Proxmox / VM storage on NAS over 10GbE

[azop]

I'm looking at deploying two Proxmox servers on the same LAN. I have a third server with FreeNAS installed and all servers have 10GbE (fiber if that matters). I'd like to host 3 VM's (MSSQL VM, Windows 10 VM and a CentOS VM) on the FreeNAS server which would host a different of pools with SATA and SSD (all 6gb/s).

Is there any reason not to do this? Would you recommend iSCSI, NFS or iSCSI w/ ZFS? Eventually it would be nice to migrate the VMs between the two servers (not as a cluster really...just move them if the host needs a reboot).

Thanks

 

[sboesch]

If you want to cluster the Proxmox servers for easy migration, you'll need a 3rd device for the quorum, this can be a Raspberry Pi. If you do not want to cluster them, this makes VM migration a step more tedious.
I used to run FreeNAS as storage, now I run Proxmox as a storage device for everything, I found there was no real need for me to have VM's and storage on FreeNAS and Proxmox, as Proxmox can be a NAS as well. Sharing in ZFS and NFS is easy, zfs set sharenfs=on pool/mypool and for SMB zfs set sharesmb=on pool/mypool. This will eliminate the need for FreeNAS and you can use all 3 hosts in a Proxmox cluster.
I present all my storage to Proxmox using NFS.

 

[josh]

If you want to cluster the Proxmox servers for easy migration, you'll need a 3rd device for the quorum, this can be a Raspberry Pi. If you do not want to cluster them, this makes VM migration a step more tedious.
I used to run FreeNAS as storage, now I run Proxmox as a storage device for everything, I found there was no real need for me to have VM's and storage on FreeNAS and Proxmox, as Proxmox can be a NAS as well. Sharing in ZFS and NFS is easy, zfs set sharenfs=on pool/mypool and for SMB zfs set sharesmb=on pool/mypool. This will eliminate the need for FreeNAS and you can use all 3 hosts in a Proxmox cluster.
I present all my storage to Proxmox using NFS.

How do you deal with the storage node being a single point of failure for the cluster? I removed the other hosts in my setup because if the host with the drives go down the entire setup goes down anyway.

 

[sboesch]

How do you deal with the storage node being a single point of failure for the cluster? I removed the other hosts in my setup because if the host with the drives go down the entire setup goes down anyway.

You need clustered storage, glusterfs is a good option, how to is here.
Ceph is also a good option, how to is here.
Or you could add another device into the mix as storage only.

 

[RobstarUSA]

I'm using this setup with a CIFS share. I typically only have 1 vm server, except when I need to do repair on it. Then I add an identical server to the cluster & move everything to the CIFS share and migrate the vms.

Then I take down the original node I'm going to upgrade do the upgrade & bring it back up & move everything back.