[SOLVED] ZFS allocate the full disk size

[spearox]

Dear Members,

I have a freshly installed Proxmox 8 with two ZFS.

Current disk usage (btw its not valid usage):


I'll add a 100GB VM disk (full empty, just adding):


Current usage:


Why my ZFS work like this?

ZFS details:

root@int-node1:~# zpool get all
NAME    PROPERTY                       VALUE                          SOURCE
rpool2  size                           952G                           -
rpool2  capacity                       29%                            -
rpool2  altroot                        -                              default
rpool2  health                         ONLINE                         -
rpool2  guid                           17368552471780374021           -
rpool2  version                        -                              default
rpool2  bootfs                         -                              default
rpool2  delegation                     on                             default
rpool2  autoreplace                    off                            default
rpool2  cachefile                      -                              default
rpool2  failmode                       wait                           default
rpool2  listsnapshots                  off                            default
rpool2  autoexpand                     off                            default
rpool2  dedupratio                     1.00x                          -
rpool2  free                           672G                           -
rpool2  allocated                      280G                           -
rpool2  readonly                       off                            -
rpool2  ashift                         12                             local
rpool2  comment                        -                              default
rpool2  expandsize                     -                              -
rpool2  freeing                        0                              -
rpool2  fragmentation                  0%                             -
rpool2  leaked                         0                              -
rpool2  multihost                      off                            default
rpool2  checkpoint                     -                              -
rpool2  load_guid                      2149096279672757092            -
rpool2  autotrim                       off                            default
rpool2  compatibility                  off                            default
rpool2  feature@async_destroy          enabled                        local
rpool2  feature@empty_bpobj            active                         local
rpool2  feature@lz4_compress           active                         local
rpool2  feature@multi_vdev_crash_dump  enabled                        local
rpool2  feature@spacemap_histogram     active                         local
rpool2  feature@enabled_txg            active                         local
rpool2  feature@hole_birth             active                         local
rpool2  feature@extensible_dataset     active                         local
rpool2  feature@embedded_data          active                         local
rpool2  feature@bookmarks              enabled                        local
rpool2  feature@filesystem_limits      enabled                        local
rpool2  feature@large_blocks           enabled                        local
rpool2  feature@large_dnode            enabled                        local
rpool2  feature@sha512                 enabled                        local
rpool2  feature@skein                  enabled                        local
rpool2  feature@edonr                  enabled                        local
rpool2  feature@userobj_accounting     active                         local
rpool2  feature@encryption             enabled                        local
rpool2  feature@project_quota          active                         local
rpool2  feature@device_removal         enabled                        local
rpool2  feature@obsolete_counts        enabled                        local
rpool2  feature@zpool_checkpoint       enabled                        local
rpool2  feature@spacemap_v2            active                         local
rpool2  feature@allocation_classes     enabled                        local
rpool2  feature@resilver_defer         enabled                        local
rpool2  feature@bookmark_v2            enabled                        local
rpool2  feature@redaction_bookmarks    enabled                        local
rpool2  feature@redacted_datasets      enabled                        local
rpool2  feature@bookmark_written       enabled                        local
rpool2  feature@log_spacemap           active                         local
rpool2  feature@livelist               enabled                        local
rpool2  feature@device_rebuild         enabled                        local
rpool2  feature@zstd_compress          enabled                        local
rpool2  feature@draid                  enabled                        local
root@int-node1:~#

root@int-node1:~# zpool list
NAME     SIZE  ALLOC   FREE  CKPOINT  EXPANDSZ   FRAG    CAP  DEDUP    HEALTH  ALTROOT
rpool2   952G   280G   672G        -         -     0%    29%  1.00x    ONLINE  -
root@int-node1:~#

What can i do to enable thin provision?

 

[spearox]

I'll missed the thin provision button. '-.-

Now enabled.


But still the usage is not 280GB.

 

[LnxBil]

But still the usage is not 280GB.

That flag only applies to new datasets. You need to unset the refreservation flag on all datasets. Look at

zfs get refreservation -r rpool2

 

[spearox]

You are right. (ofc) They still thick.
I moved every VM disk to another then i moved one back. Current status:

This is a 100GB VM disk i don't know why it say 107.37GB.


It show 20.19GB used.


The zpool show less.
root@int-node1:~# zpool list NAME SIZE ALLOC FREE CKPOINT EXPANDSZ FRAG CAP DEDUP HEALTH ALTROOT rpool2 952G 18.8G 933G - - 0% 1% 1.00x ONLINE - root@int-node1:~#

Now its nearly the same but why it says 107 instead of 100 and why 20.19 instead of 18.8?

 

[LnxBil]

I moved every VM disk to another then i moved one back.

Yes, that works, too.

Now its nearly the same but why it says 107 instead of 100 and why 20.19 instead of 18.8?

Maybe run trim inside of your guest? This should free up more "free space". There is a technical difference between what you see as free and what really is free. Consider deleting a file: only the reference is deleted not its contents.

 

[Neobin]

Now its nearly the same but why it says 107 instead of 100 and why 20.19 instead of 18.8?

Gigabyte (GB) vs. Gibibyte (GiB)