[PVE-User] WG: WG: DRBD+ Proxmox2.1 + Debian Squeeze

Muhammad Yousuf Khan sirtcp at gmail.com
Tue Aug 28 16:54:52 CEST 2012

On Tue, Aug 28, 2012 at 6:32 PM, Holger Ernst <he at ernstdatenmedien.de> wrote:
>> Betreff: Re: [PVE-User] WG: DRBD+ Proxmox2.1 + Debian Squeeze
>> > Then you have to be aware that starting a snapshot of a running
> system
>> > is like switching a running machine off and on again.
>> > So make sure the os can handle that.
>> Thanks for sharing. but what if i use DRBD isnt it like switching
>> machine off and on when my primary base machine goes down ?
> [|] That is right. Even using proxmox backup in snapshot mode will
> produce such a situation. This is happening always if you copy a running
> system without a methode to copy and restore memory content and cpu
> state.
> So propper method would be shutdown/copy/start - but this results in
> serious down-time.
> So you chose between no downtime and eventually corrupted machine and
> lot of downtime and no corruption in any way.
> But luckily most of the time the os'es are robust enough...
>> as per my very limited knowlege. DRBD is also kind of a mirror b/w two
>> computer or Raid 1.  in Rsync the problem is that i have to do it
>> manually every hour and in DRBD the changes are updated instantly.
> [|] That is correct but if you use DRBD in sync mode (that is what you
> should do) disk io is limited to the connection speed between the two
> syncing nodes. So in some scenarios it is not applicable.

Thanks for very informative feedback but things still end up to my old question

if rsync, snapshot , drbd more or less same (i mean risk ratio of the
data that they copy to another node or in shape of bakup) . regardless
of live migration or DRBD HA. then why people do not like to use
because if i restore from snapshot or copy hard drive image *.raw of
one VM to same VM on another machine then it should have work.

how old the backup is, is not a matter . but the matter is what we
have as a backup
 if Rsync copy every hour it would be more better then taking a
snapshot on weekly basis.
DRBD will take the lead as it would be doing instant data upload but 1
hour old backup would also work in my case.

please throw some light on this ........


>> please correct me if i understand some thing wrong .......
> regards, Holger
> _______________________________________________
> pve-user mailing list
> pve-user at pve.proxmox.com
> http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-user

More information about the pve-user mailing list