I'm new in this group ..."hello"!
I would first like to point out that my English is not very good. I (and google translator) do my best ;-)
I've used bacula but I want switch to bareos. Totay I have install my first bareos-system. The first backups run at the moment. My typical server are look like:
- KVM-Host (Ubuntu 12.04-TLS 64Bit)
- VM: Ipfire for Firewall, OpenVPN
- VM: OpenSuSE 12.2 (LDAP-Server, Samba-PDC)
- VM: CentOS 6.4 for Kolab Groupware
- VM: Windows7 Pro 64Bit (industry-apps from my customer)
- VM: SystemrescueCD (Live) for maintenance work (not in backup)
The storage-device for the vm's are an logical-volume in lvm,
A other physical host for bareos with ubuntu 12.04 LTS.
I install bareos (director, console, FD and SD) on a separate host and the FileDaemon on each client.
Is this a good strategy?
how should I proceed if a vm dies and I have to restore it?
My backup strategy is from the bacula-book.
best regards
sven
> At the moment there is no easy or best way to perfom a live
> snapshot/backup of a running kvm guest.
one is not completely clear to me. I have an KVM-Guest (Windows) with install the bareos-client. My bareos do a backup from all local (fix) media in the guest.
Now the windows-machine die (sometimes happens).
how do I put this machine with my backup, restore?
> There are some cumbersome ways to create live backups of a kvm
> guest, as it is described in that link:
> http://kashyapc.wordpress.com/2013/01/22/live-backup-with-external-disk-snapshots-and-libvirts-blockpull/.
> But i did not test this anyway. So we can't give you a strategy for
> kvm at hand.
I will look at me
> What do you mean with that statement "dies"? what exactly occured on the
> vm? What is stored in the windows event logs about the "crashes"? The
> machine crashes during backups or random? If the windows vm crahses
> during backups, what is keept in bareos.log about that job? Or is that a
> question about best practice ways to restore?
I mean if the windows os does not boot after a crash (random) and must be reinstalled. I need long time for:
- OS-Install
- Update's
- Softwareinstall
...
I am looking for a solution that:
- Fits in my backup strategy (full -> Diff -> Incr)
When I save the vm on the KVM host I only "see" the blockdevice (lv or file) and can't backup diffs or incr.
- Fast reinstall the complete vm (os / prog / data) after a crash
Please wait briefly. I have an idea how it could work. I try that time to put together and post to.
The kvm-host is installed with ubuntu-server 12.04-tls. On this host I have configure a lvm with the volume group 'vmg01'. The vm's use the lv's as a block device and write their partitions in there.
from the perspective of the kvm hosts I can't determine changes on the filsystem. Only one file (or device file in case og lvm).
my idea (another vm, lvmsnapshot, bareos-sript):
My bareos run on an other host (physical). I imagine the drain as before:
1. bareos-director create the lvm-snapshots on the kvm-host
2. bareos-director start my seperate vm with mounted snapshots an installed bareos-client.
3. Bareos backup the vm-volumes (full -> diff -> incr)
4. bareos director shutdown the vm and delete the snapshots
For recovery before I go opposite. I create a new lv for the vm to restore. start my helper-vm with this blockdevice and write the last version of the filesystem in the device's.
would it be possible?
> This one could be done once per week, month depending of the need.
> The restore to normal operation is the key factor of what kind of backup you
> need.
> This scenario cover the destroyed vm incident
> Normal backup : (what's inside the vm)
> Normal bareos backup jobs done in whatever kind of rotation you need.
> Like any real bare machine.
> This one cover the detroyed data inside the vm.
> What I want to be able to restore, and how far in the past I should be able
> to
> do it, plus in how much time I need to be fully back define the backup
> scenario.
I'm trying to find a way to realize this in a scenario