Hello,
I plan to upgrade my Proxmox Cluster 3.4 to 4.1 and I need advice or upgrade the way that I've chosen will work.
The current configuration:
Proxmox 3.4 cluster with nodes 5
Node 1,2 and 3 are part of a cluster Proxmox but serve only as CEPH server (Proxmox configuration) are not for running any VM.
Node 4 and 5 are compute nodes and VMs running on them.
HA is enabled and configured, but have only one resource - manage IP address.
Now my plan:
a) On node 2. I add 4 x 5TB disk and create RAID10, share it via NFS and add the compute nodes.
b) I use the function "move disk" to move virtual disks from Ceph storage to NFS storage created at step a)
c) Online migrate all VMs from node 4. to node 5.
d) Now i remove node 4. from a cluster, reinstall to PVE 4.1 and create a new cluster on it. I add NFS share from step a) on it and step by step I turn off all VMs from node 5. and manually copied VM configuration to Node 4. and run.
e) Now step by step reinstall nodes 1, 3 and 5 to 4.1 PVE and add them in to a cluster running on node 4. On nodes 1. and 3. I configure Ceph server (to which I later add a node 2.)
f) Use the "move disk" to move virtual disks from NFS to Ceph storage created in step d).
g) Reinstall node 2. to PVE 4.1 and add it to a cluster, configure Ceph and add it as another Ceph storage node.
h) Done
- Do you think it is possible to use this procedure?
- I can create only one cluster node at which copies the VMs and run?
- Is posible to have two different clusters on one network (switch) at the same time?
- May I add nodes from step e) in to the new cluster created in step d) if on this node (node 4.) have running VMs
I plan to upgrade my Proxmox Cluster 3.4 to 4.1 and I need advice or upgrade the way that I've chosen will work.
The current configuration:
Proxmox 3.4 cluster with nodes 5
Node 1,2 and 3 are part of a cluster Proxmox but serve only as CEPH server (Proxmox configuration) are not for running any VM.
Node 4 and 5 are compute nodes and VMs running on them.
HA is enabled and configured, but have only one resource - manage IP address.
Now my plan:
a) On node 2. I add 4 x 5TB disk and create RAID10, share it via NFS and add the compute nodes.
b) I use the function "move disk" to move virtual disks from Ceph storage to NFS storage created at step a)
c) Online migrate all VMs from node 4. to node 5.
d) Now i remove node 4. from a cluster, reinstall to PVE 4.1 and create a new cluster on it. I add NFS share from step a) on it and step by step I turn off all VMs from node 5. and manually copied VM configuration to Node 4. and run.
e) Now step by step reinstall nodes 1, 3 and 5 to 4.1 PVE and add them in to a cluster running on node 4. On nodes 1. and 3. I configure Ceph server (to which I later add a node 2.)
f) Use the "move disk" to move virtual disks from NFS to Ceph storage created in step d).
g) Reinstall node 2. to PVE 4.1 and add it to a cluster, configure Ceph and add it as another Ceph storage node.
h) Done
- Do you think it is possible to use this procedure?
- I can create only one cluster node at which copies the VMs and run?
- Is posible to have two different clusters on one network (switch) at the same time?
- May I add nodes from step e) in to the new cluster created in step d) if on this node (node 4.) have running VMs