diff --git a/llm_server/routes/v1/generate_stream.py b/llm_server/routes/v1/generate_stream.py
index 55fb6e4..6cd98c0 100644
--- a/llm_server/routes/v1/generate_stream.py
+++ b/llm_server/routes/v1/generate_stream.py
@@ -159,23 +159,10 @@ def do_stream(ws, model_name):
                                             }))
                                         except:
                                             # The has client closed the stream.
-                                            if request:
+                                            if response:
                                                 # Cancel the backend?
-                                                request.close()
-                                            end_time = time.time()
-                                            elapsed_time = end_time - start_time
-                                            log_prompt(ip=handler.client_ip,
-                                                       token=handler.token,
-                                                       prompt=input_prompt,
-                                                       response=generated_text,
-                                                       gen_time=elapsed_time,
-                                                       parameters=handler.parameters,
-                                                       headers=r_headers,
-                                                       backend_response_code=response_status_code,
-                                                       request_url=r_url,
-                                                       backend_url=handler.backend_url,
-                                                       response_tokens=None
-                                                       )
+                                                response.close()
+                                            # used to log here
                                             return
 
                                         message_num += 1
@@ -184,22 +171,9 @@ def do_stream(ws, model_name):
                             # If there is no more data, break the loop
                             if not chunk:
                                 break
-
-                    end_time = time.time()
-                    elapsed_time = end_time - start_time
-                    log_prompt(ip=handler.client_ip,
-                               token=handler.token,
-                               prompt=input_prompt,
-                               response=generated_text,
-                               gen_time=elapsed_time,
-                               parameters=handler.parameters,
-                               headers=r_headers,
-                               backend_response_code=response_status_code,
-                               request_url=r_url,
-                               backend_url=handler.backend_url,
-                               response_tokens=None,
-                               is_error=not response
-                               )
+                    if response:
+                        response.close()
+                    # used to log here
                 except:
                     traceback.print_exc()
                     generated_text = generated_text + '\n\n' + handler.handle_error('Encountered error while streaming.', 'exception')[0].json['results'][0]['text']
@@ -208,33 +182,19 @@ def do_stream(ws, model_name):
                         'message_num': message_num,
                         'text': generated_text
                     }))
-                    if request:
-                        request.close()
-                    log_prompt(ip=handler.client_ip,
-                               token=handler.token,
-                               prompt=input_prompt,
-                               response=generated_text,
-                               gen_time=None,
-                               parameters=handler.parameters,
-                               headers=r_headers,
-                               backend_response_code=response_status_code,
-                               request_url=r_url,
-                               backend_url=handler.backend_url,
-                               response_tokens=None,
-                               is_error=True
-                               )
-                    return
+                    # used to log here
                 finally:
                     # The worker incremented it, we'll decrement it.
                     decrement_ip_count(handler.client_ip, 'processing_ips')
                     decr_active_workers(handler.selected_model, handler.backend_url)
-                try:
-                    ws.send(json.dumps({
-                        'event': 'stream_end',
-                        'message_num': message_num
-                    }))
-                except:
-                    # The client closed the stream.
+                    try:
+                        ws.send(json.dumps({
+                            'event': 'stream_end',
+                            'message_num': message_num
+                        }))
+                    except:
+                        # The client closed the stream.
+                        pass
                     end_time = time.time()
                     elapsed_time = end_time - start_time
                     log_prompt(ip=handler.client_ip,
@@ -246,8 +206,7 @@ def do_stream(ws, model_name):
                                headers=r_headers,
                                backend_response_code=response_status_code,
                                request_url=r_url,
-                               backend_url=handler.backend_url,
-                               response_tokens=None
+                               backend_url=handler.backend_url
                                )
     finally:
         try:
diff --git a/server.py b/server.py
index 0eb0f6c..040b8d9 100644
--- a/server.py
+++ b/server.py
@@ -24,7 +24,8 @@ from llm_server.routes.server_error import handle_server_error
 from llm_server.routes.v1 import bp
 from llm_server.sock import init_socketio
 
-# TODO: make sure system tokens are excluded from 5/24 hr proompters
+
+# TODO: redis SCAN vs KEYS??
 # TODO: implement blind RRD controlled via header and only used when there is a queue on the primary backend(s)
 # TODO: is frequency penalty the same as ooba repetition penalty???
 # TODO: make sure openai_moderation_enabled works on websockets, completions, and chat completions