logging tweaks

brozzler/browser.py

@@ -233,8 +233,7 @@ class WebsockReceiverThread(threading.Thread):
                 brozzler.thread_raise(self.calling_thread, BrowsingException)
             elif message['method'] == 'Console.messageAdded':
                 self.logger.debug(
-                        '%s console.%s %s', self.websock.url,
-                        message['params']['message']['level'],
+                        'console.%s %s', message['params']['message']['level'],
                         message['params']['message']['text'])
             elif message['method'] == 'Page.javascriptDialogOpening':
                 self._javascript_dialog_opening(message)
@@ -310,10 +309,8 @@ class Browser:
         if not self.is_running():
             self.websock_url = self.chrome.start(**kwargs)
             self.websock = websocket.WebSocketApp(self.websock_url)
-            thread_name = 'WebsockThread:{}-{:%Y%m%d%H%M%S}'.format(
-                    surt.handyurl.parse(self.websock_url).port,
-                    datetime.datetime.utcnow())
-            self.websock_thread = WebsockReceiverThread(self.websock)
+            self.websock_thread = WebsockReceiverThread(
+                    self.websock, name='WebsockThread:%s' % self.chrome.port)
             self.websock_thread.start()
 
             self._wait_for(lambda: self.websock_thread.is_open, timeout=10)

brozzler/chrome.py

@@ -165,8 +165,7 @@ class Chrome:
                 stdout=subprocess.PIPE, stderr=subprocess.PIPE, bufsize=0)
         self._out_reader_thread = threading.Thread(
                 target=self._read_stderr_stdout,
-                name='ChromeOutReaderThread(pid=%s)' % self.chrome_process.pid,
-                daemon=True)
+                name='ChromeOutReaderThread:%s' % self.port, daemon=True)
         self._out_reader_thread.start()
         self.logger.info('chrome running, pid %s' % self.chrome_process.pid)
 

brozzler/frontier.py

@@ -125,7 +125,7 @@ class RethinkDbFrontier:
         self._vet_result(result, replaced=[0,1], unchanged=[0,1])
 
     def new_page(self, page):
-        self.logger.debug("inserting into 'pages' table %s", page)
+        self.logger.trace("inserting into 'pages' table %s", page)
         result = self.r.table("pages").insert(page.to_dict()).run()
         self._vet_result(result, inserted=1)
 

brozzler/worker.py

@@ -418,7 +418,7 @@ class BrozzlerWorker:
                                 repr(self._proxy(site)), site)
                         th = threading.Thread(
                                 target=self._brozzle_site, args=(browser, site),
-                                name="BrozzlingThread:%s" % site.seed,
+                                name="BrozzlingThread:%s" % browser.chrome.port,
                                 daemon=True)
                         with self._browsing_threads_lock:
                             self._browsing_threads.add(th)
@@ -432,9 +432,7 @@ class BrozzlerWorker:
                                 "all %s browsers are busy", self._max_browsers)
                         latest_state = "browsers-busy"
                 except brozzler.NothingToClaim:
-                    if latest_state != "no-unclaimed-sites":
-                        self.logger.info("no unclaimed sites to browse")
-                        latest_state = "no-unclaimed-sites"
+                    pass
                 time.sleep(0.5)
         except brozzler.ShutdownRequested:
             self.logger.info("shutdown requested")

setup.py

@@ -32,7 +32,7 @@ def find_package_data(package):
 
 setuptools.setup(
         name='brozzler',
-        version='1.1b9.dev185',
+        version='1.1b9.dev186',
         description='Distributed web crawling with browsers',
         url='https://github.com/internetarchive/brozzler',
         author='Noah Levitt',