object-reconstructor should delete fragments as soon as they are reverted

[Romain LE DISEZ]

When the object-reconstructor is trying to revert fragment N of a partition, if one fragment failed to be reverted, all fragments of that partition will stay on disk.

The reconstructor should delete all reverted fragments to free space as soon as possible.

# swift-object-replicator /etc/swift/object-server.conf -d disk-01-000 -p 238968 -v -o
object-replicator: Starting 13697
object-replicator: [worker 1/1 pid=13730] Running object replicator in script mode.
object-replicator: [worker 1/1 pid=13730] 172.16.0.196:6207/disk-00-007/238968 Unexpected response: ":ERROR: 500 'ERROR: With :UPDATES: 1 failures to 0 successes'"
object-replicator: [worker 1/1 pid=13730] 1/1 (100.00%) partitions replicated in 16.50s (0.06/sec, 0s remaining)
object-replicator: [worker 1/1 pid=13730] 2 successes, 0 failures
object-replicator: [worker 1/1 pid=13730] Object replication complete (once). (0.28 minutes)
object-replicator: Finished 13697
object-replicator: Exited 13697

# du -sh /srv/node/disk-01-000/objects/238968/
4.9G /srv/node/disk-01-000/objects/238968/

# find /srv/node/disk-01-000/objects/238968/ -type f -name '*.data' | wc -l
5171

# FIX THE BUG

# swift-object-replicator /etc/swift/object-server.conf -d disk-01-000 -p 238968 -v -o
object-replicator: Starting 14520
object-replicator: [worker 1/1 pid=14553] Running object replicator in script mode.
object-replicator: [worker 1/1 pid=14553] Removing partition: /srv/node/disk-01-000/objects/238968
object-replicator: [worker 1/1 pid=14553] 1/1 (100.00%) partitions replicated in 149.64s (0.01/sec, 0s remaining)
object-replicator: [worker 1/1 pid=14553] 3 successes, 0 failures
object-replicator: [worker 1/1 pid=14553] Object replication complete (once). (2.49 minutes)
object-replicator: Finished 14520
object-replicator: Exited 14520

# du -sh /srv/node/disk-01-000/objects/238968/
du: cannot access `/srv/node/disk-01-000/objects/238968/': No such file or directory