Suggestions on migration

carlosmp

Renowned Member
Jun 2, 2010
48
1
73
Hi,

We currently have a 2 node proxmox without HA. We just got some newer servers in, and want to migrate from the current 2 node setup and end up with a 3 node with all the bells and whistles including ceph. I'm trying to think through the process and see if this makes sense, or if I should be thinking of other things:

- Add the 3 new nodes to the existing cluster
- setup the HA/fencing on the 3 new nodes (Is this possible, leaving the other 2 nodes out?)
- configure ceph on the 3 new nodes
- migrate existing containers to one of the 3 nodes.
- kill the old nodes
- celebrate

Also, if we have a subscription on the two existing nodes can those be migrated to the new nodes?

Thanks,
Carlos.
 
Hi,

We currently have a 2 node proxmox without HA. We just got some newer servers in, and want to migrate from the current 2 node setup and end up with a 3 node with all the bells and whistles including ceph. I'm trying to think through the process and see if this makes sense, or if I should be thinking of other things:

- Add the 3 new nodes to the existing cluster
- setup the HA/fencing on the 3 new nodes (Is this possible, leaving the other 2 nodes out?)
should work.
- configure ceph on the 3 new nodes
- migrate existing containers to one of the 3 nodes.
containers?! You know that ceph is for kvm-storage only?
- kill the old nodes
- celebrate

Also, if we have a subscription on the two existing nodes can those be migrated to the new nodes?

Thanks,
Carlos.
Udo
 
should work.
containers?! You know that ceph is for kvm-storage only?
Udo

Yes, we're going to try to see if we can use the NFS export with ceph for the containers...will be interesting. It will slow the speed down a bit, but realistically, disk IO is not a concern with our environment. If this fails, we'll just setup the 3 ceph nodes as ceph only, and then setup some proxmox nodes. One of our nodes is enough to handle all our containers/kvms, maybe just need a bit more RAM.

Thanks,

Carlos.