708c9b3dae
This is a *simple* delete strategy - if the root container thinks it is empty then it is OK to delete it. The root will not consider itself empty until (a) all objects have been deleted and (b) all shards have updated their usage (i.e. the sharders have passed over them), so there may be some delay between object deletes and the root being deletable. That is analogous to a normal container being undeletable because some delete updates dropped into an async pending queue. We can do better, but this allows us to delete at some point. Change-Id: I942b3b5a8295bb9957eddb15a149263a9037dd9a |
||
|---|---|---|
| .. | ||
| controllers | ||
| __init__.py | ||
| server.py | ||