Proxmox VE How To Setup High Availability

In the video below, we show you how to setup High Availability in Proxmox VE

One of the main goals of a hypervisor cluster is to minimise downtime

If you have a standalone hypervisor and that fails, all of the virtual machines it was running will then be out of service

But hypervisors like Proxmox VE offer a high availability service to reduce downtime

Provided the servers in the cluster have access to the same shared storage and the files for the virtual machines are stored there, if a hypervisor fails then HA can automatically restart a VM on another hypervisor

Assumptions

1. A Proxmox VE cluster already exists
2. The nodes all have access to shared storage where VMs are stored
3. The time on the hypervisors is synced - Debian uses chrony for this and the servers will need access to an NTP server

Steps Taken

1. HA Groups
   HA needs a group of hypervisors it can use to run a VM on
   
   Navigate to Datacenter | HA | Groups then click the create button
   
   The ID should be meaningful e.g. if the group should prefer node 1 it could be called Prefer_Node1
   
   NOTE: You cannot rename a group after it has been created
   
   Select the nodes for the group and give them a priority to set the preferences of which node a VM should run on; The higher the priority the more preferred a node is
   
   TIP: Nodes can have equal priority to allow the resource manager to decide
   
   Just bear in mind that you need to balance the resources available when setting preferences
   
   The restricted option is to only allow a VM to be migrated by HA to a node that is part of the group
   
   NOTE: This has no affect on manual migrations
   
   The default setting is unrestricted meaning HA can use other nodes in the cluster if none of the ones in the group can be used
   
   NOTE: If restricted is enabled, a VM will be placed into a stopped state if none of the nodes in the group are available, and until one of the group’s nodes is back online
   
   The nofailback option is to prevent HA automatically migrating a VM to node with a higher priority
   This will happen if a higher preference node is offline for instance and is then brought back online
   
   For stability reasons it may be better to enable this option, because an intermittent server fault could result in it regularly rebooting leading to frequent migration attempts

2. Configure HA
   There are two options to enable HA for a VM
   
   Option 1
   Select the VM
   Click More then select Manage HA from the drop down menu
   
   Option 2
   Select Datacenter
   Click HA
   Click Add then select the VM from the list
   
   The Max Restart option is the maximum number of attempts to restart a VM on the existing node
   
   The Max Relocate option is the maximum number of attempts to migrate a VM to another node
   
   The Request State decides the state the VM should be in and usually the default choice of Started would be selected
   
   Other states are covered in the help section. Click the Help button for more details
   
   

3. Shutdown Policy
   The default shutdown policy shuts down VMs and containers when a node is shutdown from within the GUI for instance
   
   Navigate to Datacenter | Options, select HA Settings and click Edit if you’d prefer HA protected VMs to be migrated instead for instance
   
   Other options are covered in the help section. Click the Help button for more details
   
   NOTE: Rebooting a node still results in VMs being shutdown

4. Cluster Resource Scheduling
   Version 7.3 added a new option for resource allocation when HA migrates a VM that takes into account CPU and memory availability
   
   Navigate to Datacenter | Options, select Cluster Resource Scheduling and click Edit
   
   NOTE: This is still in a testing phase

Sharing is caring!