changeset 30206:d105195436c0

wireproto: compress data from a generator Currently, the "getbundle" wire protocol command obtains a generator of data, converts it to a util.chunkbuffer, then converts it back to a generator via the protocol's groupchunks() implementation. For the SSH protocol, groupchunks() simply reads 4kb chunks then write()s the data to a file descriptor. For the HTTP protocol, groupchunks() reads 32kb chunks, feeds those into a zlib compressor, emits compressed data as it is available, and that is sent to the WSGI layer, where it is likely turned into HTTP chunked transfer chunks as is or further buffered and turned into a larger chunk. For both the SSH and HTTP protocols, there is inefficiency from using util.chunkbuffer. For SSH, emitting consistent 4kb chunks sounds nice. However, the file descriptor it is writing to is almost certainly buffered. That means that a Python .write() probably doesn't translate into exactly what is written to the I/O layer. For HTTP, we're going through an intermediate layer to zlib compress data. So all util.chunkbuffer is doing is ensuring that the chunks we feed into the zlib compressor are of uniform size. This means more CPU time in Python buffering and emitting chunks in util.chunkbuffer but fewer function calls to zlib. This patch introduces and implements a new wire protocol abstract method: compresschunks(). It is like groupchunks() except it operates on a generator instead of something with a .read(). The SSH implementation simply proxies chunks. The HTTP implementation uses zlib compression. To avoid duplicate code, the HTTP groupchunks() has been reimplemented in terms of compresschunks(). To prove this all works, the "getbundle" wire protocol command has been switched to compresschunks(). This removes the util.chunkbuffer from that command. Now, data essentially streams straight from the changegroup emitter to the wire, possibly through a zlib compressor. Generators all the way, baby. There were slim to no performance changes on the server as measured with the mozilla-central repository. This is likely because CPU time is dominated by reading revlogs, producing the changegroup, and zlib compressing the output stream. Still, this brings us a little closer to our ideal of using generators everywhere.
author Gregory Szorc <gregory.szorc@gmail.com>
date Sun, 16 Oct 2016 11:10:21 -0700
parents b4074417b661
children abe723002509
files mercurial/hgweb/protocol.py mercurial/sshserver.py mercurial/wireproto.py
diffstat 3 files changed, 26 insertions(+), 7 deletions(-) [+]
line wrap: on
line diff
--- a/mercurial/hgweb/protocol.py	Mon Oct 17 19:48:36 2016 +0200
+++ b/mercurial/hgweb/protocol.py	Sun Oct 16 11:10:21 2016 -0700
@@ -73,21 +73,30 @@
         val = self.ui.fout.getvalue()
         self.ui.ferr, self.ui.fout = self.oldio
         return val
+
     def groupchunks(self, fh):
+        def getchunks():
+            while True:
+                chunk = fh.read(32768)
+                if not chunk:
+                    break
+                yield chunk
+
+        return self.compresschunks(getchunks())
+
+    def compresschunks(self, chunks):
         # Don't allow untrusted settings because disabling compression or
         # setting a very high compression level could lead to flooding
         # the server's network or CPU.
         z = zlib.compressobj(self.ui.configint('server', 'zliblevel', -1))
-        while True:
-            chunk = fh.read(32768)
-            if not chunk:
-                break
+        for chunk in chunks:
             data = z.compress(chunk)
             # Not all calls to compress() emit data. It is cheaper to inspect
             # that here than to send it via the generator.
             if data:
                 yield data
         yield z.flush()
+
     def _client(self):
         return 'remote:%s:%s:%s' % (
             self.req.env.get('wsgi.url_scheme') or 'http',
--- a/mercurial/sshserver.py	Mon Oct 17 19:48:36 2016 +0200
+++ b/mercurial/sshserver.py	Sun Oct 16 11:10:21 2016 -0700
@@ -71,6 +71,10 @@
     def groupchunks(self, fh):
         return iter(lambda: fh.read(4096), '')
 
+    def compresschunks(self, chunks):
+        for chunk in chunks:
+            yield chunk
+
     def sendresponse(self, v):
         self.fout.write("%d\n" % len(v))
         self.fout.write(v)
--- a/mercurial/wireproto.py	Mon Oct 17 19:48:36 2016 +0200
+++ b/mercurial/wireproto.py	Sun Oct 16 11:10:21 2016 -0700
@@ -85,6 +85,14 @@
         """
         raise NotImplementedError()
 
+    def compresschunks(self, chunks):
+        """Generator of possible compressed chunks to send to the client.
+
+        This is like ``groupchunks()`` except it accepts a generator as
+        its argument.
+        """
+        raise NotImplementedError()
+
 class remotebatch(peer.batcher):
     '''batches the queued calls; uses as few roundtrips as possible'''
     def __init__(self, remote):
@@ -773,9 +781,7 @@
             return ooberror(bundle2required)
 
     chunks = exchange.getbundlechunks(repo, 'serve', **opts)
-    # TODO avoid util.chunkbuffer() here since it is adding overhead to
-    # what is fundamentally a generator proxying operation.
-    return streamres(proto.groupchunks(util.chunkbuffer(chunks)))
+    return streamres(proto.compresschunks(chunks))
 
 @wireprotocommand('heads')
 def heads(repo, proto):