server: use queue blocking instead of sleep

server/httpserver.py

@@ -1,7 +1,6 @@
 import logging
 import multiprocessing as mp
 import socket
-import time
 from multiprocessing.context import Process
 from multiprocessing.queues import Queue
 from multiprocessing.synchronize import Event
@@ -10,9 +9,6 @@ from server import worker
 
 
 class HTTPServer:
-    """
-
-    """
     address: str
     port: int
     workers = []
@@ -24,11 +20,11 @@ class HTTPServer:
 
     def __init__(self, address: str, port: int, worker_count, logging_level):
         """
-        Initialize a HTTP server with the specified address, port, worker_count and logging_level
+        Initialize an HTTP server with the specified address, port, worker_count and logging_level
         @param address: the address to listen on for connections
         @param port: the port to listen on for connections
-        @param worker_count:
-        @param logging_level:
+        @param worker_count: The amount of worker processes to create
+        @param logging_level: verbosity level for the logger
         """
         self.address = address
         self.port = port
@@ -36,7 +32,9 @@ class HTTPServer:
         self.logging_level = logging_level
 
         mp.set_start_method("spawn")
-        self._dispatch_queue = mp.Queue()
+
+        # Create a queue with maximum size of worker_count.
+        self._dispatch_queue = mp.Queue(worker_count)
         self._stop_event = mp.Event()
 
     def start(self):
@@ -51,12 +49,9 @@ class HTTPServer:
     def __do_start(self):
         """
         Internal method to start the server.
-
-        @raise:
         """
         # Create socket
         self.server = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
-        # self.server.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
         self.server.bind((self.address, self.port))
 
         # Create workers processes to handle requests
@@ -75,14 +70,12 @@ class HTTPServer:
         logging.debug("Listening on %s:%d", self.address, self.port)
 
         while True:
-            if self._dispatch_queue.qsize() > self.worker_count:
-                time.sleep(0.01)
-                continue
-
             conn, addr = self.server.accept()
             conn.settimeout(5)
 
             logging.info("New connection: %s", addr[0])
+
+            # blocks when the queue is full (contains self.worker_count items).
             self._dispatch_queue.put((conn, addr))
             logging.debug("Dispatched connection %s", addr)
 
@@ -98,7 +91,7 @@ class HTTPServer:
 
         # Wake up workers
         logging.debug("Waking up workers")
-        for p in self.workers:
+        for _ in self.workers:
             self._dispatch_queue.put((None, None))
 
         logging.debug("Closing dispatch queue")
@@ -120,7 +113,6 @@ class HTTPServer:
 
         A worker process is created with start method "spawn", target `worker.worker` and the `self.logging_level`
         is passed along with the `self.dispatch_queue` and `self._stop_event`
-
         """
         for i in range(self.worker_count):
             logging.debug("Creating worker: %d", i + 1)
@@ -129,6 +121,3 @@ class HTTPServer:
                                  self._stop_event))
             p.start()
             self.workers.append(p)
-
-            time.sleep(0.2)
-        time.sleep(1)