Many thanks and some ideas.

[endurance1968]

Just found your script as I was struggling to analyse where all my storage is gone....

using your script results in e.g. for one VM:
[DIR] Path to datastore: /media/backup
[CHUNK] Chunk path: /media/backup/.chunks
[DIR] Search path: /media/backup/vm/109
[THREADS] Threads: 12

[SAVE] Saving all used chunks
[INDEX] Index 28/28 (100.00%) | [TIME] 0m 00s
[SUM] Summing up chunks
[CHUNK] Chunk 71468/71468 (100.00%) | [TOTAL] Size so far: 226.3GiB | [TIME] 0m 02s
[TOTAL] Total size: 243004268683 Bytes (226.3GiB)
[TIME] Evaluation duration: 0m 02s
[DETAIL] Chunk usage summary:
Unique chunks : 71468 10.34% | 226.3GiB
Duplicate refs : 619510 89.66% | 2.1TiB
Total references : 690978 | 2.4TiB

I build some script around to get a better overview - now it looks like (focus on the unique part):

`
NAMESPACE TYPE ID NAME Unique% UniqueSize

---

root vm 116 win11-office 42.49% 211.4GiB
...
root vm 109 lx-ncloud 10.34% 226.3GiB
...
PVE-Hosts host - pve-config-pve-hv-07 100.00% 2.2MiB
`
hmm - format of table is lost.. so a tiny screenshot

means it prints out VMs, CTS and hosts sorted based on uniqe size, % or weighted unique % * size as table. So you can easliy identify which VM etc. uses potentially the space on disk.

Which at least for me summerizes it more intuitivly also displaying the VM name (for this a restored temporarily stuff from PBS and extracted the name - found no other working way).