only consider OSDs with sufficient capacity for assignment of new folders

xtreemfs_client/dataDistribution.py

@@ -186,6 +186,7 @@ class DataDistribution(object):
         # find out which folders are not assigned yet
         new_folders = []
         for a_folder in folders:
+            # TODO adding folders to OSDs might violate their capacity
             containing_osd = self.get_containing_osd(a_folder.id)
             if containing_osd is not None:
                 containing_osd.add_folder(a_folder.id, a_folder.size)
@@ -199,7 +200,7 @@ class DataDistribution(object):
         # this information must be returned
         osds_for_new_folders = []
 
-        # totally random OSD assignment, even ignoring OSD capacities
+        # totally random OSD assignment, ignoring OSD capacities
         # (might lead to I/O errors when too many groups are assigned to an OSD)
         if random_osd_assignment and ignore_osd_capacities and not ignore_folder_sizes:
             if debug:
@@ -216,10 +217,7 @@ class DataDistribution(object):
             if debug:
                 print("using random osd assignment, respecting osd capacities")
             for a_folder in new_folders:
-                suitable_osds = []  # list of OSDs with enough capacity
-                for one_osd in self.OSDs.values():
-                    if one_osd.capacity - one_osd.total_folder_size - a_folder.size >= 0:
-                        suitable_osds.append(one_osd)
+                suitable_osds = self.get_suitable_osds(a_folder.size)  # list of OSDs with enough capacity
                 suitable_random_osd = random.choice(suitable_osds)
                 suitable_random_osd.add_folder(a_folder.id, a_folder.size)
                 osds_for_new_folders.append((a_folder.id,
@@ -426,6 +424,18 @@ class DataDistribution(object):
 
         return movements
 
+    def get_suitable_osds(self, folder_size):
+        """
+        create a list of OSDs with at least folder_size free capacity.
+        :return:
+        """
+        suitable_osds = []
+        for one_osd in self.OSDs.values():
+            if one_osd.capacity - one_osd.total_folder_size - folder_size >= 0:
+                suitable_osds.append(one_osd)
+
+        return suitable_osds
+
     def get_lpt_osd(self, folder_size):
         """
         calculate the processing time of all OSDs, using the sum of their current total_folder_size and folder_size.
@@ -433,7 +443,7 @@ class DataDistribution(object):
         """
         best_processing_time = None
         best_processing_time_osd = -1
-        for one_osd in self.OSDs.values():
+        for one_osd in self.get_suitable_osds(folder_size):
             processing_time = (one_osd.total_folder_size + folder_size) / one_osd.bandwidth
             if (best_processing_time is None) or processing_time < best_processing_time_osd:
                 best_processing_time = one_osd