VM / Container migration between unclustered hosts

[Karl Dane]

Hi,

We have one three-node PVE cluster and several standalone PVE hosts. We'd like to be able to migrate VMs and containers between the standalone hosts and the cluster in a similar way to how we already can between nodes in the cluster using the web interface.

Right now I can do something like this:

root@standalone1:~# vzdump 10001 -stdout -mode stop | ssh root@clusternode1 "qmrestore - 10001 -storage Storage"

but this prevents other less unix-oriented admins from performing the same task.

Is it possible to connect the standalone hosts to the cluster while not taking full part? We'd like to maintain a degree of separation in case the cluster as a whole suffers any problems. Or is there another approach that would make more sense in this situation?

Best regards,

Karl Dane

 

[guletz]

Hi,

We have one three-node PVE cluster and several standalone PVE hosts. We'd like to be able to migrate VMs and containers between the standalone hosts and the cluster in a similar way to how we already can between nodes in the cluster using the web interface.

Right now I can do something like this:

root@standalone1:~# vzdump 10001 -stdout -mode stop | ssh root@clusternode1 "qmrestore - 10001 -storage Storage"

but this prevents other less unix-oriented admins from performing the same task.

Is it possible to connect the standalone hosts to the cluster while not taking full part? We'd like to maintain a degree of separation in case the cluster as a whole suffers any problems. Or is there another approach that would make more sense in this situation?

Best regards,

Karl Dane

So the less unix-dissoriented admins can not do a copy/paste and then replace with the proper values ? In my opinion any PC user can do this. If they can not do this simple task ... I think they are not admins. By the way, this admins can know how to power up a monitor?

Now assuming that you do not have any admins around, you can create a bash script who can do what you need(passwordless and with arcfour cipher if you want speed) for the desired VMs.

 

[dcsapak]

Is it possible to connect the standalone hosts to the cluster while not taking full part? We'd like to maintain a degree of separation in case the cluster as a whole suffers any problems. Or is there another approach that would make more sense in this situation?

this is not directly possible at the moment

there are some patches on the dev-list for cross cluster migration, but i guess they still need work, so no idea if/when they get merged
if you use zfs, you can use zfs-send/receive (or pve-zsync) to replicate one vm to another zpool

also we want to have something to manage multiple clusters, but this is only in the planning stage

 

[Karl Dane]

this is not directly possible at the moment

there are some patches on the dev-list for cross cluster migration, but i guess they still need work, so no idea if/when they get merged
if you use zfs, you can use zfs-send/receive (or pve-zsync) to replicate one vm to another zpool

also we want to have something to manage multiple clusters, but this is only in the planning stage

Thank you for the helpful reply. I'll investigate the zfs solution.

 

[guletz]

Thank you for the helpful reply. I'll investigate the zfs solution.

zfs send/recive is the fastes .