datastore is 100% full - help wanted!

chudak

Renowned Member
May 11, 2019
350
25
68
Hi all,

My PBS datastore (PBS_ZFS) is 100% full.
I manually removed a bunch of chunk files from .chunks/, but df -h still shows no free space and also via GUI.

I also tried to run prune/garbage-collection, but prune jobs could not be created because the datastore is full.

Questions:

Why does deleting chunks directly not free space?

What’s the correct way to reclaim space in this situation?

Thanks for any guidance.
 
Last edited:
Will this work?

# 0) Stop PBS
systemctl stop proxmox-backup proxmox-backup-proxy

# 1) Remove datastore entry from PBS
proxmox-backup-manager datastore remove PBS_ZFS

# 2) Destroy the dataset (wipes all backups), then recreate it
zfs destroy -r PBS_ZFS
zfs create PBS_ZFS

# 3) Start PBS back up
systemctl start proxmox-backup proxmox-backup-proxy

# 4) Re-add datastore to PBS
proxmox-backup-manager datastore create PBS_ZFS /mnt/datastore/PBS_ZFS

# 5) Verify
proxmox-backup-manager datastore list
df -h /mnt/datastore/PBS_ZFS

@Chris could you pls help?
 
Last edited:
My AI took me to rabbit hole but simple google search said

"To resolve a full Proxmox Backup Server (PBS) datastore, you must perform garbage collection (GC) to reclaim space used by deleted backups. You may need to temporarily move some .chunks files to another drive to free space, run pruning and GC, and then return the files"

and seems to work
Odd ....