Hello,
I am trying to identify, "rogue" or not attached disks in our ceph/proxmox cluster.
I run a few auditing type functions on our Ceph storage to find out which VMs are using the most space (as in used, not preallocated), and removing them.
I find that:
#rbd -p <pool name> du --merge-snapshots
seems to identify which disks are using the most space ("real" usage, not "preallocated" usage).
Usually, I'd look at the VMID in the disk name and find the associated VM that way. But for a few disks, I noticed they did not have a VM in the pool, suggesting a prior erase process went awry.
Is there a good way to audit the disks stored in a ceph pool to see if they're NOT associated ( attached, and "unused") with any VMs? We have about 1000 VMs in our cluster and I'd rather not check each by hand!
Thank you.
I am trying to identify, "rogue" or not attached disks in our ceph/proxmox cluster.
I run a few auditing type functions on our Ceph storage to find out which VMs are using the most space (as in used, not preallocated), and removing them.
I find that:
#rbd -p <pool name> du --merge-snapshots
seems to identify which disks are using the most space ("real" usage, not "preallocated" usage).
Usually, I'd look at the VMID in the disk name and find the associated VM that way. But for a few disks, I noticed they did not have a VM in the pool, suggesting a prior erase process went awry.
Is there a good way to audit the disks stored in a ceph pool to see if they're NOT associated ( attached, and "unused") with any VMs? We have about 1000 VMs in our cluster and I'd rather not check each by hand!
Thank you.