Delete a volume about flocker HOT 12 CLOSED

If a local volume is deleted what happens to volumes on remote nodes?

My opinion is that remote copies of volumes should aim to track the masters - so that if the master is deleted all remote copies are deleted as well. This would avoid confusing conflicts when a volume is deleted and created again with the same name.

from flocker.

That makes sense. On the other hand: imagine a host gets taken over by an attacker. You don't want data to be automatically destroyed on other hosts if one host is compromised. Maybe when master deletes it the remotes should rename it? Or maybe ZFS-level names should be UUIDs, and then just because name chosen by human is reused there's still no conflicts (but that may result in confusing UI).

In any case this is out of scope for this issue, which is just the low-level single node delete operation.

from flocker.

I would be in favour of renames. One of the lessons I'd like to take away from HybridCluster is that having opaque data structures does not make administration easy. I'm referring here to the unix timestamp snapshot names. It's easy to say that the tools should be able to perform all the tasks you would want to perform but that isn't realistic for a constrained developer team.

Either way I'll make a story.

from flocker.

Actually, probably not strictly necessary for first release.

from flocker.

Or, might turn out to be when we do push back and forth between two nodes, but not sure yet so not going to start on this.

from flocker.

Jean-Paul had some vague notion of a flocker-clean command...

from flocker.

The basic logic for deleting a ZFS IFilesystem instance would be:

1. Delete all snapshots that do not have clones off of them.
2. If the filesystem has an origin snapshot, and that snapshot has no other clones, delete the snapshot.
3. Delete the filesystem.

This will result in both clones and non-clones cleaning up all references that they own. Clone logic may be something to spin off into different issue.

from flocker.

The logic above would probably be exposed via flocker-volume only, for now:

$ flocker-volume delete <owner uuid> <name>

from flocker.

is this for user-facing functionality? We don't really have docs for flocker-volume yet - so we also need to add some if we're going to expose user-facing functionality this way.

from flocker.

Quite possibly, it's not clear to me yet how deletion would work on level of flocker-deploy (merely doing a handoff shouldn't necessarily delete origin, e.g.).

from flocker.

I guess probably not user facing now; will file issue for the the user-facing command separately.

from flocker.

We are moving our development planning to JIRA. This issue is now being tracked at https://clusterhq.atlassian.net/browse/FLOC-14. You are welcome to file additional issues in GitHub if that's easier for you.

from flocker.