Overview
(BDM) gives precise control over how storage is presented to an instance at creation time. Use BDM to boot from a persistent volume, attach pre-existing data volumes, or configure the ephemeral scratch disk included in certain flavors.Prerequisites
- An active Xloud project with
memberoradminrole - Xloud Block Storage volumes (for volume-backed launches) or images (for image-backed launches)
- CLI users: Xloud CLI configured and credentials sourced
BDM Attributes Reference
| Attribute | Description | Values |
|---|---|---|
source_type | Origin of the block device | image, volume, snapshot, blank |
destination_type | Where the device is stored | volume (persistent) or local (ephemeral) |
boot_index | Boot priority (0 = primary boot device) | Integer; -1 for non-bootable |
uuid | ID of the source image, volume, or snapshot | UUID string |
volume_size | Size of the created volume in GB | Integer |
delete_on_termination | Whether to delete the volume when the instance is deleted | true or false |
Boot from Volume
Booting from a persistent volume is recommended for production workloads. The root disk survives instance termination whendelete_on_termination is set to false, enabling
instance snapshots, backups, and re-attachment to replacement instances.
- Dashboard
- CLI
Configure the boot source
On the Source tab:
- Set Select Boot Source to
Image. - Set Create New Volume to
Yes. - Set the Volume Size to at least the image’s minimum disk requirement.
- Set Delete Volume on Instance Delete based on your retention policy.
| Setting | Recommendation |
|---|---|
| Create New Volume | Yes — creates a persistent boot volume |
| Volume Size | ≥ image minimum disk size (shown in image metadata) |
| Delete Volume on Instance Delete | No for production; Yes for ephemeral workloads |
Attach Additional Volumes at Launch
Attach pre-existing data volumes when creating an instance to have formatted storage available immediately after boot.- Dashboard
- CLI
The Dashboard does not expose the full BDM
--block-device API at launch time for
additional volumes. Attach pre-existing volumes after launch:Navigate to Project → Volumes → Volumes, find the target volume, and select
Manage Attachments → Attach to Instance.Ephemeral Disk Configuration
Ephemeral disks are temporary storage included in certain flavors. They exist only for the lifetime of the instance and are reset or removed on rebuild, resize, or termination.| Property | Value |
|---|---|
| Device path | /dev/vdb (first ephemeral); /dev/vdc if a persistent volume is also attached |
| Persistence | Deleted when the instance is terminated or rebuilt |
| Use case | Scratch space, temporary build artifacts, swap |
| Flavor control | Ephemeral disk size is defined by the flavor, not configurable at launch |
BDM Scenarios Summary
Production workload — persistent root disk
Production workload — persistent root disk
Recommended configuration: Boot from volume with
delete_on_termination=false.
The root disk survives instance deletion, enabling backup, snapshot, and re-attachment
workflows.CLI: launch with persistent boot volume
Stateless workload — ephemeral root disk
Stateless workload — ephemeral root disk
Recommended configuration: Boot from image with
delete_on_termination=true.
Suitable for auto-scaling groups, CI runners, or stateless app tiers where instances
are disposable.CLI: launch with ephemeral root (deleted on terminate)
Database tier — separate data volume
Database tier — separate data volume
Recommended configuration: Boot from volume (OS) + separate data volume (database
files). This separates the OS lifecycle from the data lifecycle — terminate and
rebuild the OS volume while the data volume persists.
CLI: launch with separate data volume
Next Steps
Xloud Block Storage
Create, manage, and snapshot persistent block storage volumes
Launch an Instance
Full instance launch walkthrough including boot source and volume configuration
Rescue an Instance
Mount and repair volumes when an instance fails to boot
Compute User Guide
Overview of all compute operations and instance lifecycle management