Incremental rising memory consumption over time

I see RAM consumption slowly rise over time, as shown below:

Is this normal for HA, or perhaps a memory leak? For the latter, this forum suggests that Add-Ons are the first place to check. The sole Add-On I have currently is the ‘Advanced SSH & Web Terminal’, which shows “Add-on RAM Usage 0.4 %”. That seems not to be a problem.

I’m running on a RPI3B (1 GB RAM); I’ve attempted to upgrade to RPi4xx but haven’t had any luck booting HA on that.

Where else might I look?

What is the scale?
Either way, it could be a mis-behaving integration. See if this tells you anything.

Thanks @Sir_Goodenough for the fast reply. To answer you, the scale from the image in my post is visible on the left axis, roughly 66%-76% RAM usage.

If I cycle HA, RAM returns to ~66% and then slowly crawls upwards over several days as shown.

Thanks for the debug link, I’ll explore that, but I should emphasize that I haven’t seen any instability or crashing, instead simply slow incremental RAM consumption.

Constant 75% ram usage without something like frigate or plex or kodi transcoding things is high.

Apologies for my error – I corrected my post above to note increasing RAM, not CPU as I spuriously put in one line. CPU is <10% at all times.

same answer

How much RAM do you have? That would determine whether that constant percent is high.

I’ve ran into custom integrations having memory leaks in the past. But I also found that a lot of my issues went away when I simply upgraded to a device with more RAM than the 1GB that my Pi 3B+ had (while increasing swap size helped, it still seemed to struggle). Unless you have a pretty simple setup, I think more than that is really required.

RAM is there to be used. Used memory can improve performance, unused memory doesn’t do anything until it’s used. I wouldn’t worry about it if you don’t see any issues.

I checked the usage on my HA, and I can say it also steadily rises over time, until it suddenly goes down. Overall it’s usually between 40 and 50%, but it peaked at 65% a few days ago, just to go back to 45% afterwards. This is with 2 GB and a rather small HA system, running on proxmox.

% utilization doesn’t mean much without knowing how much memory your system has.

I have an old RPi 3B+ with 1G, and I saw almost exactly the same thing. Started around HA Core 2024.4.x or so. Before that I sat at 50% virtually all the time. I’ve updated to 2024.5.x and 2024.6.x since then. It still pushes 80%, instead of the previous 50%, but at least now it stays there, and doesn’t keep slowly increasing until the system crashes like it did originally.

One of the things you can do is install the “Glances” add-on. That way you can see which container shows the increasing memory utilization. In my case it was “homeassistant,” so no way to blame an add-on. I just assume it was a bug, quietly fixed but in a less efficient way than before. Since most developers have moved far beyond the RPi these days, they probably don’t even know it happened.

Good to know @Fanful, your cyclic memory experience is helpful for context.

Thanks @CaptTom, ‘Glances’ is a helpful diagnostic.

For reference like you I’m using Core 2024.6.4. Also: Supervisor 2024.06.2, Operating System 12.4, Frontend 20240610.1.