[PVE-User] taking advantage of the lvm snapshot

[Holger Ernst]

Hi Tonci,

I am using a python script to sync a snapshot of a Windows 2008 machine over a laser line (100Mbit/s) to another physical location.
You find it here: https://gist.github.com/1843764

It takes aprox. 90 min to sync 500GB (depending on amount of changed data) while using all of the line.

Regards, Holger

> Auftrag von Tonci Stipicevic
> Gesendet: Samstag, 7. Juli 2012 00:06
> An: pve-user at pve.proxmox.com
> Betreff: [PVE-User] taking advantage of the lvm snapshot
> 
> Hello to all,
> 
> I'm running pve2.1 cluster with shared lvm2 storage (nexenta) and am
> very satisfied so far .
> 
> scheduled (and manual)  backups are working very good and backup files
> are copied to another shared nfs storage.
> 
> So when we have to return to the backup first we have to restore the
> archive what could last depending on the amount of data .
> 2nd.. every backup takes time .. but when we make backup of the VM that
> are on the lvm volumes then 1st step is snapshoting the logical VMID
> volume  which makes very useful quasi-backup of the VM.
> We are using it to make archive of the VM in the same time the original
> VM is running which makes zero-downtime .. which is also very very good.
> 
> But using this backed-up VM again means restoring it again which also
> takes time.
> 
> So my question would be whether this snapshot (i.e. 101-s) could be used
> to make copy (delta-replica) to  another-equivalent logical volume (i.e.
> 201 ) on the second storage   so that this 201-VM is ready to run
> without restoring the archive. So VM201 will always be down so that
> logvolume cannot be locked ... (Am I right?)
> 
> I hope dd could do that job but dd cannot transfer only changed blocks
> next time copying is done ?  or it can ?
> 
> So my idea would be to make fast cloning of VMs on the another storage
> that are usable right away !   is there any scenario to achieve this ?
> 
> This could be something called offline-VM-cloning (for me it is not
> necessary to achieve drbd features )
> 
> Since we had snapshot that is done in moment it would very useful if we
> could transfer only changes over to the 201 volume  ( after the initial
> full replication )
> 
> 
> Thank you very much in advance for your attention
> and help
> 
> srdačan pozdrav / best regards
> 
> Tonči Stipičević, dipl. ing. elektr.
> 
> SUMA Informatika d.o.o.
> Badalićeva 27 10000 Zagreb
> fax: 01 5560007
> mob: 091 1234003
> 
> _______________________________________________
> pve-user mailing list
> pve-user at pve.proxmox.com
> http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-user