Proxmox Backup Server

Links

• PBS Homepage

Documentation

• PBS documentation
• PBS Technical overview
• Managing PBS remotes & sync

Installation

• Installing PBS

Upgrading

• Upgrading PBS 2 to 3

Processes

Service daemons

See Service Daemons

Command line tools

See Command-line Tools

Under the hood

Structure of datastore:

/mnt/datastore/yourDS

contains:

.chunks

The actual data

vm

Metadata per vm: client.log.blob drive-scsi0.img.fidx index.json.blob qemu-server.conf.blob

HOWTO

Move datastore

• Make sure there is no activity on that datastore, i suggest using rsync. As i can tell it should be safe to use rsync option --ignore-existing when updating a copy.
• Edit '/etc/proxmox-backup/datastore.cfg' to change path to datastore.
• Reload configuration (this will interrupt running garbage collectors) :

systemctl reload proxmox-backup-proxy

Upgrade proxmox 6 to 7

https://pve.proxmox.com/wiki/Upgrade_from_6.x_to_7.0

WARN: 'noout' flag not set - recommended to prevent rebalancing during cluster-wide upgrades

https://docs.ceph.com/en/quincy/rados/troubleshooting/troubleshooting-osd/

ceph osd set noout

Restore disk from PBS

Use pbs-restore

What's so busy?

proxmox-backup-manager list

and then

proxmox-backup-manager task log <everying from and including "UPID:">

PBS Backups

Create new backup repository

On PBS server:

Create account

Configuration->Access Control->User Management->Add

Create datastore

Create space, normally something like

zfs create pbspool/mystore

Datastore->Add Datastore

New subdirectory will be created

Backup path must be full path to storage + datastore name

Configuration->Access Control->Permissions
Datastore->yournewstore->Permissions

to give the user access, Role is usually DatastoreBackup

On PBS guest:

Datacenter->Storage->Add->Proxmox backup server

For "Datastore" enter the name as used on the PBS. For fingerprint (on pbs server):

proxmox-backup-manager cert info | grep Fingerprint

Then:

Backup->Add storage: your-pbs

Remove PBS datastore

proxmox-backup-manager datastore remove yourdatastore

this does not delete the data!

Move a PBS datastore

rsync -auv /mnt/datastore/old /mnt/datastore/new
systemctl stop proxmox-backup-proxy
rsync -auv /mnt/datastore/old /mnt/datastore/new
vi /etc/proxmox-backup/datastore.cfg
systemctl start proxmox-backup-proxy

and after checking everything you can

rm -vf /mnt/datastore/old

Move vm data to another datastore

Create local sync user

localsync@pbs

Permissions: datastorereader on source, datastorebackup on target

PBS API

• API Tokens
• PBS API viewer

PBS equivalent for pvesh:

proxmox-backup-debug 

example:

proxmox-backup-debug api get /status/datastore-usage

proxmox-backup-debug api get /nodes/localhost/tasks

Create cookie

PBS=your.pbs.server
curl --silent --insecure --data "username=zabbix@pbs&password=n0tmyp4ss" https://$PBS:8007/api2/json/access/ticket | jq --raw-output '.data.ticket' | sed 's/^/PBSAuthCookie=/' > cookie

Test cookie

curl -XGET --silent -k -b $(cat cookie ) https://$PBS:8007/api2/json/

Recovering files from backup

export PBS_REPOSITORY=mypbs:mydatastore
proxmox-backup-client snapshot list

Output like: vm/109/2022-01-25T20:18:02Z │ 200.00 GiB │ client.log drive-ide0.img index.json qemu-server.conf

# proxmox-backup-client map vm/109/2022-01-25T20:18:02Z drive-ide0.img

Image 'data2:vm/109/2022-01-25T20:18:02Z/drive-ide0.img' mapped on /dev/loop0

# lsblk

NAME         MAJ:MIN RM   SIZE RO TYPE MOUNTPOINT
loop0          7:0    0   200G  1 loop
├─loop0p1    259:0    0   251M  1 part
├─loop0p2    259:1    0   3.9G  1 part
└─loop0p3    259:2    0 195.9G  1 part

 

mount /dev/loop0p3 /mnt/loop/

mount: /mnt/loop: cannot mount /dev/loop0p3 read-only.

mount -o loop,ro,norecovery /dev/loop0p3 /mnt/loop/

and that should be it. After you're done:

umount /mnt/loop
proxmox-backup-client unmap /dev/loop0

 

TBD: explain how t o use kpartx and friends to acces LVM volumes

 

FAQ

Error messages

error fetching datastores - 401 Unauthorized (500)

Check username/password :)

Limit bandwidth

On cluster node

In /etc/vzdump.conf

# In KB/s
bwlimit: 1500

Garbage collection

What does the Prune all button do?

It will give you a dialog with the usual 'Keep xxx' options.

TASK ERROR: update atime failed, disk full

Try pruning some Check if some sync is running

unable to access non-existent chunk

check if this returns?

Pending removals

proxmox-backup-manager garbage-collection status mystore

error pruning backups - check log (500)

Possibly a permissions thing, which is good. Leave pruning to the backup server

command error: http upgrade request timed out

??

Connection error - server offline?

This can also happen when you can actually connect to server:8007 Check

 systemctl status proxmox-backup

Connection error 401: No ticket

Browser issues, try clearing cache/cookies(?) or another browser.

Noout flag(s) set

ceph osd unset noout

qmp command 'query-backup' failed - got wrong command id

maybe bwlimit?

Garbage collection failed: unlinking chunk ... ENOENT: No such file or directory

Ignore?

Sync errror: owner check failed

Check in datastore/yourrepository/vm/*/owner

There's command

proxmox-backup-client change-owner

tcp connect error: deadline has elapsed

Probably connection timeout.

proxmox-backup-manager remote list not showing remote

Try

proxmox-backup-manager remote show

Check

/etc/proxmox-backup/remote.cfg

Retrieve PBS password

On a pve node check /etc/pve/priv/storage/

group lock failed: Permission denied

Probably a mismatch between path in /etc/proxmox-backup/datastore.cfg and actual location. You also need to restart proxmox-backup-proxy.service after changing either of those.