Il 2021-04-16 14:35 Nikolaos Milas ha scritto:
On 16/4/2021 10:10 π.μ., Felix Kölzow wrote:
you may take a look at _rear_ (it is that easy how it looks like)
You may also want to check mondo:
http://www.mondorescue.org/about.shtml (the website is not updated regularly!)
I am using it for many years, since CentOS 5 (until now with CentOS 8).
There may be a few glitches, but otherwise I am pretty satisfied.
It supports bare metal recovery and can be used for cloning too.
Cheers, Nick
Hi, while relax-and-recover, mondorescue and similar tools have their place, I mostly find them outdated for VM backup. The main reason is that they have difficult providing low RTO and RPO.
For my current low-cost linux hypervisors I use a ZFS filesystem, with backups done leaveraing a mix of both ZFS send/recv and rsnapshot. More specifically: - rolling hourly snapshots are taked via sanoid; - the primary datastore (with raw disk images) is replicated with 1min interval (thanks to send/recv) to a second, standby server. This guarantee a byte-for-byte copy of all the virtual machines backing files *and* the capability to quickly restart all VMs in case the first server dies; - a third machine does rsnapshot-based backup of the data inside the VMs themselves, for added convenience in recovering a single file and for longer historical retention.
This backup scheme as served me very well. Just my 2c. Regards.