Comment 8 for bug 1503161

Conditions to reproduce:
1) Create geo-replicated cluster with 2 regions, 3 replicas.
2) Upload a test object and find out its placement with swift-get-node.
3) Delete the object.
4) Connect to the API endpoint of the region that is expected to have 1 replica and re-upload the object.
5) Before the replication to the other region is completed, delete the object via the API of the region that has only 1 replica.
Result: 404 Not Found. The resource could not be found.

Why is this problematic?
a) From a cloud customer point of view, the object has been deleted and will never reappear.
b) From a cloud provider point of view, the object has been deleted. When the replicator gets to this object it will notice it has been deleted will not replicate it to the other region.
c) Cloud customers are getting the same error message they get for when they try to delete an object that does not exist.

We are interested in fixing this and would like to hear the opinion of more experienced developers on what the best path would be.