default to 100 proxy threads, 1 warc writer thread

see https://github.com/internetarchive/warcprox/wiki/benchmarking-number-of-threads

setup.py

@@ -40,7 +40,7 @@ except:
 
 setuptools.setup(
         name='warcprox',
-        version='2.4b2.dev167',
+        version='2.4b2.dev168',
         description='WARC writing MITM HTTP/S proxy',
         url='https://github.com/internetarchive/warcprox',
         author='Noah Levitt',

warcprox/controller.py

@@ -141,11 +141,9 @@ class WarcproxController(object):
         self.playback_proxy = Factory.playback_proxy(
             self.proxy.ca, self.options)
 
-        # default number of warc writer threads = sqrt(proxy.max_threads)
-        # pulled out of thin air because it strikes me as reasonable
-        # 1=>1 2=>1 5=>2 10=>3 50=>7 100=>10 200=>14 500=>22 1000=>32 2000=>45
+        # https://github.com/internetarchive/warcprox/wiki/benchmarking-number-of-threads
         if not self.options.writer_threads:
-            self.options.writer_threads = int(self.proxy.max_threads ** 0.5)
+            self.options.writer_threads = 1
 
         self.build_postfetch_chain(self.proxy.recorded_url_q)
 

warcprox/mitmproxy.py

@@ -501,35 +501,14 @@ class MitmProxyHandler(http_server.BaseHTTPRequestHandler):
 class PooledMixIn(socketserver.ThreadingMixIn):
     logger = logging.getLogger("warcprox.mitmproxy.PooledMixIn")
     def __init__(self, max_threads=None):
-        '''
-        If max_threads is not supplied, calculates a reasonable value based
-        on system resource limits.
-        '''
         self.active_requests = set()
         self.unaccepted_requests = 0
-        if not max_threads:
-            # man getrlimit: "RLIMIT_NPROC The maximum number of processes (or,
-            # more precisely on Linux, threads) that can be created for the
-            # real user ID of the calling process."
-            try:
-                import resource
-                rlimit_nproc = resource.getrlimit(resource.RLIMIT_NPROC)[0]
-                rlimit_nofile = resource.getrlimit(resource.RLIMIT_NOFILE)[0]
-                max_threads = min(rlimit_nofile // 10, rlimit_nproc // 2)
-                # resource.RLIM_INFINITY == -1 which can result in max_threads == 0
-                if max_threads <= 0 or max_threads > 5000:
-                    max_threads = 5000
-                self.logger.info(
-                        "max_threads=%s (rlimit_nproc=%s, rlimit_nofile=%s)",
-                        max_threads, rlimit_nproc, rlimit_nofile)
-            except Exception as e:
-                self.logger.warn(
-                        "unable to calculate optimal number of threads based "
-                        "on resource limits due to %s", e)
-                max_threads = 100
-                self.logger.info("max_threads=%s", max_threads)
-        self.max_threads = max_threads
-        self.pool = concurrent.futures.ThreadPoolExecutor(max_threads)
+        if max_threads:
+            self.max_threads = max_threads
+        else:
+            self.max_threads = 100
+        self.pool = concurrent.futures.ThreadPoolExecutor(self.max_threads)
+        self.logger.info("%s proxy threads", self.max_threads)
 
     def status(self):
         if hasattr(super(), 'status'):

warcprox/writerthread.py

@@ -54,7 +54,7 @@ class WarcWriterProcessor(warcprox.BaseStandardPostfetchProcessor):
         self.batch = set()
 
     def _startup(self):
-        self.logger.info('%s threads', self.pool._max_workers)
+        self.logger.info('%s warc writer threads', self.pool._max_workers)
         warcprox.BaseStandardPostfetchProcessor._startup(self)
 
     def _get_process_put(self):