Add Incus server

Configure Ceph

Console

Resources

Deployment templates

NameCreated byCreated

Users

Admin = full control · Lab = create/run/delete instances · View only = read-only.

UsernameGroup

Edit

New instance file

Available resources on this server

New storage pool

⚠ A source starting with /dev/ will erase and reformat that whole device. Leave source blank for a loop file, or point at an existing pool/VG/dataset/dir to reuse it.

Raw config (advanced)

New profile

Config

Advanced config

Cloud-init (optional)

Works the same for containers and VMs. A profile is the natural way to give the same cloud-init setup to every instance that uses it — it only applies on each instance's first boot.

Devices

New network

Raw config (advanced)

Resource
Restore preview

Restore recovery point

This replaces the instance's current state.

Changes made after this recovery point may be lost. Incus volumes attached separately may not be reverted.

Instance
—
Created
—
Type
—
State
—
Protected workflow

Protect before change

Recommended

Create a recovery point first, then continue automatically when it completes.

Alerts

No active alerts.

Recent

—

Alert configuration

Webhook (optional)

Fleet digest

A periodic summary of fleet-wide alerts, confidence, and updates, sent to the same webhook above.

Confirm

Confirm

Notice

Enter a value

Ports

HostContainerProtocol

Logs


    
  

  
    

      
Explain

      
      
    

    

  

  
    

      
Config

      
      
    

    
Resource limits

    

      
      
      
      
    

    

      Cloud-init
      
⚠ Cloud-init only runs on an instance's very first boot. If this instance has already started before, changing these fields here has no effect — there's no way to re-trigger it, restart or otherwise. Set cloud-init at instance-create time (or in a profile) instead.

      

        
        
        
        
        
      

      
      
      
    

    

      Advanced config keys
      

        
KeyValue

      

      

        
        
        
      

    

    

      
      
    

    
    
Profiles

    

    

      
      
    

    
Networks

    

      
InterfaceNetwork

    

    

      
      
      
      
    

  

  
    
Migrate instance

    

    
    
    

  

  
    
Enable clustering

    
This turns this host into a single-member cluster. Other hosts can then join it.

    
    
    
    

    
  
  
    
Add member — join instructions

    
Run this on the host you want to join (it must already be a registered, reachable Incus host). The app does not execute the join.

    
    
      
  1. On the joining host, run incus admin init
    2. 
      
  2. Choose “yes” to join an existing cluster, use this host's address, and paste the token below.
    3. 
    

    
    

  

  
    

      
Image launch workflow
New instance
Choose an image and configure the workload.

      
      
      
      

        
        
        

        
      

      
      
      
      
      
      
      
      
      
      
      
      

      
      
      

        Cloud-init (optional)
        
Works the same for containers and VMs. Only applies on this instance's first boot, so this is the only place setting it actually takes effect — editing it later on the Config dialog has no effect.

        

          
          
          
          
          
        

        
        
Select a (cloud) image version first. These templates work for Linux containers and VMs on their first boot. They do not install cloud-init in a default image.

        
        
        
        
A password enables console login and sudo with that password. SSH password login stays disabled. Generated YAML contains the password; do not share it or save it in a reusable deployment template. Key-only accounts have password login locked and passwordless sudo.

        
        

        
        
        
      

      
      
Review
Workload
Container
Image
debian/12
Name
Not named yet

      
        
        
        
        
      
      

    

  

  
    

      
⚠ Evacuate the only available cluster member

      

      
      

        
        
      

    

  

  
    

    
Cluster recovery — restoring a stuck Incus host

    
If an environment is unreachable and the daemon logs show no available cowsql leader, the Incus API is down and the panel cannot fix it remotely. These are host-level steps to run over SSH. Back up before recovery, and confirm subcommands with incus admin cluster --help for your version.

    
    
      
  1. Diagnose
        
    sudo journalctl -u incus -n 80 --no-pager
    
        
        
    Look for no available cowsql leader.
    2. 
      
  2. Force the daemon down (a stuck daemon ignores a graceful stop)
        
    sudo systemctl kill --signal=SIGKILL incus.service
sudo systemctl stop incus.socket
ps aux | grep -i incusd | grep -v grep
sudo pkill -9 incusd   # only if the line above still shows incusd
    
        
    3. 
      
  3. Back up the database
        
    sudo cp -a /var/lib/incus/database /var/lib/incus/database.bak-$(date +%s)
# snap install: /var/snap/incus/common/incus/database
    
        
    4. 
      
  4. Inspect, then recover (daemon stopped)
        
    sudo incus admin cluster list-database
sudo incus admin cluster recover-from-quorum-loss
    
        
    5. 
      
  5. Restart and verify
        
    sudo systemctl start incus.socket incus.service
sleep 20 && incus cluster list
    
        
    6. 
      
  6. Un-evacuate (if the member shows EVACUATED)
        
    incus cluster restore <member-name>
incus start <name>
    
        
    7.