Is there an existing issue for this?

• There is no existing issue for this feature

What are you currently unable to do

Related to #1837

Problem statement: When instance(s) have many thounsands of snapshots, it's infeasibly slow for them to be deleted since they are removed one-by one (a full round trip) for each deletion.

Proposed solution: Refactor the deletion of snapshots logic so that multiple expired snaoshots may be deleted in bulk, rather than serially. Perhaps Lean on the Automatic snapshot names

When a snapshot policy generates many (thousands) of snapshots, for example with a regular cadence schedule , it can take many

There are a few options to delete snaphots within Incus:

• Change the attached snapshot expiry to something less than before- incus will then gradually remove snapshots
• Remove (delete) the instance, after which incus will start deleting associated snapshots one-by-one

What the user must not do, if my understanding is correct, is manually zfs delete any Incus deleted snapshots- whilst this would be very fast, Inus would loose state in it's sqlite style database, causing a confused state.

What do you think would need to be added

Proposed solution: Refactor the deletion of snapshots logic so that multiple expired snaoshots may be deleted in bulk, rather than serially. Perhaps Lean on the Automatic snapshot names

There's two scenarios I can think of for this:

• Primary use case: When Incus has generated a large volume of snapshots (> 10,000) (e.g.) and wants to keep the instance in situ, but delete all the snapshots.
• Secondary use case: When an incus instance has been deleted, to expediate the deletion of the snapshots