It should be possible to remove Swift Accounts after their tenants have been deleted

Bug #1194540 reported by Julian Fischer
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
OpenStack Identity (keystone)
Invalid
Undecided
Unassigned
OpenStack Object Storage (swift)
Won't Fix
Undecided
Unassigned

Bug Description

Consider the following scenario:
Create a tenant, create a user, create a directory, upload a file, delete the user, delete the tenant.

Now it makes sense to send DELETE to the swift account before deleting the tenant.
However, one might forget it or an application error could occur.

So it could be imaginable that there are Swift Accounts whose tenants are gone and nobody remembers their tenant id.
In this case all related data in swift is inaccessible.

This should not be possible.

Possible solutions:
a) Make it possible to retrieve a list of swift accounts -> A script could be used to compare with keystone tenants and check for orphan swift accounts.

b) Create a keystone callback / hook that notifies Swift to mark accounts as deleted once their corresponding keystone tenants have been deleted.
This feature should be optional so that swift operators can either activate or deactivate it.

Revision history for this message
Kun Huang (academicgareth) wrote :

I think this is the job or client applications or keystone. Let's see keystone developer's view on the keystone callback

Revision history for this message
Dolph Mathews (dolph) wrote :

As of havana, keystone emits notifications on project (tenant) deletion for exactly this use case :)

BP: https://blueprints.launchpad.net/keystone/+spec/notifications

Docs: http://docs.openstack.org/developer/keystone/event_notifications.html

Changed in keystone:
status: New → Invalid
Revision history for this message
Kun Huang (academicgareth) wrote :

Thanks Dolph :)

Revision history for this message
Kun Huang (academicgareth) wrote :

But swift doesn't use any mq now.

Revision history for this message
Dolph Mathews (dolph) wrote :

Keystone got on the mq just to provide a solution to this issue

Revision history for this message
John Dickinson (notmyname) wrote :

This is normally handled by utilization/billing tools that walk the drives to see what's there and report on usage in the cluster.

Changed in swift:
status: New → Won't Fix
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.