PVE with Ceph: Datacenter Summary shows bad storage usage/percentage

B

broth-itk

New Member
Dec 30, 2024
17
6
3
Dear Proxmox Team,

please look carefully at following screenshots from my cluster (8.4.1, all updates, enterprise repo):

Datacenter -> Summary

1751298684951.png

... and Datacenter -> Ceph


1751298712626.png



As you can see, the values are different but yet the same.
The first screenshot does not take into account the 3x replication from the Ceph pool.

I'd suggest two solutions to this issue:

1. Quick and not so dirty fix

Use the values from Ceph in Datacenter -> Summary in calculation.
Doing so, the real storage allocation is shown.

2. Improved fix

Like 1 but show both values, example

14.06 TiB of 41.92 TiB raw capacity
(4.70 TiB user data)


I'm aware that the numbers include other storage pools as well like rpool (zfs mirror).
Nevertheless the Datacenter -> Summary page is misleading since in reality way more storage is used.
By no means we have more than 40TiB available ;-)

What do you think?

Best regards,
Bernhard
 
Last edited:
On our system Datacenter > Summary > Storage is larger (more space used and more total space). Since the difference is 7 TiB that's not directly attributable to boot drive space that I see at first glance, unless maybe it's counting free space on both of the ZFS mirrored disks, on all servers...? And maybe something else?

Can I ask, how is your Ceph replication set?

I see different measurements in different places, for example node > (Ceph storage name) shows approx. 31% (not 33%) of the node> Ceph total which is presumably related to 3:2 replication, and maybe DB/WAL since those are counted as used space if they are not on the OSD disk.

But overall the measurements do seem inconsistent.
 
You must log in or register to reply here.
Top Bottom