zfs rpool/data via commandline / pvesm status fail

[tirili]

Looking around the installation, we find references to rpool/data. But creating the pool using commandline causes an error....

I did

/sbin/zpool create -f -o ashift=12 rpool mirror /dev/sda4 /dev/sdb4
pvesm add zfspool pve$(hostname -s)-data -pool rpool/data
pvesm set pve$(hostname -s)-data -sparse
pvesm set pve$(hostname -s)-data -content images,rootdir

zpool list gives some info:

# zpool list
NAME SIZE ALLOC FREE EXPANDSZ FRAG CAP DEDUP HEALTH ALTROOT
rpool 3.56T 456K 3.56T - 0% 0% 1.00x ONLINE -

# cat /etc/pve/storage.cfg
dir: local
path /var/lib/vz
content rootdir,iso,vztmpl,images
maxfiles 0

zfspool: pvex0720-data
pool rpool/data
content rootdir,images
sparse 1


but pvesm status fails:

# pvesm status
zfs error: cannot open 'rpool/data': dataset does not exist
Name Type Status Total Used Available %
local dir active 1998672 252060 1625372 12.61%
pvex0720-data zfspool inactive 0 0 0 0.00%

How can I get it active?

 

[dcsapak]

has the rpool a 'data' dataset?
what does zfs list say?

 

[tirili]

Hello Dominik, rpool does not have a data dataset. It was a clean install over debian, and all commands are provided in my post.
I have access to the system this afternoon, then I will provide the zfs list output.
Did I miss a command while creation?

 

[dcsapak]

if you want to use the data dataset (as in your config -> 'rpool/data') it must exist
you can create it with
zfs create rpool/data

 

[tirili]

# zfs list
NAME USED AVAIL REFER MOUNTPOINT
rpool 348K 3.45T 96K /rpool

 

[tirili]

Now I have:

# zfs list
NAME USED AVAIL REFER MOUNTPOINT
rpool 348K 3.45T 96K /rpool
# zfs create rpool/data
# zfs list
NAME USED AVAIL REFER MOUNTPOINT
rpool 456K 3.45T 96K /rpool
rpool/data 96K 3.45T 96K /rpool/data

# pvesm status
Name Type Status Total Used Available %
local dir active 1998672 257988 1619444 12.91%
pve-data zfspool active 3705667224 96 3705667128 0.00%


Thank you so much.
Which is the reason/advantage to work with subvolumes like rpool/data, and not directly in rpool ?

 

[dcsapak]

Which is the reason/advantage to work with subvolumes like rpool/data, and not directly in rpool ?

you can set features/properties of the dataset independent of the rpool (e.g., size limits)