I’ve got Home Assistant installed in Proxmox using the HassOS image, but I’ve noticed that my local hard disk consumption continues to grow rapidly in Proxmox. My HA instance believes it is only consuming 30% of the available space, but the actual allocation in the local-lvm storage went from 40GB to 52GB in 3 weeks.
I have no idea what is consuming this space as there is nothing else running on this box other than Proxmox and the HassOS image. What could be causing this?
And more importantly, what can I do to prevent it from filling up? I had the SSD filled accidentally previously when I was running a Windows VM as well, but I’ve removed that and nothing new is active or configured on this device.
It’s 2.5GB, but I’ve switched to MySQL in Docker on my NAS. I thought that might of been the problem first, but the HA instance shows 70% of free space available (60GB allocated).
same here. I could really use a pointer on how to clear the space or how to automate that process.
I’m wondering if it’s because I’m using an SSD and it’s not automatically trimming or something…
My HA instance previously became unresponsive due to full disk, which I determined was due to Proxmox snapshots filling the space. I’ve disabled that now, so I’m not sure how I will fix once it does fill up local-lvm the next time…
So there is a big discrepancy between what HA is thinking it is using and what proxmox thinks that it is using. I’ve set up trim on both the HassOS and proxmox, but it does not solve the problem.
Faced a similar issue. Proxmox, Home Assistant VM (it’s vm100 for me), and Ubuntu Server (vm101) are on the same SSD. My local-lvm storage went to 100% usage, and I got an I/O error on the HA VM. It was stuck, and I couldn’t restart it from the Proxmox GUI.
Here are some commands that you can use to unlock the VM:
This should allow you to boot the HA VM. However, if you stop here, your local-lvm storage will eventually fill up again. The problem is somehow connected to the TRIM mechanism, which deals with deletion/garbage collection on SSDs. Normally, TRIM is launched once a week with trim.timer/trim.service, but something is broken for me. So I launched TRIM manually. Go to VM → Shell:
# HA OS:
login
fstrim -av
# Ubuntu Server - fstrim -av does nothing so run:
systemctl start fstrim.service
# when fstrim.service finishes:
systemctl stop fstrim.service
Executing these commands took some time, but once finished, the occupied space on local-lvm went from 99% (remember, we added 1 GB with lvresize) to 49%. I still don’t know how to fix TRIM automation, but at least I have more time now.