[PVE-User] HA feature

Gilles Mocellin gilles.mocellin at free.fr
Mon Aug 23 15:30:54 CEST 2010


Le Monday 23 August 2010 15:12:08 Germain Maurice, vous avez écrit :
>   Hi,
> 
> I know there is no release date planned for Proxmox 2 (especially for
> the HA feature).
> However, the feature i really want to use is to recover as soon as
> possible, but not automatically, all the VMs which resided
> on a fail node on an available node in the cluster.
> 
> I use AoE shared storage, so, all VMs disks are known from proxmox slaves.
> I'm wondering and testing if i synchronise /etc/qemu-server/*.conf files
> from master to slaves, would be a good workaround.
> 
> On each slave, i plan to set a crontab task as this :
> * * * * * rsync -e ssh -au master:/etc/qemu-server/ /etc/qemu-server/
> 
> With this, the slaves now know all the VMs created on the master. With
> this workaround, you have now to create the VMs on
> the master and one minute after, the slaves are aware of the new VM.
> 
> The small problem is the state (online/offline) of VMs between master
> and slaves are not synchronised (network conflicts could occur).
> So, you have to be carefully about that.
> 
> What do you think about this workaround ? Any improvement of it ?
> Best regards,

Hi,

As soon as a conf file is in /etc/qemu-server, proxmox shows it as a new VM.
In this case, it would have the same ID as the original... Not good.

What I do is a regular backup of /etc/qemu-server, /etc/vz/, /etc/pve in the 
NFS share where I do my backups with proxmox.

So every not can access this backup, and I can restore the conf files of a 
failed node to another.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 198 bytes
Desc: This is a digitally signed message part.
URL: <http://pve.proxmox.com/pipermail/pve-user/attachments/20100823/bc3befd3/attachment-0013.sig>


More information about the pve-user mailing list