I’m trying to set up the backup image ‘snapshot’ in client linux centos 7, I install the client with option 2
Please select the snapshot mechanism to be used for backups:
- dattobd volume snapshot kernel module from https://github.com/datto/dattobd
- LVM - Logical Volume Manager snapshots
- Use no snapshot mechanism
The file /usr/local/etc/urbackup/snapshot.cfg:
the backup of directories is ok,no errors occurred but the backup of snapshot occored erros
Starting scheduled full file backup…
Creating snapshot of “/etc” failed
Using default stripesize 64.00 KiB.
Calculated size of logical volume is 0 extents. Needs to be larger.
Creating snapshot of “etc” failed.
When I run the command ‘Run full backup image’ in the WEB manager does not show me errors in the log
Do I need to change or create more settings in snapshot.cfg? I did some research on google but not found solutions…
Any ideas !!
My understanding is you can’t do image backups on Linux.
huumm… This backup is in a XEN VM, the directories backups are ok dont show erros
The VM disk is an LVM, in this case the configuration is different?
It s file backup with snapshot, not full disk image as you think about it in windows.
Because it s hard to get a good constancy for a full disk image on linux, i dont think there’s a lot of software that even support it online.
Hi @orogor what would you reommend. I have a Centos 7.7 and I would like to take image backup (like in windows) . then later I would like to restore it to new machine.
I’m like new born in terms liunx OS’s.
any suggestion would be appreciated.
For it to work there would be some requirement on the client
It would need to be formated as btrs or zfs or using lvm, and that for all the partitions, even /boot, then snapshots it, then backup the snapshot , then destroy it.
Maybe it’s possible with Ext3/4, ReiserFS, JFS, XFS using fsfreeze and dump without lvm, but that would make you unable to write to that fs during the backup.
ext2 will not do, neither fat32.
You would also need to dump the partition table using something like sfdisk -d
In that sense maybe btrfs or zfs using whoel disk without partition is the simplest
Then you could use urbackupclientctl with an option that can backup stdout (can’t remember it).
Also the storage space would be prohibitive there would need either a special support for each fs and use zfs/btrfs send/receive; or hash the fs in large chuncks like 100MB, create a map of the fs, then track changes that way, so there is some dedup going on between backups.
As for restoring then it should be possible if you have all the data available, but you’d need to do it outside urbackup.
The theory is that it would work, but i never actually made it work, then test it.
I remember dumping an fs this way but not doing the restore and ending up making a feature request to write plugins and support server side scripts @uroni so that could try to integrate what i did at that time.