Make IPResolver run the update in a separate thread

A separate thread will cache the results of parsing the range
file, and the IPResolver will hit the cache instead of blocking while recomputing the
ranges everytime. The thread updates every 600s, and retry every 60s on
failures.

util/ipresolver/__init__.py

@@ -1,8 +1,10 @@
 import logging
 import json
+import time
 
 from collections import namedtuple, defaultdict
 
+from threading import Thread, Lock
 from abc import ABCMeta, abstractmethod
 from six import add_metaclass
 from cachetools import ttl_cache, lru_cache
@@ -19,6 +21,11 @@ ResolvedLocation = namedtuple('ResolvedLocation', ['provider', 'region', 'servic
 logger = logging.getLogger(__name__)
 
 _DATA_FILES = {'aws-ip-ranges.json': 'https://ip-ranges.amazonaws.com/ip-ranges.json'}
+_UPDATE_INTERVAL = 600
+_FAILED_UPDATE_RETRY_SECS = 60
+
+CACHE = {}
+CACHE_LOCK = Lock()
 
 def update_resolver_datafiles():
   """ Performs an update of the data file(s) used by the IP Resolver. """
@@ -33,6 +40,20 @@ def update_resolver_datafiles():
       f.write(response.text)
       logger.debug('Successfully wrote %s', filename)
 
+def _get_aws_ip_ranges():
+  try:
+    with open('util/ipresolver/aws-ip-ranges.json', 'r') as f:
+      return json.loads(f.read())
+  except IOError:
+    logger.exception('Could not load AWS IP Ranges')
+    return None
+  except ValueError:
+    logger.exception('Could not load AWS IP Ranges')
+    return None
+  except TypeError:
+    logger.exception('Could not load AWS IP Ranges')
+    return None
+
 
 @add_metaclass(ABCMeta)
 class IPResolverInterface(object):
@@ -62,6 +83,9 @@ class IPResolver(IPResolverInterface):
   def __init__(self, app):
     self.app = app
     self.geoip_db = geoip2.database.Reader('util/ipresolver/GeoLite2-Country.mmdb')
+    self._worker = _UpdateIPRange(_UPDATE_INTERVAL)
+    if not app.config.get('TESTING', False):
+      self._worker.start()
 
   @ttl_cache(maxsize=100, ttl=600)
   def is_ip_possible_threat(self, ip_address):
@@ -106,28 +130,23 @@ class IPResolver(IPResolverInterface):
 
     return location_function(ip_address)
 
-  def _get_aws_ip_ranges(self):
-    try:
-      with open('util/ipresolver/aws-ip-ranges.json', 'r') as f:
-        return json.loads(f.read())
-    except IOError:
-      logger.exception('Could not load AWS IP Ranges')
-      return None
-    except ValueError:
-      logger.exception('Could not load AWS IP Ranges')
-      return None
-    except TypeError:
-      logger.exception('Could not load AWS IP Ranges')
-      return None
-
-  @ttl_cache(maxsize=1, ttl=600)
   def _get_location_function(self):
-    aws_ip_range_json = self._get_aws_ip_ranges()
-    if aws_ip_range_json is None:
+    try:
+      cache = CACHE
+      sync_token = cache.get('sync_token', None)
+      if sync_token is None:
+        logger.debug('The aws ip range has not been cached from %s', _DATA_FILES['aws-ip-ranges.json'])
+        return None
+
+      all_amazon = cache['all_amazon']
+      regions = cache['regions']
+    except KeyError:
+      logger.exception('Got exception trying to hit aws ip range cache')
+      return None
+    except Exception:
+      logger.exception('Got exception trying to hit aws ip range cache')
       return None
 
-    sync_token = aws_ip_range_json['syncToken']
-    all_amazon, regions = IPResolver._parse_amazon_ranges(aws_ip_range_json)
     return IPResolver._build_location_function(sync_token, all_amazon, regions, self.geoip_db)
 
   @staticmethod
@@ -175,3 +194,30 @@ class IPResolver(IPResolverInterface):
       regions[region].add(cidr)
 
     return all_amazon, regions
+
+
+class _UpdateIPRange(Thread):
+  """Helper class that uses a thread to loads the IP ranges from Amazon"""
+  def  __init__(self, interval):
+    Thread.__init__(self)
+    self.interval = interval
+
+  def run(self):
+    while True:
+      try:
+        logger.debug('Updating aws ip range from "%s"', 'util/ipresolver/aws-ip-ranges.json')
+        aws_ip_range_json = _get_aws_ip_ranges()
+      except:
+        logger.exception('Failed trying to update aws ip range')
+        time.sleep(_FAILED_UPDATE_RETRY_SECS)
+        break
+
+      sync_token = aws_ip_range_json['syncToken']
+      all_amazon, regions = IPResolver._parse_amazon_ranges(aws_ip_range_json)
+
+      with CACHE_LOCK:
+        CACHE['sync_token'] = sync_token
+        CACHE['all_amazon'] = all_amazon
+        CACHE['regions'] = regions
+
+      time.sleep(self.interval)