I have been noticing high memory usage for about a day or so. Tried to disable automations, add-ons, to remove almost everything from my configuration but the memory calculation as per the system monitor sensor is still high ~ 75%.
I am running Home Assistant 0.107.5 (latest) in Docker on Open Media Vault distribution and RPI4 4GB memory hardware. The strange thing is that when I ssh in OMV and check the memory, used by the containers, homeassistant uses about 80MB which seems ok. I am attaching some photos with the output of the top command and from the HomeAssistant UI:
I really don’t see what’s eating up the memory (if the 75% calculation of the System Monitor is correct as I am starting to doubt it…). Would appreciate if you guys can give me tips on what I can check in order to spot the culprit.
For my PI3b+ … HA is unsuable after a couple of hours… Got less than 80Mb of free memory… before I had 300… It’s so unstable that I had to turn off the 3B+… and I will kill soon my microSD. There is a huge memory leak problem IMHO with last releases
I have also faced this, luckily I run mine from a VMware esxi server it had 8gb memory and maxes out Nearly all the time. This has only been since the upgrade. It’s not a major issue for me as I just changed the memory to 16gb to solve the issue but it used to be minimal memory now it’s crazy since the upgrade.
Yep, As of 106 I believe the memory has been crazy.
I used to sit at less then 1gig used, now its always 3+ and if i leave it open all day, it will get to 100% used and swaping almost 1g as well. So its leaking on the client side as well.
Only thing open, 1 instance of chrome with ha loaded in it.
For clairifcations purpose: the first picture is from one of my desktops, all that its used for is editing and ha related stuff.
Second and third is my Ha server.
Same issue here.
I see peaks in the memory usage every 6 hours, sometimes eating all the available memory and blocking the Pi3b.
I have no automations that run every 6 hours so I don’t know how to find this thing.
I hope someone can tell me how I can solve this.
I also have this, which is causing multiple restarts per day as RAM utilization hits the limit for the HASS container, it swaps until CPU gets too high and then gets restarted. Unfortunately I can’t pinpoint when it happened as my hass box had to be restored from an older backup recently and I immediately upgraded it to 1.107.7, but I’ve installed 1.108.4 now to see if the situation has improved.
Just wanted to chime in and note I’ve had constant issues with HA recently. I only noticed it after updating to 0.107 or 0.108 I feel. I had to upgrade out of 107 due to an integration being broken and fixed in 108, and it’s been chuuuuuggging for me both in the GUI and automations. Light automations can be delayed 20+ seconds.
Running rpi3b+, with my main db offloaded to my nas. influxdb is running onboard still.
I am also seeing elevated RAM consumption and CPU usage. I’m on a RPI 3B, running Core .108.5 and the latest versions of the Supervisor and operating system. I noticed the increase when I want from .106 to .108.
Typically, my CPU usage was in the 4-8% range, and now it’s more like the 8-15% range. Free memory starts at about 470 MB after a restart and slowly declines until it seems to stabilize at between 360-380 MB free. Performance doesn’t seem to be impacted and that’s certainly still a reasonably low level of typical CPU utilization and enough free memory, but the sudden jump is still concerning to me.
Yes, I am… this seems to be an issue that other posts are mentioning started occurring after 0.107.7 The resulting impact means that I’m having to reboot HA every 20 hours in order to recover what appears to be retained memory. I plan on rebuilding using 0.109.3 today to see if the issue persists, however as a point of reference I’m currently at 80% usage.
Updated to 0.109.3… Forgot to mention that there are no errors being logged aside from Rest API calls that are not receiving valid responses. The memory usage drops are HA reboots, but as can be seen, memory usage increases with time, I’ll monitor over the next 24hours to see if the RPi 3+ is either locking up or non responsive and reseting…
I am also seeing a massive increase in resource utilization. I am running my system as a VM in Proxmox I have disabled several of the add-ons to see if it is any of them, but I continue to see a creep up of Memory usage and a pretty high base CPU load. The unfortunate part of this is that it keeps that core of my CPU boosting and is burning power for seemingly no reason.
Hello !
I’ve also noticed a big increase in CPU and RAM usage since a couple of release.
I’m running Hassio in a debian VM on a Proxmox host (i7-8550 and 32Go DDR4 2666MHz). At the same time, the frontend is starting to be “heavy” and not as responsive as it was in the past: sometime it’s more than 2-3 second to change lovelace page, go to the supervisor or the configuration tab.
So I’ve upgraded my VM from 1 core with 2Go of RAM to 2 cores with 4Go of RAM but the issue persist and the memory usage is following the hardware upgrades. So it’s not really a solution and don’t want to waste 2 more Go on my server…
I have some suspicions that InfluxDB is what’s causing these problems. I’ve a few times upgraded to the next minor release and Influx hasn’t autostarted, and memory is under 50 right now, compared to where it was the past few days and last night when I took this screenshot.
I have no idea if this issue is even on anyone’s radar to get resolved, Influx or otherwise. It’s super frustrating.
Any tips on debugging source of memory issues in HA? I am seeing HA use up all memory and pretty much hang system. I am using a RPI 3b+ with HA running in docker. My experience is that the memory usage seems to shoot up when the UI is requested, but a slow and steady leak even if you reboot and never launch UI. I was a bit slow to the upgrades as the graph below shows, but problem seems to have started with my upgrade to 100.2 and has gotten worse with each upgrade since. Latest release 109.6 is pretty much unusable. Going from not rebooting for months to every 12 or less hours is a real setback! Thanks for help!
I have the exact same problem. I already narrowed down the recorder to only record the automations and two sensors, but I see a clear increase of memory usage every hour during the night when not much is going on rather than one automation that determines the max day temperature every 15 minutes.
