Clustering shares resources?

[kapcom01]

Hello, I would like to know if I run out of CPU resources, can I add another node to help my current node?
So is it possible to use another mashine instead of upgrading my current's mashine CPU?
Thanks.

 

[hotwired007]

if you are wanting to split several VMs between nodes thats what clustering is for.

If you are wanting to split a single VM to run accross several nodes - you cannot do this with proxmox.

Can you provide more information for what you are wanting to do?

 

[kapcom01]

I get it. So if I have many many openvz containers and my CPU cannot handle all of them anymore, I can add another node and migrate some of the containers to the new node. Right?
Thanks.

 

[martin.g]

Correct. But no shared ram or cpu of different nodes for a single VM.

 

[hotwired007]

thats correct - although to migrate them if you are using local hard disks will have to be done with the container powered off.

 

[kapcom01]

OK. Thank you very much!

- - - Updated - - -

thats correct - although to migrate them if you are using local hard disks will have to be done with the container powered off.

Thanks!

 

[stan]

By "migrate" do you mean backup and then restore?

Thanks,

Stan

 

[stavros]

thats correct - although to migrate them if you are using local hard disks will have to be done with the container powered off.

This is only correct for KVM containers, OpenVZ can be moved to different machines with local storage while running.

 

[martin.g]

No, you can do "live migration". Right click on the VM and select Migrate. Then you select the correct target host (and check "Online migration")

 

[stavros]

No, you can do "live migration". Right click on the VM and select Migrate. Then you select the correct target host (and check "Online migration")

That is certainly correct when shared storage is used (iSCSI or NFS), but unless I am mistaken block device migration has not yet been implemented in Proxmox and therefore online migration would not work for *local* storage.

 

[martin.g]

I do not use shared storage either. VMs are on the local storage and they can be live migrated without problems. There may be a part of a second when the final switch is done but my ssh connections persist during this switch.

As far as I know there is done a snapshot in background, then the snapshot is copied to the other server. The latest changes are synchronized by rsync after that and then the switch is done.


weekly_ and daily_ storage is on a NFS share but the VM are on the local storage.

 

[stavros]

I do not use shared storage either. VMs are on the local storage and they can be live migrated without problems. There may be a part of a second when the final switch is done but my ssh connections persist during this switch.

As far as I know there is done a snapshot in background, then the snapshot is copied to the other server. The latest changes are synchronized by rsync after that and then the switch is done.

These are OpenVZ containers, not KVM VMs See my previous answer.