[PVE-User] Incremental vzdump
andrew at rocky.edu
Wed Mar 3 16:46:46 CET 2010
I think that you will find that if you take one snap shot every couple of
weeks and pair that with a incremental backup soultion (I use backuppc) you
will find that to be really elegant solution. I stagger my vzdump images so
that two get taken per night and pair that with backuppc. It works I have
had to restore a couple times and the process is easy, just restore the
recent image with vzdump then click the restore button on backuppc.
Hope that helps,
On Wed, Mar 3, 2010 at 8:32 AM, Adam <adam at blackfoot.co.uk> wrote:
> Thanks for your reply.
> We do appreciate, and agree with, the intent to not implement a full backup
> solution. However I think there is perhaps a space between third party
> backup solutions and the ability to do OpenVZ/Proxmox specific backups on a
> per container level.
> We currently have a third party solution to incrementally backup the entire
> proxmox hosts, but do miss this ability to easily restore a specific
> container, which is where our thoughts are coming from.
> As mentioned in my original email we effectively just want to replace the
> final 'tar' part of the vzdump from LVM snapshot to an archive, with an
> rsync to a dir. Rather than re-implement the entire process we perhaps think
> this would be best suited somewhere within the vzdump code, making use of
> the existing LVM snapshot/storage models internal to Proxmox.
> I think perhaps we'll look at putting something together for our specific
> purposes, see how things work out in terms of performance, and then look at
> the possibility of contributing a patch if things work well.
> - Adam
> On 03/03/2010 12:38, Martin Maurer wrote:
>> You are on the right list (there is only one in the moment).
>> We do not think that Proxmox VE (vzdump) should be a full backup
>> solutions, there are already a lot of them around with millions of features.
>> The project will focus on the core (Virtualization& Management). If you
>> keep the container small, vzdump works quite well - use bind mounts to
>> include the big data parts.
>> Just think of Bacula and also other open source solution - vzdump can
>> never cover all this. I would use a combination of a backup and vzdump, you
>> already went this way.
>> But yes, we accept patches and enhancements for vzdump, I suggest you post
>> in detail what you plan here (before you start implementing) and you can
>> post patches or if needed, you can get access to svn also.
>> Br, Martin
/-\ ndrew Niemantsverdriet
Rocky Mountain College
1511 Poly Dr.
Billings MT, 59102
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the pve-user