create s3 bucket

aws-iam/destroy.py

@@ -26,6 +26,26 @@ def destroy(target):
         # if there was an exception, then the account didn't have an alias
         pass
 
+    response = s3 = boto.connect_s3()
+    bucket = s3.get_bucket(target)
+    keys = bucket.list()
+    bucket.delete_keys([k.name for k in keys])
+
+    ### The delete_bucket is intentionally commented out because the
+    ### delete behavior is surprising relative to other delete
+    ### operations in the API. Creating a bucket with the same name as
+    ### one that has been deleted may not even be possible in some
+    ### cases. Typically it takes a few hours for the bucket names to
+    ### be cleared so in practice it is often possible to create a new
+    ### bucket with the same name. Read more at:
+    ###    http://psykidellic.github.io/posts/amazon-s3---careful-with-bucket-names/
+    ###    https://forums.aws.amazon.com/thread.jspa?threadID=37532
+    ### Interestingly, create_bucket does not behave as expected,
+    ### either; if a bucket already exists, the call to create_bucket
+    ### does not raise an exception about a duplicate
+    ### already existing.
+    # s3.delete_bucket(target)
+
     for category in ['instructors','students']:
         group = target + '-' + category
         response = iam.get_group(group)

aws-iam/provision.py

@@ -57,6 +57,9 @@ def create_iam_users(target, category):
     for r in resources:
         response = apply_policy(group, category + '-' + r)
 
+    s3 = boto.connect_s3()
+    bucket = s3.get_bucket(target)
+
     data = []
     for e in email_addresses:
         response = iam.create_user(e)
@@ -89,6 +92,8 @@ def save_credentials(target, category, creds):
 
 def provision(target):
     create_signin_url(target)
+    s3 = boto.connect_s3()
+    bucket = s3.create_bucket(target)
     for category in ['instructors','students']:
         creds = create_iam_users(target, category)
         save_credentials(target, category, creds)