Disk Metrics - Used Space Defined

Can someone help me better understand what the number in the picture below represents? I have HA installed on a Proxmox VirtualBox VM with 50gb allocated for storage. My MariaDB is around 500mb. The Disk Metrics screen indicates I’m at 40% storage spaced used. What does this number represent? How can I get more details?

The metric is as it states that 39.8% of the allocated space to your HA install has been used by your integrations, automatons etc that you have setup and installed in the VM where the data is stored.

Thank you. How can I get more details about what is using space?

Other than looking at logs and your integrations etc to see what you don’t need I’m not sure if there is a more specific way to look into that at the moment, it may be something for them to code in at a future date if more people request it for an option in the UI.

Thanks. Running du -hd 1 / I get the following results. This show a total of 4gb used, but my VM is 50gb.

I have much the same issue. I installed HAOS into a VM with 32GB disk space, and have run out of space. After removing 3 Full Backups (1.2GB each) I am looking at my system and wondering where the rest of the space went.

In my case I have the Samba and SSH add-ons installed (though I don’t expect SSH to give meaningful values from within a Docker container). The disk is shown as 32.7GB used, leaving only 33.5MB free. But what is taking up that 32.7GB ?

  • the whole /config folder (including 470.7 MB database file, ESPHome, node-RED) claims to contain 3071 items totalling 526.7MB … but then says 32.7GB used.
  • the entire /backups folder now contains 10 items totalling 8.8GB.
  • /addon_configs contains 9399 items totalling 111.3 MB
  • /share contains 629 items totalling 374.1 MB
  • /addons and /ssl are both empty
  • /media contains a massive 29,482 items … but actually is 4 local files totalling 1.5GB plus a mapped network drive - so not relevant here.

SO… HA’s data files (including backups) comprises only a total of 11.3 of the 32GB.

  • DataBase is only 470MB, so a Recorder:Purge and packing the DB won’t gain much.
  • I could transfer all my backups to a mapped network drive, but that gains only 8GB, and how long before that 8GB is also eaten up ?
    • Curiously the size of a Full Backup has tripled from 1.2GB last month to 3.3GB today ! In my latest Full backup file largest file is core_whisper.tar.gz at 2.8GB.
    • The Full backup seems to contain separate .tar.gz files for HA core and each of the add-ons. Largest are Rhasspy at 696MB, share 277MB, Homeassistant 197MB and nodered 27.4MB. Really not particularly big files.

      Are these the full Docker containers, or just the relevant data files ?
    • Recently more Whisper models have been made available, and (thinking I have plenty of processing resource) I changed to the bigger (and supposedly more accurate) “medium.en” model. Well, that explains the increase in size of Full Backup. Previous core_whisper .tar.gz was only 941kB.
  • There are methods for reducing the amount being logged by “logger” settings in configuration.yaml
logger:
  # The next line sets the default logging level, for all python modules.
  # It seems "recommended" to set it to avoid too much logging.
  default: warning
  • and also for other modules (such as ZHA
logger:
  logs:
    custom_components.zha_toolkit: debug

But honestly, trimming HA’s data files isn’t going to make the sort of difference I am looking for.


What is the history of my free disk space ?


and looking closer at the last month

Not much clue there , free disk space seems fairly static for several months, then jumps suddenly - maybe when a major upgrade or new addon was installed. I have been trying HA Voice Assist, and been re-focussing on some ESPHome devices … maybe that accounts for it.


HAOS uses a cut-down version of linux, with Docker and the various HA programs

Logging in to the console and running fdisk


Apart from the duplicated (small) kernel and system partitions, nothing surprising here.

Some forum threads suggest issues with Docker. I personally have chosen to avoid the docker learning curve, but I have picked up that HAOS uses Docker behind the scenes to run HA’s Addons in their own separate environments. Is the benefit of apps running independently, at the expense of higher RAM and disk used to store duplicates of the prerequisites ?

Some threads suggest Docker commands to remove extra files - but I didn’t notice any change.


What about generic linux system files ?

  • my system log built to 535K lines 499MB in 9 days - includes debug statements from ZHA, zigpy and Rhasspy … yet none of the HAOS logs (including Host) go back further than the last reboot (2 days). System log can be trimmed with “journalctl --vacuum-size=200M” or “journalctl --vacuum-time=3d”
  • zigpy debug statements added a lot of lines to the journalctl
  • I found a 5.2GB swapfile in /mnt/data/ Given that I allocated 16GB RAM to this machine, should it even need to swap ? If I was using a SDcard, would the swapfile be killing the SDcard ?

I found also a series of folders containing the audio of Voice Commands i have given to Jarvis at
smb://192.168.1.98/share/assist_pipeline/5211d8ced7ea109af5597d3f71f4fc8e/Jarvis%20local%20Voice%20Assist/168545720414634 contains 01_stt-stt.faster_whisper.wav

I deleted quite a few of the older folders/files, but doesn’t seem to have made much difference as each file is only around 100kB. I recall when setting it all up that there was an option to save these recordings … but now I can’t find it to turn off. Maybe that was with the previous homeassistant-satelite ?

Is there any news about your problem? Does it still exist? What was the solution?

Not sure who you are asking … but yes, disk space usage is still tight - though I now archive and delete backups after a month.

The biggest single user is Whisper. Its documentation says

Backups

Whisper model files can be quite large, so they are automatically excluded from backups.

but that is not true, as reported in this github issue.