Merge pull request #3 from mgoodfellow/delete-previous-backups

Delete previous backups

README.rst

@@ -61,7 +61,8 @@ file
           "Name": "MySQL Backup",
           "Command": "/home/bob/backups/backup-prep-script.sh",
           "Src": "/home/bob/backups/database/mysql_backup.sql",
-          "OutputPrefix": "main_db"
+          "OutputPrefix": "main_db",
+          "PreviousBackupsCount": 2
         },
         {
           "Name": "Websites Backup",
@@ -74,6 +75,9 @@ file
 If emails are not required, then omit the ``EMAIL_FROM`` and
 ``EMAIL_TO`` fields of the configuration file.
 
+If the ``PreviousBackupsCount`` is not set, then it will default to keeping
+1 previous backup. It can be set to 0, which will only keep the current backup.
+
 *Note*: When on Windows, it is better to pass the paths using forward
 slashes (/) as then escaping isn’t required (as with backslashes). The
 script will normalize the paths in these cases. However, when providing

S3Backup/config_loader.py

@@ -48,7 +48,6 @@ def config_setup(config_file):
 
     with open(config_file) as json_data_file:
         data = json.load(json_data_file)
-    print(data)
 
     failed = False
 
@@ -64,8 +63,6 @@ def config_setup(config_file):
         if optional_value in data:
             configuration[optional_value] = data[optional_value]
 
-    print(configuration)
-
     if failed:
         raise Exception('Missing keys from data. See log for details.')
 

S3Backup/plan.py

@@ -32,7 +32,7 @@
 from S3Backup import hash_file
 
 required_plan_values = ['Name', 'Src', 'OutputPrefix']
-optional_plan_values = ['Command']
+optional_plan_values = ['Command', 'PreviousBackupsCount']
 
 logger = logging.getLogger(name='Plan')
 
@@ -51,6 +51,13 @@ def __init__(self, raw_plan, configuration):
         self.name = raw_plan['Name']
         self.src = raw_plan['Src']
         self.command = None
+
+        if 'PreviousBackupsCount' in raw_plan:
+            self.previous_backups_count = int(raw_plan['PreviousBackupsCount'])
+        else:
+            self.previous_backups_count = 1
+
+        self.output_file_prefix = raw_plan['OutputPrefix']
         self.output_file = '%s_%s.zip' % (raw_plan['OutputPrefix'], time.strftime("%Y-%m-%d_%H-%M-%S"))
 
         self.new_hash = None
@@ -69,6 +76,7 @@ def run(self):
                 3) Perform hash check to see if there are any changes (which would require an upload)
                 4) Upload destination file to S3 bucket
                 5) Update hash file with new hash
+                6) Check if any previous backups need removing
         """
         logger.info('Running plan "%s"', self.name)
 
@@ -92,6 +100,9 @@ def run(self):
 
                 updated = True
 
+            # 6) Remove any previous backups if required
+            self.__clear_old_backups()
+
         finally:
             self.__cleanup()
 
@@ -138,6 +149,40 @@ def __zip_files(self):
 
         logger.info('Output file created')
 
+    def __clear_old_backups(self):
+        try:
+            conn = boto.s3.connect_to_region(
+                self.CONFIGURATION['AWS_REGION'],
+                aws_access_key_id=self.CONFIGURATION['AWS_KEY'],
+                aws_secret_access_key=self.CONFIGURATION['AWS_SECRET'])
+
+            bucket = conn.get_bucket(self.CONFIGURATION['AWS_BUCKET'])
+
+            backup_keys = []
+
+            for key in bucket.list(prefix=self.output_file_prefix):
+                backup_keys.append(key.name)
+
+            backup_keys.sort()
+
+            logger.info('There are %d previous backups', len(backup_keys))
+
+            max_backups = self.previous_backups_count + 1   # Because this is run after current backup uploaded
+
+            if len(backup_keys) > max_backups:
+                backups_to_remove = len(backup_keys) - max_backups
+                logger.info('Removing %d previous backups', backups_to_remove)
+
+                for i in range(backups_to_remove):
+                    logger.info('Removing previous backup: %s', backup_keys[i])
+                    bucket.delete_key(backup_keys[i])
+            else:
+                logger.info('No previous backups require removal')
+
+        except Exception, e:
+            logger.error('Failed to clear out previous backups from S3: %s', e)
+            raise
+
     def __upload(self):
         try:
             conn = boto.s3.connect_to_region(

config.json

@@ -11,7 +11,8 @@
       "Name": "MySQL Backup",
       "Command": "mysqldump -u bob -p password > mysql_backup.sql",
       "Src": "c:/mysql_backup.sql",
-      "OutputPrefix": "main_db"
+      "OutputPrefix": "main_db",
+      "PreviousBackupsCount": 2
     },
 	{
       "Name": "Website Backup",