Two Node Failover and Hardware Recommendations?

[CHR15M]

Hello! I am testing out Proxmox and really like what I am seeing so far. (can't wait for the native Veeam integration) Let me preface this with I am a VMWare guy as I imagine many here are/were. I have a vision for some of our remote sites to have a domain controller that can failover as well as a little capacity for a VM or two if needed. Does ProxMox work well with a two node failover scenario? (thinking replication to a second host) I have tested with three nodes and ceph it works really well but would like to keep cost down as much as possible and high availability isn't a must. (not saving lives here) I am also curious what you all are using for hardware. I am a Dell guy so traditionally I would go with 1U or 2U servers but am also looking to minimize the footprint in our racks. I have stumbled across the MS-01 which looks promising with all of the network connectivity that it has in a small package but I don't know much about the company. Any and all opinions are welcome!

 

[sw-omit]

As a original VMWare-person here too, I would suggest to (in case you haven't already) take a look at the cluster-manager wiki [1], as there is one big difference you need to be aware with between vmware and proxmox clusters: There is no "master" (vcenter) in proxmox, so MORE then half of the nodes need to be in sync for the cluster to work.
So with 2 nodes, if one of them goes down, you're at half, so not MORE then half, so the cluster goes read-only.
Everything that runs will stay running, but changes (like starting of VM's) will be blocked without extra steps.
On top of that, because of low latency requirement, having the servers far apart / over a VPN, will have possible consequences as well.

That said though, using them as two separate servers and using either backup-restore or pve-zsync [2] are probably more stable in that regard.

As for hardware, can't really help you with that, used to HP and SuperMicro myself and they work good for our purpose, but haven't diven into them much.

[1] https://pve.proxmox.com/wiki/Cluster_Manager
[2] https://pve.proxmox.com/wiki/PVE-zsync

 

[CHR15M]

As a original VMWare-person here too, I would suggest to (in case you haven't already) take a look at the cluster-manager wiki [1], as there is one big difference you need to be aware with between vmware and proxmox clusters: There is no "master" (vcenter) in proxmox, so MORE then half of the nodes need to be in sync for the cluster to work.
So with 2 nodes, if one of them goes down, you're at half, so not MORE then half, so the cluster goes read-only.
Everything that runs will stay running, but changes (like starting of VM's) will be blocked without extra steps.
On top of that, because of low latency requirement, having the servers far apart / over a VPN, will have possible consequences as well.

That said though, using them as two separate servers and using either backup-restore or pve-zsync [2] are probably more stable in that regard.

As for hardware, can't really help you with that, used to HP and SuperMicro myself and they work good for our purpose, but haven't diven into them much.

[1] https://pve.proxmox.com/wiki/Cluster_Manager
[2] https://pve.proxmox.com/wiki/PVE-zsync

Thanks for the recommendation. I will certainly read over that this afternoon. I am wondering once the Veeam integration goes live if I can do some replication magic between the two hosts like you mentioned with the built in tools. (nonclustered) How do you like the SuperMicro servers? They seem to have a wide range of hardware options.

 

[johnblue]

I have stumbled across the MS-01 which looks promising with all of the network connectivity that it has in a small package but I don't know much about the company.

For anyone that is curious about the MS-01:

https://www.youtube.com/watch?v=W3i5pKLwYak

 

[sw-omit]

How do you like the SuperMicro servers? They seem to have a wide range of hardware options.

They do their job, have some nice features, but also is missing some for me basic options like being able to set the auto-start after power-failure from the ILO/IPMI (has to be done during boot-bios). Still prefer HPE over SuperMicro, but of the costs are often higher at HPE too.

 

[esi_y]

Does ProxMox work well with a two node failover scenario? (thinking replication to a second host)

Yes albeit you really would want to have a Q device (external is not an issue). Alternatively, especially if you do not care for HA, you can consider some special settings for two node clusters, see e.g. here:

https://forum.proxmox.com/threads/asking-for-your-experience.152283/#post-690773