Proxmox High Availability
June 15, 2024
Like many other hypervisor solutions, Proxmox has its own form of high availability.
High availability varies across solutions, typically in terms of how responsive they are to failures, along with how they manage VM placement with grouping rules (affinity, anti-affinity and so on).
Most, including Proxmox, are dependent on the VM storage being shared between hypervisor nodes.
Understanding HA behaviour
Proxmox HA will migrate and start up machines, based on priority assignment, and in the event of a node failure.
It is important to note that the HA behaviour does not run an active-passive VM failover copy, but rather it manages the start up of the VM based on the priority nodes to ensure that it is available. The shared storage allows the hosts to achieve this.
Any application running on a VM in this way will need to auto-startup and will require a level of resilience to be built into the application itself.
High Availability Setup
To configure high availability in Proxmox you will need to have.
- A cluster of 2 or more Proxmox nodes.
- Shared storage, with the VM you want to run in HA running using this.
The setup steps consist of:
- Creating an HA Group and adding the Proxmox nodes to this.
- Setting the priority of the nodes in the HA group as desired.
- Adding the VM to HA and selecting the HA Group.
Creating the HA Group
In the Proxmox datacentre, go to HA and into groups. You can now click on “Create” to create an HA group.
Proxmox nodes can now be assigned to the HA group and given a priority - VMs will migrate their highest priority node, and will start on the highest available node in the event of a failure.
Note: It is possible to create multiple HA Groups and assign these to the same nodes. You may want to do this with different priority settings (eg node 1 priority 1 in HA group 1, node 2 priority 1 HA group 2) to control which VMs should be placed on different nodes, but to spread the overall workload throughout the cluster.
Adding VMs to HA
Under the VM view, click on “More” and then “Manage HA”.
You can then configure the HA Group and request state, which should generally be set to started to make use of the feature. The Max Restart and Max Relocate settings are essentially retry settings that determine how many times to retry in the event of a failure.
Video Demonstration
A video demonstration of the HA setup, migration and failover behaviour can be seen here: