From cd967749de2e15b1e058598e6983453542fc0cba Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Sybren=20A=2E=20St=C3=BCvel?= <sybren@stuvel.eu>
Date: Thu, 21 Jul 2016 14:22:29 +0200
Subject: [PATCH] Revert "Use threading when generating links & renaming files
 at GCS."

This reverts commit 2fc89a728de6e8b02800fcae69dd09b62a0c1293.
---
 pillar/application/modules/file_storage.py | 28 ++++------------------
 pillar/application/utils/gcs.py            | 23 +++---------------
 2 files changed, 8 insertions(+), 43 deletions(-)

diff --git a/pillar/application/modules/file_storage.py b/pillar/application/modules/file_storage.py
index 256dbadf..08b37ba9 100644
--- a/pillar/application/modules/file_storage.py
+++ b/pillar/application/modules/file_storage.py
@@ -1,5 +1,4 @@
 import datetime
-import functools
 import logging
 import mimetypes
 import os
@@ -7,7 +6,6 @@ import tempfile
 import uuid
 import io
 from hashlib import md5
-import threading
 
 import bson.tz_util
 import eve.utils
@@ -378,28 +376,12 @@ def _generate_all_links(response, now):
     project_id = str(
         response['project']) if 'project' in response else None  # TODO: add project id to all files
     backend = response['backend']
+    response['link'] = generate_link(backend, response['file_path'], project_id)
 
-    def generate_link_for(store_in):
-        log.debug('    generating link for %s', store_in['file_path'])
-        store_in['link'] = generate_link(backend, store_in['file_path'], project_id)
-
-    # Generate links in parallel using the thread pool.
-    # TODO: use asyncio for this, once we're on Python 3.5, to limit the total number
-    # of concurrent threads.
-    threads = []
-    thread = threading.Thread(target=functools.partial(generate_link_for, response))
-    threads.append(thread)
-
-    variations = response.get('variations', ())
-    for variation in variations:
-        thread = threading.Thread(target=functools.partial(generate_link_for, variation))
-        threads.append(thread)
-
-    # Start & wait for all the parallel tasks.
-    for thread in threads:
-        thread.start()
-    for thread in threads:
-        thread.join(120)
+    variations = response.get('variations')
+    if variations:
+        for variation in variations:
+            variation['link'] = generate_link(backend, variation['file_path'], project_id)
 
     # Construct the new expiry datetime.
     validity_secs = current_app.config['FILE_LINK_VALIDITY'][backend]
diff --git a/pillar/application/utils/gcs.py b/pillar/application/utils/gcs.py
index 83e5f074..23b336c7 100644
--- a/pillar/application/utils/gcs.py
+++ b/pillar/application/utils/gcs.py
@@ -1,9 +1,7 @@
-import functools
 import os
 import time
 import datetime
 import logging
-import threading
 
 from bson import ObjectId
 from gcloud.storage.client import Client
@@ -161,8 +159,6 @@ def update_file_name(node):
     if node['properties'].get('status', '') == 'processing':
         return
 
-    files_collection = current_app.data.driver.db['files']
-
     def _format_name(name, override_ext, size=None, map_type=u''):
         root, _ = os.path.splitext(name)
         size = u'-{}'.format(size) if size else u''
@@ -170,6 +166,7 @@ def update_file_name(node):
         return u'{}{}{}{}'.format(root, size, map_type, override_ext)
 
     def _update_name(file_id, file_props):
+        files_collection = current_app.data.driver.db['files']
         file_doc = files_collection.find_one({'_id': ObjectId(file_id)})
 
         if file_doc is None or file_doc['backend'] != 'gcs':
@@ -196,26 +193,12 @@ def update_file_name(node):
                 continue
             storage.update_name(blob, name)
 
-    # Generate links in parallel using the thread pool.
-    threads = []
-    log.debug('Creating _update_name tasks')
-
     # Currently we search for 'file' and 'files' keys in the object properties.
     # This could become a bit more flexible and realy on a true reference of the
     # file object type from the schema.
     if 'file' in node['properties']:
-        thread = threading.Thread(target=functools.partial(
-            _update_name, node['properties']['file'], {}))
-        threads.append(thread)
+        _update_name(node['properties']['file'], {})
 
     if 'files' in node['properties']:
         for file_props in node['properties']['files']:
-            thread = threading.Thread(target=functools.partial(
-                _update_name, file_props['file'], file_props))
-            threads.append(thread)
-
-    # Start & wait for all the parallel tasks.
-    for thread in threads:
-        thread.start()
-    for thread in threads:
-        thread.join(120)
+            _update_name(file_props['file'], file_props)