Ceph (RBD) trash support

[hydro-b]

Description
Since Ceph Luminous (12.2.x) there is support for putting RBD images in a "trash" bin (namespace). Instead of directly deleting an image from a pool the image can be put in the "trash" instead.

Use case
By putting an IMAGE in the "trash" it provides the user / admin the possibility to undo unwanted delete actions without relying on backups. This might avoid accidental data loss in case someone accidentally deletes an IMAGE or a TEMPLATE including disk images.

Interface Changes
This change will need adjustments from drivers, Sunstone as well as CLI / API commands. An overview of IMAGES in "trash" as well as a way to definitively get rid of them.

Progress Status

• Branch created
• Code committed to development branch
• Testing - QA
• Documentation
• Release notes - resolved issues, compatibility, known issues
• Code committed to upstream release/hotfix branches
• Documentation committed to upstream release/hotfix branches

[xorel]

Merge to master:

• F #3147: Ceph - move to trash instead delete #3341
• https://github.com/OpenNebula/infra/pull/258
• https://github.com/OpenNebula/development/pull/627
• F OpenNebula/one#3147: Document CEPH_TRASH docs#627

[rsmontero]

Hi @hydro-b

Overview of the implementation

• New CEPH_TRASH attribute for the datastores and disk
• A disk includes CEPH_TRASH = yes all rbd rm operations are replaced by rbd trash move
• This attribute can be set per DISK to only move to trash that given disk, or globally at DS level
• When a disk contains snapshots all the snapshots are also moved to trash

Compared with your original request we are not implementing an OpenNebula side inspection of the CEPH_TRASH.

Let us know what do you think