USB Tool) btw, because that creates and restores a 1:1 copy of your system, including all “bad stuff”).
Yes, every time I restart homey, I keep fingers crossed to not see red ring … once it starts, memory usage is what you can see in the picture..and keeps going up every time.
I sent a diagnostic report to Athom, hoping they discover something.
The rising graph in my first post starts in February, back when I did a factory reset & (cloud backup) restore, for the same reason. In particular, if i remember right, it was after a fimware update. At the restart I had to factory reset again, because Homey got stuck in bootloop.
I don’t know if It could depend, in some way, on my advanced flows or zigbee devices. Many of them are Tuya relays with energy metering, reporting voltage, current and power.
This situation is becoming very frustrating.
UPDATE: just did a factory reset and restore from cloud backup. Homey keeps bootlooping with red pulsating ring and then reboots..This is not normal.
Best is to produce system debug and share the logs with Athom. 1.18GB after reboot is clearly not OKay
…ensure that you have same version of backup like the firmware, if it’s not on the same version, you might get bootloops.
I an starting to believe that 105 zigbee devices are a little too close to the limit for homey to handle.
Maybe this is part of the problem. Temporarily disabling Tuya Cloud and Tuya zigbee app and all flows that use those devices frees up some memory. Not much but but system free memory passes from 200-300 MB to 500 MB andh homey process falls below 1 GB.
Which is still too much, but this tells something.
If you have too many Tuya ZigBee devices, you can also use a Tuya Zigbee or MultiMode gateway and connect your devices via the official Tuya app or Tuya Cloud app.
I already use a Tuya Zigbee gateway to interface homey with Tuya Thermostats and motion sensors that are not recognized by Tuya Zigbee app
I have currently ~85 Zigbee, that’s not the issue
I’m having a conversation with Arno, from Athom support. He suggests to try to disable logging of all Tuya zigbee relays, which have power metering, because they could stress homey with a lot of data for insights (voltage.current, energy and power)
But how can I disable logging of those devices in insights? I can’t find a suitable option in insight section nor in developer tools.
You can delete Insights logs in the web app by right-clicking the logs and selecting Delete
But is deleting the same as disabling?
There is no option to disable logging in advanced settings. Only the possibility to set the minimum report interval, whose maximum value is 86400 seconds, which i set.
In insights, I have the possibility to delete data, download data or eliminate the whole parameter, which disappears from insights, but there is no possibility to restore it.
