SOLVED I dun goofed - Recovering space from root dataset

[Akshunhiro]

Hi all, doing a storage shuffle and now trying to move stuff back.

Created a new pool and started copying stuff over before realising it was going into the root dataset and not the child.

In an annoying spot now with excess wear on the pretty much day-old drives but now stuck on how I can recover the space apparently used by the dataset.

I want to put everything into "chimera/data" but there's still 20T (and climbing) referenced in the parent dataset.

Code:

zfs list -t all -o space
NAME          AVAIL   USED  USEDSNAP  USEDDS  USEDREFRESERV  USEDCHILD
chimera       23.1T  20.3T      272K   15.0T             0B      5.40T
chimera@test      -   272K         -       -              -          -
chimera/data  23.1T  5.40T        0B   5.40T             0B         0B

Code:

zfs list
NAME           USED  AVAIL     REFER  MOUNTPOINT
chimera       20.4T  23.1T     15.0T  /mnt/chimera
chimera/data  5.43T  23.1T     5.43T  /mnt/chimera/data

Code:

zfs get copies
NAME          PROPERTY  VALUE   SOURCE
chimera       copies    1       default
chimera@test  copies    -       -
chimera/data  copies    1       local

EDIT: output of list -t filesystem

Code:

zfs list -t filesystem -o space
NAME          AVAIL   USED  USEDSNAP  USEDDS  USEDREFRESERV  USEDCHILD
chimera       22.9T  20.6T        0B   15.0T             0B      5.60T
chimera/data  22.9T  5.60T        0B   5.60T             0B         0B

How do I purge the 15.0T REFER/USEDDS - Been Googling for ages but not finding much on this specific problem (except don't copy data to the root dataset haha)

 

[danb35]

Snapshots?

 

[Akshunhiro]

Negatory, chimera@test is the only one

 

[Akshunhiro]

Yep, definitely have teh dumb

Purging some bigger folders with rm -r is now reducing the usage for the parent dataset

 

[joeschmuck]

How (what command) did you use to start moving the dataset data to "/mnt/chimera" ?
Are you "copying" or "moving" the data? There is a difference. "moving" can lead to a more complicated solution, "copying" is an easy solution.

Purging some bigger folders with rm -r is now reducing the usage for the parent dataset

So you turned on recursive. This will also delete data in "/mnt/chimera/data" if I understand my "-r" correctly. Of course I'm making an assumption, I'm not there to see exactly what you are doing.

Best of luck.

 

[Akshunhiro]

Gonna level with you, I'm using unRAID. I used zfs send | zfs recv for migrating pools on TrueNAS recently (that was a lesson in itself especially with encryption) but using rsync on unRAID after creating a raidz2 pool.

I then messed with the mount points and symlinks which doubled-up my data dataset (when I thought I was copying to chimera/data, I was copying into a folder called data on chimera).

Resetting the mount point and renaming the data folder allowed me to see everything.

I was then doing a move from the parent to the child (probably just quicker copying from the array with rsync again though).