Managing replication state for deleted objects

1. A system, comprising:

at least one processor;

at least one memory;

an operations log, stored in the at least one memory, configured to store a plurality of log operation entries for storage operations targeting data objects in a first object data store;

a delete function, stored in the at least one memory for execution by the at least one processor, configured to add a delete marker to the operations log to render a data object stored in the first object data store inaccessible to a client application;

a replication state machine, stored in the at least one memory for execution by the at least one processor, configured to:

identify a replication state of the data object; and

store the replication state of the data object in the operations log; and

a garbage collector, stored in the at least one memory for execution by the at least one processor, configured to:

query, from the operations log, the replication state of the data object, wherein the replication state indicates that a replication of the data object from the first object data store to a second object data store of a data storage system is incomplete;

determine that the replication state of the data object has changed to indicate that the replication is complete; and

responsive to the replication state indicating that the replication is complete, physically delete the data object from the first object data store.

2. The system of claim 1 , wherein the replication state machine is further configured to set, responsive to the first object data store storing the data object, the replication state of the data object to a null value.

3. The system of claim 1 , wherein the replication state machine is further configured to:

monitor a replication manager for the replication manager to add the data object to a replication queue; and

set, responsive to the replication manager adding the data object to the replication queue, the replication state to a replication pending value.

4. The system of claim 1 , wherein the replication state machine is further configured to:

monitor a replication manager for the replication manager to complete replication of the data object; and

set, responsive to the replication manager completing replication of the data object, the replication state to a successful replication value.

5. The system of claim 1 , wherein the replication state machine is further configured to:

monitor a replication manager for the replication manager to identify a replication error for the data object; and

set, responsive to the replication manager identifying a replication error for the data object, the replication state to a replication failed value.

6. The system of claim 1 , wherein the replication state machine is further configured to store, in the operations log, a replication state identifier for the replication state of the data object in an object creation entry for the data object.

7. The system of claim 1 , further comprising:

a replication manager, stored in the at least one memory for execution by the at least one processor, configured to:

access the data object using a replication read command for data objects rendered inaccessible to the client application; and

replicate the data object from the first object data store to the second object data store of the data storage system.

8. The system of claim 7 , wherein the replication read command is configured to detect a delete marker and a preceding write operation without a corresponding replication state identifier indicating successful replication.

9. The system of claim 1 , wherein the garbage collector is further configured to:

initiate garbage collection for data objects rendered inaccessible to the client application and stored in the first object data store; and

delay, responsive to querying the replication state of the data object indicating replication is not complete, physical deletion of the data object from the first object data store.

10. A computer-implemented method, comprising:

determining to delete a data object stored in a data storage system, wherein the data storage system includes an operations log configured to store a plurality of log operation entries for storage operations targeting data objects in a first object data store;

rendering the data object inaccessible to a client application by adding a delete marker to the operations log;

querying, from the operations log, a replication state of the data object, wherein the replication state indicates that a replication of the data object from the first object data store to a second object data store of the data storage system is incomplete;

determining that the replication state of the data object has changed to indicate that the replication is complete; and

responsive to the replication state indicating that the replication is complete, physically deleting the data object from the first object data store.

11. The computer-implemented method of claim 10 , further comprising:

storing the data object in the first object data store; and

setting, responsive to storing the data object in the first object data store, the replication state of the data object to a null value.

12. The computer-implemented method of claim 10 , further comprising:

monitoring a replication queue for an addition of the data object to the replication queue; and

setting, responsive to adding the data object to the replication queue, the replication state to a replication pending value.

13. The computer-implemented method of claim 10 , further comprising:

monitoring for complete replication of the data object; and

setting, responsive to complete replication of the data object, the replication state to a successful replication value.

14. The computer-implemented method of claim 10 , further comprising:

monitoring for a replication error for the data object; and

setting, responsive to identification of the replication error for the data object, the replication state to a replication failed value.

15. The computer-implemented method of claim 10 , further comprising:

storing, in the operations log, a replication state identifier for the replication state of the data object in an object creation entry for the data object.

16. The computer-implemented method of claim 10 , further comprising:

accessing the data object using a replication read command for data objects rendered inaccessible to the client application.

17. The computer-implemented method of claim 16 , further comprising:

detecting, responsive to the replication read command accessing the data object, a delete marker and a preceding write operation marker without a corresponding replication state identifier indicating successful replication; and

replicating, responsive to detecting the delete marker and the preceding write operation marker without a corresponding replication state identifier indicating successful replication, the data object to the second object data store.

18. The computer-implemented method of claim 10 , further comprising:

initiating garbage collection for data objects rendered inaccessible to the client application and stored in the first object data store; and

delaying, responsive to querying the replication state of the data object indicating replication is not complete, physical deletion of the data object from the first object data store.

19. The computer-implemented method of claim 18 , further comprising:

completing, responsive to querying the replication state of the data object indicating replication is complete, garbage collection for data objects rendered inaccessible to the client application and stored in the first object data store.

20. A system, comprising:

a first object data store configured to store a data object;

an operations log configured to store a plurality of log operation entries for storage operations targeting data objects in the first object data store;

means for determining to delete the data object stored in the first object data store;

means for rendering the data object inaccessible to a client application by adding a delete marker to the operations log;

means for querying, from the operations log, a replication state of the data object, wherein the replication state indicates that a replication of the data object from the first object data store to a second object data store is incomplete;

means for determining that the replication state of the data object has changed to indicate that the replication is complete; and

means for, responsive to the replication state indicating that the replication is complete, physically deleting the data object from the first object data store.