My backup strategy with Proxmox Backup Server

Updated at PBS version 4.0.14 with S3 backup still in beta

3‑2‑1 Backup Rule: How Have I Implemented It?

• 3 copies of my data: Proxmox VE, NAS QNAP, EXTERNAL HDD
• 2 different type of media: nothing fancy, no cloud just different physical disc (both the QNAP and the EXTERNAL HDD have WD RED drives but are separated from each other)
• 1 copy off-site: I bring the EXTERNAL HDD to my workplace once a week and keep it there

Addition protection:

• The NAS QNAP is on different vLAN (air-gapped) with strict firewall rules for access
• It is also powered off for almost half a day (from 18:00 to 11:00)

Backup Jobs

Daily

• VM: Important ones
• Mode: Snapshot
• Schedule: Everyday at 15:00
• Destination: pbs-nfs-qnap (Refer to the QNAP section)

Weekly

• VM: Important ones
• Mode: Snapshot
• Schedule: Sunday at 11:30
• Destination: pbs-ext-hdd (Refer to the EXTERNAL-HDD section)

QNAP

• NFS share mounted every day at 12:30 with command: mount -t nfs IP:/Bck-PVE /mnt/NAS (using crontab)
• The PBS datastore is nfs-qnap and point at /mnt/NAS
• Scheduled Prune Job everyday at 13:00 keeping 5 daily and 4 weekly backup
• Scheduled Verification Job every Saturday at 14:00

EXTERNAL-HDD

• The PBS datastore is ext-hdd and point at /mnt/datastore/ext-hdd
• Scheduled Prune Job every Sunday at 16:00 keeping the lastest 4 backup
• Scheduled Verification Job every Sunday at 14:30

Configuration:

In the Web UI under Administration > Storage/Disk > Directory I've created a directory named ext-hdd, selected xfs as filesystem and pointed it at the correct disk. It's also needed to check the option "Removable datastore".

With this every time the disk is plugged it's automatically mounted to the datastore.

The problem is that as of today the unmount need to be done manually. Refer to the official docs: https://pbs.proxmox.com/docs/storage.html#removable-datastores

I then created a crotab that execute the unmount script every Sunday at 16:45:

    crontab -e
    45 16 * * SUN /usr/local/bin/unmount_hdd_datastore.sh ext-hdd

Refer to the file unmount_hdd_datastore.sh on this repo for the code.

The script needs to executable:

    chmod +x /usr/loca/bin/unmount_hdd_datastore.sh

To test the script run:

    ./unmount_hdd_datastore.sh ext-hdd

With ext-hdd being the name of the datastore and an output similar to this should apper:

[2025-08-30 14:27:12] [info] New logging session
------------------------------------------------------------
[2025-08-30 14:27:12] [info] Checking presence of UUID for datastore 'ext-hdd' …
[2025-08-30 14:27:12] [info] UUID for 'ext-hdd' is present (disk attached).
[2025-08-30 14:27:12] [info] Attempting to unmount datastore 'ext-hdd' …
[2025-08-30 14:27:12] [info] Datastore 'ext-hdd' successfully unmounted.

The full log file is located at /var/log/pbs-datastore-unmount.log.

For cleaning purpose I've also configured logrotate:

nano /etc/logrotate.d/pbs-datastore-unmount
                                                            
/var/log/pbs-datastore-unmount {
    monthly
    rotate 4
    compress
    missingok
    notifempty
    create 0640 root adm
}