Hello,
I'm a sysadmin of a small company and I'm looking to replace the current really OLD proxmox cluster (proxmox 4.3-9).
I would like to make a new 3 nodes cluster with ceph,
but I'm quite new to proxmox since le current cluster was made by another company a long time ago when I was not in this company.
I'm thinking to buy 3 of this:
- HPE ProLiant DL380 Gen10 4214R 1P 32 GB-R P408i-a NC 8 SFF 800 W (https://buy.hpe.com/it/it/servers/p...r-p408i-a-nc-8sff-800w-ps-server/p/P24842-B21)
this server has:
Intel® Xeon® Silver 4214R
32 GB RAM
4x ethernet NIC 1GB
no storage
- another 32 GB of RAM for each server
- 2 X 1TB SSD for each server for ceph storage
- 1 X 500 GB SSD for each server for proxmox and linux installation
- 1 X network card 10 GB with 2 NIC for each server
For the CEPH network,
I'm thinking to make a full mesh network with RSTP configuration,
and to connect each 10GB NIC of a server to the other 2 servers.
I will use the 4x1GB ethernet NICs of each server to connect them to the current network for the normal access from the users.
In the cluster there will be 4 VM (but the number will increase in the future and I know, I will need to update all these windows OLD instances...):
- 1x windows 2012 terminal server (VM1)
- 1x windows 2008 database server (VM2)
- 1x windows 2008 for an old lotus domino instance not used anymore but that we need to keep (VM3)
- 1x windows 2012 web server for our ecommerce website (VM4)
I would like to balance the VM on the different nodes,
so I will put the VM1 in the node 1, the VM2 on the node 2, and the VM3 and VM4 on the node 3
The backup will be managed by a separate Proxmox Backup Server in a separate machine (I'm thinking to use one of the servers of the current proxmox cluster that I will wipe and install the Proxmox Backup Server)
Can I please ask you a couple of questions?
- do you think that my configuration will be ok or do you see any flaws in it?
- do you think that proxmox will be compatible with that server?
- a lot of data is daily exchanged between VM1 and VM2, but in my opinion it would be better to have them on different nodes for performance reasons.
Can I set the data exchanged between the VMs (and so by the different nodes/servers) to go on the 10 GB link?
(I mean, that network will be used for ceph, would it be a nice idea to use it also for exchange data between VMs?)
- will I be able to import the current VM that I have on proxmox 4.3-9 to the new proxmox version?
- I have an UPS that is APC.
Currently the UPS, in the case of energy interruption, gracefully shuts down the servers.
I would like to do the same on my new proxmox cluster, but I have troubles to understand how to set the different shutdowns and starts with ceph.
In my opinion, in the energy gets interrupted, the cluster should be shut down in the following order:
- first all the VMs of the different nodes should be shut down
- then the node3 should be shut down
- then the node2 should be shut down
- then the node1 should be shut down
and when the energy is back:
- all the nodes/servers will start back automatically, and also proxmox and ceph would start automatically
- then all the VMs should start automatically
is this configuration correct?
Thank you very much for all the help and sorry for all the questions, it looks simple but this is my first proxmox cluster
I'm a sysadmin of a small company and I'm looking to replace the current really OLD proxmox cluster (proxmox 4.3-9).
I would like to make a new 3 nodes cluster with ceph,
but I'm quite new to proxmox since le current cluster was made by another company a long time ago when I was not in this company.
I'm thinking to buy 3 of this:
- HPE ProLiant DL380 Gen10 4214R 1P 32 GB-R P408i-a NC 8 SFF 800 W (https://buy.hpe.com/it/it/servers/p...r-p408i-a-nc-8sff-800w-ps-server/p/P24842-B21)
this server has:
Intel® Xeon® Silver 4214R
32 GB RAM
4x ethernet NIC 1GB
no storage
- another 32 GB of RAM for each server
- 2 X 1TB SSD for each server for ceph storage
- 1 X 500 GB SSD for each server for proxmox and linux installation
- 1 X network card 10 GB with 2 NIC for each server
For the CEPH network,
I'm thinking to make a full mesh network with RSTP configuration,
and to connect each 10GB NIC of a server to the other 2 servers.
I will use the 4x1GB ethernet NICs of each server to connect them to the current network for the normal access from the users.
In the cluster there will be 4 VM (but the number will increase in the future and I know, I will need to update all these windows OLD instances...):
- 1x windows 2012 terminal server (VM1)
- 1x windows 2008 database server (VM2)
- 1x windows 2008 for an old lotus domino instance not used anymore but that we need to keep (VM3)
- 1x windows 2012 web server for our ecommerce website (VM4)
I would like to balance the VM on the different nodes,
so I will put the VM1 in the node 1, the VM2 on the node 2, and the VM3 and VM4 on the node 3
The backup will be managed by a separate Proxmox Backup Server in a separate machine (I'm thinking to use one of the servers of the current proxmox cluster that I will wipe and install the Proxmox Backup Server)
Can I please ask you a couple of questions?
- do you think that my configuration will be ok or do you see any flaws in it?
- do you think that proxmox will be compatible with that server?
- a lot of data is daily exchanged between VM1 and VM2, but in my opinion it would be better to have them on different nodes for performance reasons.
Can I set the data exchanged between the VMs (and so by the different nodes/servers) to go on the 10 GB link?
(I mean, that network will be used for ceph, would it be a nice idea to use it also for exchange data between VMs?)
- will I be able to import the current VM that I have on proxmox 4.3-9 to the new proxmox version?
- I have an UPS that is APC.
Currently the UPS, in the case of energy interruption, gracefully shuts down the servers.
I would like to do the same on my new proxmox cluster, but I have troubles to understand how to set the different shutdowns and starts with ceph.
In my opinion, in the energy gets interrupted, the cluster should be shut down in the following order:
- first all the VMs of the different nodes should be shut down
- then the node3 should be shut down
- then the node2 should be shut down
- then the node1 should be shut down
and when the energy is back:
- all the nodes/servers will start back automatically, and also proxmox and ceph would start automatically
- then all the VMs should start automatically
is this configuration correct?
Thank you very much for all the help and sorry for all the questions, it looks simple but this is my first proxmox cluster