If your organization uses a block storage solution, such as SolidFire, for root disks for your virtual machines, it may have requested that the option to boot from a compute server drive be disabled in the Dashboard during the build process for your Metacloud deployment. This would prevent Metacloud Dashboard users from launching instances from compute server drives.
However, there is no way to disable this option in the API or the command line interface (CLI). If you use one of these methods to launch instances without specifying a block-device root disk, you will, by default, boot instances from compute server drives. As instances accumulate, this could result in instances eventually using up the ephemeral storage on your Metacloud Hypervisors (MHVs).
Specifying a block-device Option
When launching a new instance with the CLI, make sure to use the
--block-device parameter to boot the instance from volume, as in the following CLI example:
nova boot newInstance \
--flavor m1.small \
--security-groups default \
--nic net-id=<UUID> \
For more information about launching instances with the CLI, see Launching an Instance from a Volume.
If you are using the API, include the
block-device-mapping-v2 parameter in the create request body to boot a server from a volume. See OpenStack Docs: Compute API for more information.
NOTE: To find out whether a given instance was booted from volume, see Determining the Boot Source of an Instance.