Cache improvements and 15 min TTL

Author: yuanzhou

src/app.py

@@ -63,9 +63,6 @@
 requests.packages.urllib3.disable_warnings(category = InsecureRequestWarning)
 
 # For performance improvement and not overloading the server, especially helpful during Elasticsearch index/reindex
-# We use the cache as long as it exists
-# Only clear the expired one based on TTL setting passively upon lookup rather than the actual expiration time
-# This approach takes advantage of the cache and also prevents from memory overflow
 entity_cache = {}
 
 
@@ -4009,20 +4006,18 @@ def query_target_entity(id, user_token):
     current_datetime = datetime.now()
     current_timestamp = int(round(current_datetime.timestamp()))
 
-    # Use the cached data as long as it exists, no matter if it's expired or not
-    # Otherwise make a fresh request and add to the cache pool
-    # But we'll clear the expired cache based on TTL setting passively
-    if id in entity_cache:
-        entity_dict = entity_cache[id]['entity']
+    # Check if the cached entity is found and still valid based on TTL upon request, delete if expired
+    if (id in entity_cache) and (current_timestamp > entity_cache[id]['created_timestamp'] + SchemaConstants.REQUEST_CACHE_TTL):
+        del entity_cache[id]
 
-        # We can still return the cache entity even if it's expired.
-        # Just need to clear the expired data from cache so the next lookup makes a new call and stores the new data to cache
-        if current_timestamp <= entity_cache[id]['created_timestamp'] + SchemaConstants.REQUEST_CACHE_TTL:
-            logger.info(f'Using the entity data of {id} from cache at time {current_datetime}')
-        else:
-            logger.info(f'Using the entity data of {id} from cache at time {current_datetime} then clear it based on TTL setting')
+        logger.info(f'Deleted the expired entity cache of {id} at time {current_datetime}')
+
+    # Use the cached data if found and still valid
+    # Otherwise, make a fresh query and add to cache
+    if (id in entity_cache) and (current_timestamp <= entity_cache[id]['created_timestamp'] + SchemaConstants.REQUEST_CACHE_TTL):
+        entity_dict = entity_cache[id]['entity']
 
-            del entity_cache[id]
+        logger.info(f'Using the valid cache data of entity {id} at time {current_datetime}')
     else:
         logger.info(f'Cache not found or expired. Making a new query to retrieve {id} at time {current_datetime}')
 

src/schema/schema_constants.py

@@ -1,6 +1,6 @@
 class SchemaConstants(object):
-    # Expire the request cache after the time-to-live (seconds), default 4 hours
-    REQUEST_CACHE_TTL = 14400
+    # Expire the request cache after the time-to-live (seconds), default 15 minutes
+    REQUEST_CACHE_TTL = 900
 
     # Constants used by validators
     INGEST_API_APP = 'ingest-api'

src/schema/schema_manager.py

@@ -32,9 +32,6 @@
 _neo4j_driver = None
 
 # For handling cached requests to uuid-api and external static resources (github raw yaml files)
-# We use the cache as long as it exists
-# Only clear the expired one based on TTL setting passively upon lookup rather than the actual expiration time
-# This approach takes advantage of the cache and also prevents from memory overflow
 request_cache = {}
 
 
@@ -1557,18 +1554,18 @@ def make_request_get(target_url, internal_token_used = False):
     current_datetime = datetime.now()
     current_timestamp = int(round(current_datetime.timestamp()))
 
-    # Use the cached data as long as it exists, no matter if it's expired or not
-    # Otherwise make a fresh request and add to the cache pool
-    # But we'll clear the expired cache based on TTL setting passively
-    if target_url in request_cache:
-        response = request_cache[target_url]['response']
+    # Check if the cached response is found and still valid based on TTL upon request, delete if expired
+    if (target_url in request_cache) and (current_timestamp > request_cache[target_url]['created_timestamp'] + SchemaConstants.REQUEST_CACHE_TTL):
+        del request_cache[target_url]
 
-        if current_timestamp <= request_cache[target_url]['created_timestamp'] + SchemaConstants.REQUEST_CACHE_TTL:
-            logger.info(f'Using the cached HTTP response of GET {target_url} at time {current_datetime}')
-        else:
-            logger.info(f'Using the cached HTTP response of GET {target_url} at time {current_datetime} then clear it based on TTL setting')
+        logger.info(f'Deleted the expired HTTP response cache of GET {target_url} at time {current_datetime}')
+
+    # Use the cached data if found and still valid
+    # Otherwise, make a fresh query and add to cache
+    if (target_url in request_cache) and (current_timestamp <= request_cache[target_url]['created_timestamp'] + SchemaConstants.REQUEST_CACHE_TTL):
+        response = request_cache[target_url]['response']
 
-            del request_cache[target_url]
+        logger.info(f'Using the cached HTTP response of GET {target_url} at time {current_datetime}')
     else:
         logger.info(f'Cache not found or expired. Making a new HTTP request of GET {target_url} at time {current_datetime}')