SSD / Dataset mysteriously full?

Newb question:
I have 2 datasets on an SSD, one for ixapplications and the other for data whcih is currently empty.

Yesterday however it suddenly jumped from 1% to 73% full.
All I did was update photoprism, which is using a different ZFS for imports storage etc.

The ixapplications is only using 7Gb and the other dataset is empty.
How can I identify what is filling up my SSD?
I also deleted all the old snapshots, but these were small anyway,
and I deleted photoprism and the ixapplication… no difference.

I can’t see what’s using up all the space.


You can start with these to find out what is consuming (or “reporting”) the space:

zpool list -o name,size,alloc,free,bcloneused,bclonesaved,checkpoint SSD

zfs list -r -t filesystem,volume -o space SSD

Both of the graphics you posted show the space used by files etc. Files can be compressed, snapshotted, reserved, use block cloning all of which change how much disk space they only use.

The only way to tell what is actually used is to use zpool list as @winnielinnie has suggested.

Thanks, however under shell in Truenas Scale, when I use those commands, I get “zsh: command not found: zpool” and “zsh: command not found: zfs”

EDIT: Ok… so you need to type “SUDO” that’s what was missing

What was the output?

The offender is
SSD/ix-apps/docker 238G 648G

The problem is, ix-apps is a hidden folder on the SSD.
I was playing around with Portainer, but then deleted it. I think it’s residual files left over.

Do you know how I can delete this hidden docker folder?
It’s 648G of wasted space.

EDIT:

Getting somewhere now.
Electric eel switched ix-applications over to a hidden “ix-apps”

Access via
cd /mnt/.ix-apps
Now I just need to work out what specifically chewed up my space

Not a folder, a dataset, but I know what you mean. Good idea to be careful with ZFS terminology, especially when it comes to deleting things.


It does not show on the Pools / Datasets page in SCALE?


With the command-line, first create a checkpoint:

zpool checkpoint SSD

Check that it exists:

zpool get checkpoint

zpool status SSD | grep checkpoint

If the checkpoint exists, you can now destroy the dataset:

zfs destroy SSD/ix-apps/docker

If you messed up, you can export the pool and rewind to the checkpoint. (“Messed up” can be anything from typing in the wrong command, or deleting the wrong target, or having an immediate regret or change of heart.)

If you’re happy with the result, you can (and should) discard the checkpoint:

zpool checkpoint -d SSD
1 Like

Thanks for your help @winnielinnie In the end, I rolled back and also deleted the entire docker and reinstalled my apps.

The thing that confused me was I used to have a dataset “ix-applications” but after the upgrade they migrated everything into the hidden ix-apps folder. Usually my data for apps in a main ZFS dataset, but I think portainer had other ideas,

I’m pretty sure it was Portainer and deleting that didn’t remove all the data associated with it (even when selecting delete ix-application data).

Anyway, mystery data is gone. Thanks.