add rebalance_lpt method and test

tests/test_dataDistribution.py

@@ -188,6 +188,23 @@ class TestDataDistribution(unittest.TestCase):
         average = (sum(folder_sizes) * num_folders) / (num_osds * len(osd_capacities))
         self.assertEqual(average, distribution.get_average_total_folder_size())
 
+    def test_rebalance_lpt(self):
+        folder_sizes = [1]
+        num_folders = 8
+        osd_capacities = [10]
+        num_osds = 4
+
+        distribution = dataDistribution.DataDistribution()
+        distribution.add_osd_list(create_test_osd_list(num_osds, osd_capacities))
+        distribution.add_folders(create_test_folder_list(num_folders, folder_sizes), random_osd_assignment=True)
+        distribution.rebalance_lpt()
+
+        osds = distribution.get_osd_list()
+        total_folder_sizes = list(map(lambda x: distribution.OSDs[x].total_folder_size, osds))
+        # we should obtain a perfectly balanced distribution
+        self.assertEqual(min(total_folder_sizes), max(total_folder_sizes))
+
+
 def create_test_osd_list(num_osds, osd_capacities):
     test_osds = []
     for i in range(0, num_osds):

xtreemfs_client/dataDistribution.py

@@ -231,6 +231,27 @@ class DataDistribution(object):
                                          least_used_osd.uuid))
         return osds_for_new_folders
 
+    def rebalance_lpt(self, rebalance_factor=1, osd_information=None, capacity=''):
+        movements = {}
+        folders_to_be_reassigned = []
+        reassignment_limit = self.get_average_total_folder_size() * rebalance_factor
+
+        # for each OSD, remove the smallest folder until its total_folder_size does not exceed the reassignment_limit
+        for osd in self.OSDs.values():
+            while osd.total_folder_size > reassignment_limit:
+                folder_id, folder_size = osd.get_smallest_folder()
+                folders_to_be_reassigned.append(folder.Folder(folder_id, folder_size, None))
+                movements[folder_id] = osd.uuid
+                osd.remove_folder(folder_id)
+
+        new_assignments = self.add_folders(folders_to_be_reassigned,
+                                           osd_information=osd_information, ratio_parameter=capacity)
+
+        for folder_id, target in new_assignments:
+            movements[folder_id] = (movements[folder_id], target)
+
+        return movements
+
     def update_folder(self, folder, size):
         """
         updates the size of a given folder