let the OS pick an available port, to avoid what appear to be timing …

…issues causing multiple browsers to choose the same port

brozzler/browser.py

@@ -30,6 +30,7 @@
 import base64
 from brozzler.chrome import Chrome
 import surt
+import socket
 
 class BrowsingException(Exception):
     pass
@@ -41,10 +42,12 @@ class BrowsingTimeout(BrowsingException):
     pass
 
 class BrowserPool:
+    '''
+    Manages pool of browsers. Automatically chooses available port for the
+    debugging protocol.
+    '''
     logger = logging.getLogger(__module__ + '.' + __qualname__)
 
-    BASE_PORT = 9200
-
     def __init__(self, size=3, **kwargs):
         '''
         Initializes the pool.
@@ -54,13 +57,8 @@ def __init__(self, size=3, **kwargs):
             **kwargs: arguments for Browser(...)
         '''
         self.size = size
-        self._available = set()
+        self.kwargs = kwargs
         self._in_use = set()
-
-        for i in range(0, size):
-            browser = Browser(port=BrowserPool.BASE_PORT + i, **kwargs)
-            self._available.add(browser)
-
         self._lock = threading.Lock()
 
     def acquire(self):
@@ -74,30 +72,35 @@ def acquire(self):
             NoBrowsersAvailable if none available
         '''
         with self._lock:
-            try:
-                browser = self._available.pop()
-            except KeyError:
+            if len(self._in_use) >= self.size:
                 raise NoBrowsersAvailable
+
+            # choose available port
+            sock = socket.socket()
+            sock.bind(('0.0.0.0', 0))
+            port = sock.getsockname()[1]
+            sock.close()
+
+            browser = Browser(port=port, **self.kwargs)
             self._in_use.add(browser)
             return browser
 
     def release(self, browser):
+        browser.stop()  # make sure
         with self._lock:
-            self._available.add(browser)
             self._in_use.remove(browser)
 
     def shutdown_now(self):
         self.logger.info(
                 'shutting down browser pool (%s browsers in use)',
                 len(self._in_use))
         with self._lock:
-            for browser in self._available:
-                browser.stop()
             for browser in self._in_use:
                 browser.stop()
+            self._in_use.clear()
 
     def num_available(self):
-        return len(self._available)
+        return self.size - len(self._in_use)
 
     def num_in_use(self):
         return len(self._in_use)

brozzler/chrome.py

@@ -28,7 +28,6 @@
 import signal
 import sqlite3
 import json
-import psutil
 import tempfile
 
 class Chrome:
@@ -61,24 +60,6 @@ def __enter__(self):
     def __exit__(self, *args):
         self.stop()
 
-    def _find_available_port(self, default_port=9200):
-        try:
-            conns = psutil.net_connections(kind='tcp')
-        except psutil.AccessDenied:
-            return default_port
-
-        if not any(conn.laddr[1] == default_port for conn in conns):
-            return default_port
-
-        for p in range(9999,8999,-1):
-            if not any(conn.laddr[1] == p for conn in conns):
-                self.logger.warn(
-                        'port %s already in use, using %s instead',
-                        default_port, p)
-                return p
-
-        return default_port
-
     def _init_cookie_db(self, cookie_db):
         cookie_dir = os.path.join(self._chrome_user_data_dir, 'Default')
         cookie_location = os.path.join(cookie_dir, 'Cookies')
@@ -140,7 +121,6 @@ def start(self, proxy=None, cookie_db=None):
 
         new_env = os.environ.copy()
         new_env['HOME'] = self._home_tmpdir.name
-        self.port = self._find_available_port(self.port)
         chrome_args = [
                 self.chrome_exe,
                 '--remote-debugging-port=%s' % self.port,

setup.py

@@ -32,7 +32,7 @@ def find_package_data(package):
 
 setuptools.setup(
         name='brozzler',
-        version='1.1b9.dev192',
+        version='1.1b9.dev193',
         description='Distributed web crawling with browsers',
         url='https://github.com/internetarchive/brozzler',
         author='Noah Levitt',
@@ -71,7 +71,6 @@ def find_package_data(package):
             'surt>=0.3.0',
             'rethinkstuff>=0.1.5',
             'rethinkdb>=2.3,<2.4',
-            'psutil==4.3.0',
             'cerberus==1.0.1',
             'jinja2',
         ],

tests/test_units.py

@@ -25,7 +25,6 @@
 import brozzler.chrome
 import socket
 import logging
-import psutil
 import yaml
 
 @pytest.fixture(scope='module')
@@ -57,23 +56,6 @@ def test_robots(httpd):
     site = brozzler.Site(seed=url, user_agent='im/a bAdBOt/uh huh')
     assert not brozzler.is_permitted_by_robots(site, url)
 
-def test_find_available_port():
-    x = brozzler.chrome.Chrome(None, None)
-    try:
-        psutil.net_connections(kind='tcp')
-    except psutil.AccessDenied:
-        logging.warn(
-                'skipping _find_available_port() test because '
-                'psutil.net_connections(kind="tcp") raised AccessDenied')
-        return
-    assert x._find_available_port(9800) == 9800
-    sock = socket.socket()
-    sock.bind(('localhost', 9800))
-    sock.listen(0)
-    assert x._find_available_port(9800) >= 9990
-    sock.close()
-    assert x._find_available_port(9800) == 9800
-
 def test_scoping():
     test_scope = yaml.load('''
 max_hops: 100