Hi,
In the web-interface, for any VM/CT you have a entry with name "Replication"!
Good luck!
Best practice for Proxmox self-backup
- Thread starter speedbird
- Start date
-
- Tags
- backup proxmox host
Hi,
In the web-interface, for any VM/CT you have a entry with name "Replication"!
Good luck!
Hello
Thank for the script
I try to backup the server with it .
Then I try to restore to same machine but with fresh install (same version of proxmox) but pve-cluster refuses to start
I receive pmxcfs error 255
Tried to run pmxcfs I receive :
fuse: mountpoint is not empty
fuse: if you are sure it is safe , use the 'nonempty' mount option
This happenes (if i understood well) because /etc/pve is not empty (probably as it has my setup)
How to continue from there?
How to try to mount the nonempty /etc/pve ?
Thank you
Thank for the script
I try to backup the server with it .
Then I try to restore to same machine but with fresh install (same version of proxmox) but pve-cluster refuses to start
I receive pmxcfs error 255
Tried to run pmxcfs I receive :
fuse: mountpoint is not empty
fuse: if you are sure it is safe , use the 'nonempty' mount option
This happenes (if i understood well) because /etc/pve is not empty (probably as it has my setup)
How to continue from there?
How to try to mount the nonempty /etc/pve ?
Thank you
move your copy of /etc/pve away:
`mv /etc/pve /etc/pve.bck`
restart the pve-cluster service/pmxcfs:
`systemctl restart pve-cluster`
once this works - copy the files you need from /etc/pve.bck to /etc/pve
First BIG THX to all here & esp DerDanilo, & we WILL USE this !!
adding my 2 cents
Instead of pick & choose which dir & so & such .. take the entire root fs, so, if using, gd 'ol fashioned LVM and old school way of using fsarchiver cmd line utility like this:
# create the LV snapshot of the root file system
$ lvcreate -L 10G -s -n snapROOTlabel /dev/pve/root
# run the fsarchiver util or put it into script
$ fsarchiver -j 16 -z 9 -L savethisrootfssnapshotlabel savefs ./nameoffsarchiverfsbackup.fsa /dev/mapper/pve-snapROOTlabel
# -j 16 means use 16 cores (multi threading), -z 9 means use lzma level nine compression (opting to use lzma legacy instead of something like zstd or etc .. ), -L is just labeling it ..
# get rid of the snapshot LV after thE backup script finishes ..
$ lvremove /dev/pve/snapROOTlabel
adding my 2 cents
Instead of pick & choose which dir & so & such .. take the entire root fs, so, if using, gd 'ol fashioned LVM and old school way of using fsarchiver cmd line utility like this:
# create the LV snapshot of the root file system
$ lvcreate -L 10G -s -n snapROOTlabel /dev/pve/root
# run the fsarchiver util or put it into script
$ fsarchiver -j 16 -z 9 -L savethisrootfssnapshotlabel savefs ./nameoffsarchiverfsbackup.fsa /dev/mapper/pve-snapROOTlabel
# -j 16 means use 16 cores (multi threading), -z 9 means use lzma level nine compression (opting to use lzma legacy instead of something like zstd or etc .. ), -L is just labeling it ..
# get rid of the snapshot LV after thE backup script finishes ..
$ lvremove /dev/pve/snapROOTlabel
No, the script currently doesn't have a restore function.
But the backup ist really just an archive of the backuped folders, so you can extract it and put the files of your chosing back.
E.g. the vm<ID>.conf files to recreate all your guests. Of course, the VM/container data (vol or subvols) needs to be there as well, as it is not part of this backup.
I would recommend cherry-picking the part of the configuration you put back, as a newly installed system is sometimes an opportunity for decluttering. At least for me it is.
First, thanks to DerDanilo for the scripts and work! I am doing backups of the PVE installation to PBS but I prefer to shut down my virtual machines when collecting the backup as I have some databases I prefer not be touched by the VM's while taking the backup.
I have added some bits and bobs to for my own needs:
I have only touched the 'prox_config_backup.sh' script. Perhaps someone could find some of my additions useful.
https://github.com/Lennong/proxmox-stuff/blob/master/prox_config_backup.sh
I have added some bits and bobs to for my own needs:
- Shutdown only the running VM's
- Wait for the VM's to fully stop before doing the PVE system backup
- Rsync over SSH between PVE and PBS
- Startup only the VM's that was running before
I have only touched the 'prox_config_backup.sh' script. Perhaps someone could find some of my additions useful.
https://github.com/Lennong/proxmox-stuff/blob/master/prox_config_backup.sh
Last edited:
@DerDanilo, I know this is an old post but I see that the backup script was updated a year ago. Will the backup script support configs such as SDN, Ceph (config not data), OVS switch, etc.?
The updates are commits from other contributors.
I wrote an Ansible role to backup everything in /etc (including the mounted /etc/pve in case of clusters) to a PBS.
Hence I am not using this simple bash script anymore.
Restoring server configuration is not a one click tasks, especially if hardware changed. Hence I did not put much effort into automating that.
You are welcome to open a PR with improvements.
I went a bit further and rewrote the whole thing. Found out about proxmox-backup-client which helps out a great deal. When just throwing up the data we miss out on the dedup features but when using proxmox-backup-client it gets properly recognized by PBS datastore.
Wrapped a bit of shell spagetti around it to be able to control what VM's to start or stop while doing the backup:
https://github.com/Lennong/PVE-Backup/blob/main/pve-to-pbs.sh