Configuration tips

The main configuration information is provided in OpenNebula configuration. Here you can find additional configuration details.

Extras

!!! Warning !!! These features are changing the defalt behavior of OpenNebula and should be used when the StorPool addon is the only storage driver, and when used for both SYSTEM and IMAGE datastores!

VM snapshots

The feature is working only with recent version of StorPool. Please contact StorPool Support for details.

The StorPool addon can re-purpose the OpenNebula’s interface for VM snapshots to do atomic snapshots on all VM disks at once. In addition, there is an option to enable FSFREEZE before snapshotting and FSTHAW after the snapshots are taken.

To enable the feature do the following changes.

  1. In /etc/one/oned.conf:

    1. Edit the VM_MAD configuration for kvm and change the ARGUMENTS line as follows:

          ARGUMENTS      = "-t 15 -r 0 kvm -l snapshotcreate=snapshot_create-storpool,snapshotrevert=snapshot_revert-storpool,snapshotdelete=snapshot_delete-storpool",
      
    2. Edit the VM_MAD configuration for kvm and change KEEP_SNAPSHOTS to ‘yes’:

          KEEP_SNAPSHOTS = "yes",
      
  2. In /var/lib/one/remotes/addon-storpoolrc:

    1. Add VMSNAPSHOT_OVERRIDE configuration variable:

      VMSNAPSHOT_OVERRIDE=1

    2. To enable the optional fsfreeze/fsthaw:

      VMSNAPSHOT_FSFREEZE=1
      

Please restart the opennebula service to activate the changes.

Using StorPool SYSTEM datastore when there are other SYSTEM datastores available

Using scheduler filters

To tell the OpenNebula’s scheduler to deploy the VMs only on StorPool backed SYSTEM datastores, set SCHED_DS_REQUIREMENTS in the VM template:

SCHED_DS_REQUIREMENTS="TM_MAD=storpool"

Using sunstone, it is located in “Update VM template” -> “Scheduling” -> “Placement” -> “Datastore Requirements”. Set the “Expression” to TM_MAD=storpool.

Using Cluster definition

Define an OpenNebula cluster selecting only StorPool backed SYSTEM and IMAGE datastores. Keep in mind that in this case the images used in the VM templates must be on the datastore selected in the cluster!