sapling/mercurial/hgweb/protocol.py

199 lines
6.5 KiB
Python
Raw Normal View History

#
# Copyright 21 May 2005 - (c) 2005 Jake Edge <jake@edge2.net>
# Copyright 2005-2007 Matt Mackall <mpm@selenic.com>
#
# This software may be used and distributed according to the terms of the
2010-01-20 07:20:08 +03:00
# GNU General Public License version 2 or any later version.
2015-10-31 16:07:40 +03:00
from __future__ import absolute_import
import cgi
protocol: send application/mercurial-0.2 responses to capable clients With this commit, the HTTP transport now parses the X-HgProto-<N> header to determine what media type and compression engine to use for responses. So far, we only compress responses that are already being compressed with zlib today (stream response types to specific commands). We can expand things to cover additional response types later. The practical side-effect of this commit is that non-zlib compression engines will be used if both ends support them. This means if both ends have zstd support, zstd - not zlib - will be used to compress data! When cloning the mozilla-unified repository between a local HTTP server and client, the benefits of non-zlib compression are quite noticeable: engine server CPU (s) client CPU (s) bundle size zlib (l=6) 174.1 283.2 1,148,547,026 zstd (l=1) 99.2 267.3 1,127,513,841 zstd (l=3) 103.1 266.9 1,018,861,363 zstd (l=7) 128.3 269.7 919,190,278 zstd (l=10) 162.0 - 894,547,179 none 95.3 277.2 4,097,566,064 The default zstd compression level is 3. So if you deploy zstd capable Mercurial to your clients and servers and CPU time on your server is dominated by "getbundle" requests (clients cloning and pulling) - and my experience at Mozilla tells me this is often the case - this commit could drastically reduce your server-side CPU usage *and* save on bandwidth costs! Another benefit of this change is that server operators can install *any* compression engine. While it isn't enabled by default, the "none" compression engine can now be used to disable wire protocol compression completely. Previously, commands like "getbundle" always zlib compressed output, adding considerable overhead to generating responses. If you are on a high speed network and your server is under high load, it might be advantageous to trade bandwidth for CPU. Although, zstd at level 1 doesn't use that much CPU, so I'm not convinced that disabling compression wholesale is worthwhile. And, my data seems to indicate a slow down on the client without compression. I suspect this is due to a lack of buffering resulting in an increase in socket read() calls and/or the fact we're transferring an extra 3 GB of data (parsing HTTP chunked transfer and processing extra TCP packets can add up). This is definitely worth investigating and optimizing. But since the "none" compressor isn't enabled by default, I'm inclined to punt on this issue. This commit introduces tons of tests. Some of these should arguably have been implemented on previous commits. But it was difficult to test without the server functionality in place.
2016-12-25 01:29:32 +03:00
import struct
2015-10-31 16:07:40 +03:00
from .common import (
HTTP_OK,
)
from .. import (
util,
wireproto,
)
stringio = util.stringio
urlerr = util.urlerr
urlreq = util.urlreq
HGTYPE = 'application/mercurial-0.1'
protocol: send application/mercurial-0.2 responses to capable clients With this commit, the HTTP transport now parses the X-HgProto-<N> header to determine what media type and compression engine to use for responses. So far, we only compress responses that are already being compressed with zlib today (stream response types to specific commands). We can expand things to cover additional response types later. The practical side-effect of this commit is that non-zlib compression engines will be used if both ends support them. This means if both ends have zstd support, zstd - not zlib - will be used to compress data! When cloning the mozilla-unified repository between a local HTTP server and client, the benefits of non-zlib compression are quite noticeable: engine server CPU (s) client CPU (s) bundle size zlib (l=6) 174.1 283.2 1,148,547,026 zstd (l=1) 99.2 267.3 1,127,513,841 zstd (l=3) 103.1 266.9 1,018,861,363 zstd (l=7) 128.3 269.7 919,190,278 zstd (l=10) 162.0 - 894,547,179 none 95.3 277.2 4,097,566,064 The default zstd compression level is 3. So if you deploy zstd capable Mercurial to your clients and servers and CPU time on your server is dominated by "getbundle" requests (clients cloning and pulling) - and my experience at Mozilla tells me this is often the case - this commit could drastically reduce your server-side CPU usage *and* save on bandwidth costs! Another benefit of this change is that server operators can install *any* compression engine. While it isn't enabled by default, the "none" compression engine can now be used to disable wire protocol compression completely. Previously, commands like "getbundle" always zlib compressed output, adding considerable overhead to generating responses. If you are on a high speed network and your server is under high load, it might be advantageous to trade bandwidth for CPU. Although, zstd at level 1 doesn't use that much CPU, so I'm not convinced that disabling compression wholesale is worthwhile. And, my data seems to indicate a slow down on the client without compression. I suspect this is due to a lack of buffering resulting in an increase in socket read() calls and/or the fact we're transferring an extra 3 GB of data (parsing HTTP chunked transfer and processing extra TCP packets can add up). This is definitely worth investigating and optimizing. But since the "none" compressor isn't enabled by default, I'm inclined to punt on this issue. This commit introduces tons of tests. Some of these should arguably have been implemented on previous commits. But it was difficult to test without the server functionality in place.
2016-12-25 01:29:32 +03:00
HGTYPE2 = 'application/mercurial-0.2'
HGERRTYPE = 'application/hg-error'
def decodevaluefromheaders(req, headerprefix):
"""Decode a long value from multiple HTTP request headers."""
chunks = []
i = 1
while True:
v = req.env.get('HTTP_%s_%d' % (
headerprefix.upper().replace('-', '_'), i))
if v is None:
break
chunks.append(v)
i += 1
return ''.join(chunks)
class webproto(wireproto.abstractserverproto):
def __init__(self, req, ui):
self.req = req
self.response = ''
self.ui = ui
self.name = 'http'
def getargs(self, args):
knownargs = self._args()
data = {}
keys = args.split()
for k in keys:
if k == '*':
star = {}
for key in knownargs.keys():
if key != 'cmd' and key not in keys:
star[key] = knownargs[key][0]
data['*'] = star
else:
data[k] = knownargs[k][0]
return [data[k] for k in keys]
def _args(self):
args = self.req.form.copy()
postlen = int(self.req.env.get('HTTP_X_HGARGS_POST', 0))
if postlen:
args.update(cgi.parse_qs(
self.req.read(postlen), keep_blank_values=True))
return args
argvalue = decodevaluefromheaders(self.req, 'X-HgArg')
args.update(cgi.parse_qs(argvalue, keep_blank_values=True))
return args
def getfile(self, fp):
length = int(self.req.env['CONTENT_LENGTH'])
for s in util.filechunkiter(self.req, limit=length):
fp.write(s)
def redirect(self):
self.oldio = self.ui.fout, self.ui.ferr
self.ui.ferr = self.ui.fout = stringio()
def restore(self):
val = self.ui.fout.getvalue()
self.ui.ferr, self.ui.fout = self.oldio
return val
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.
2016-10-16 21:10:21 +03:00
def _client(self):
return 'remote:%s:%s:%s' % (
self.req.env.get('wsgi.url_scheme') or 'http',
urlreq.quote(self.req.env.get('REMOTE_HOST', '')),
urlreq.quote(self.req.env.get('REMOTE_USER', '')))
protocol: send application/mercurial-0.2 responses to capable clients With this commit, the HTTP transport now parses the X-HgProto-<N> header to determine what media type and compression engine to use for responses. So far, we only compress responses that are already being compressed with zlib today (stream response types to specific commands). We can expand things to cover additional response types later. The practical side-effect of this commit is that non-zlib compression engines will be used if both ends support them. This means if both ends have zstd support, zstd - not zlib - will be used to compress data! When cloning the mozilla-unified repository between a local HTTP server and client, the benefits of non-zlib compression are quite noticeable: engine server CPU (s) client CPU (s) bundle size zlib (l=6) 174.1 283.2 1,148,547,026 zstd (l=1) 99.2 267.3 1,127,513,841 zstd (l=3) 103.1 266.9 1,018,861,363 zstd (l=7) 128.3 269.7 919,190,278 zstd (l=10) 162.0 - 894,547,179 none 95.3 277.2 4,097,566,064 The default zstd compression level is 3. So if you deploy zstd capable Mercurial to your clients and servers and CPU time on your server is dominated by "getbundle" requests (clients cloning and pulling) - and my experience at Mozilla tells me this is often the case - this commit could drastically reduce your server-side CPU usage *and* save on bandwidth costs! Another benefit of this change is that server operators can install *any* compression engine. While it isn't enabled by default, the "none" compression engine can now be used to disable wire protocol compression completely. Previously, commands like "getbundle" always zlib compressed output, adding considerable overhead to generating responses. If you are on a high speed network and your server is under high load, it might be advantageous to trade bandwidth for CPU. Although, zstd at level 1 doesn't use that much CPU, so I'm not convinced that disabling compression wholesale is worthwhile. And, my data seems to indicate a slow down on the client without compression. I suspect this is due to a lack of buffering resulting in an increase in socket read() calls and/or the fact we're transferring an extra 3 GB of data (parsing HTTP chunked transfer and processing extra TCP packets can add up). This is definitely worth investigating and optimizing. But since the "none" compressor isn't enabled by default, I'm inclined to punt on this issue. This commit introduces tons of tests. Some of these should arguably have been implemented on previous commits. But it was difficult to test without the server functionality in place.
2016-12-25 01:29:32 +03:00
def responsetype(self, v1compressible=False):
"""Determine the appropriate response type and compression settings.
The ``v1compressible`` argument states whether the response with
application/mercurial-0.1 media types should be zlib compressed.
Returns a tuple of (mediatype, compengine, engineopts).
"""
# For now, if it isn't compressible in the old world, it's never
# compressible. We can change this to send uncompressed 0.2 payloads
# later.
if not v1compressible:
return HGTYPE, None, None
# Determine the response media type and compression engine based
# on the request parameters.
protocaps = decodevaluefromheaders(self.req, 'X-HgProto').split(' ')
if '0.2' in protocaps:
# Default as defined by wire protocol spec.
compformats = ['zlib', 'none']
for cap in protocaps:
if cap.startswith('comp='):
compformats = cap[5:].split(',')
break
# Now find an agreed upon compression format.
for engine in wireproto.supportedcompengines(self.ui, self,
util.SERVERROLE):
if engine.wireprotosupport().name in compformats:
opts = {}
level = self.ui.configint('server',
'%slevel' % engine.name())
if level is not None:
opts['level'] = level
return HGTYPE2, engine, opts
# No mutually supported compression format. Fall back to the
# legacy protocol.
# 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.
opts = {'level': self.ui.configint('server', 'zliblevel', -1)}
return HGTYPE, util.compengines['zlib'], opts
def iscmd(cmd):
return cmd in wireproto.commands
def call(repo, req, cmd):
p = webproto(req, repo.ui)
protocol: send application/mercurial-0.2 responses to capable clients With this commit, the HTTP transport now parses the X-HgProto-<N> header to determine what media type and compression engine to use for responses. So far, we only compress responses that are already being compressed with zlib today (stream response types to specific commands). We can expand things to cover additional response types later. The practical side-effect of this commit is that non-zlib compression engines will be used if both ends support them. This means if both ends have zstd support, zstd - not zlib - will be used to compress data! When cloning the mozilla-unified repository between a local HTTP server and client, the benefits of non-zlib compression are quite noticeable: engine server CPU (s) client CPU (s) bundle size zlib (l=6) 174.1 283.2 1,148,547,026 zstd (l=1) 99.2 267.3 1,127,513,841 zstd (l=3) 103.1 266.9 1,018,861,363 zstd (l=7) 128.3 269.7 919,190,278 zstd (l=10) 162.0 - 894,547,179 none 95.3 277.2 4,097,566,064 The default zstd compression level is 3. So if you deploy zstd capable Mercurial to your clients and servers and CPU time on your server is dominated by "getbundle" requests (clients cloning and pulling) - and my experience at Mozilla tells me this is often the case - this commit could drastically reduce your server-side CPU usage *and* save on bandwidth costs! Another benefit of this change is that server operators can install *any* compression engine. While it isn't enabled by default, the "none" compression engine can now be used to disable wire protocol compression completely. Previously, commands like "getbundle" always zlib compressed output, adding considerable overhead to generating responses. If you are on a high speed network and your server is under high load, it might be advantageous to trade bandwidth for CPU. Although, zstd at level 1 doesn't use that much CPU, so I'm not convinced that disabling compression wholesale is worthwhile. And, my data seems to indicate a slow down on the client without compression. I suspect this is due to a lack of buffering resulting in an increase in socket read() calls and/or the fact we're transferring an extra 3 GB of data (parsing HTTP chunked transfer and processing extra TCP packets can add up). This is definitely worth investigating and optimizing. But since the "none" compressor isn't enabled by default, I'm inclined to punt on this issue. This commit introduces tons of tests. Some of these should arguably have been implemented on previous commits. But it was difficult to test without the server functionality in place.
2016-12-25 01:29:32 +03:00
def genversion2(gen, compress, engine, engineopts):
# application/mercurial-0.2 always sends a payload header
# identifying the compression engine.
name = engine.wireprotosupport().name
assert 0 < len(name) < 256
yield struct.pack('B', len(name))
yield name
if compress:
for chunk in engine.compressstream(gen, opts=engineopts):
yield chunk
else:
for chunk in gen:
yield chunk
rsp = wireproto.dispatch(repo, p, cmd)
if isinstance(rsp, str):
req.respond(HTTP_OK, HGTYPE, body=rsp)
return []
elif isinstance(rsp, wireproto.streamres):
if rsp.reader:
gen = iter(lambda: rsp.reader.read(32768), '')
else:
gen = rsp.gen
protocol: send application/mercurial-0.2 responses to capable clients With this commit, the HTTP transport now parses the X-HgProto-<N> header to determine what media type and compression engine to use for responses. So far, we only compress responses that are already being compressed with zlib today (stream response types to specific commands). We can expand things to cover additional response types later. The practical side-effect of this commit is that non-zlib compression engines will be used if both ends support them. This means if both ends have zstd support, zstd - not zlib - will be used to compress data! When cloning the mozilla-unified repository between a local HTTP server and client, the benefits of non-zlib compression are quite noticeable: engine server CPU (s) client CPU (s) bundle size zlib (l=6) 174.1 283.2 1,148,547,026 zstd (l=1) 99.2 267.3 1,127,513,841 zstd (l=3) 103.1 266.9 1,018,861,363 zstd (l=7) 128.3 269.7 919,190,278 zstd (l=10) 162.0 - 894,547,179 none 95.3 277.2 4,097,566,064 The default zstd compression level is 3. So if you deploy zstd capable Mercurial to your clients and servers and CPU time on your server is dominated by "getbundle" requests (clients cloning and pulling) - and my experience at Mozilla tells me this is often the case - this commit could drastically reduce your server-side CPU usage *and* save on bandwidth costs! Another benefit of this change is that server operators can install *any* compression engine. While it isn't enabled by default, the "none" compression engine can now be used to disable wire protocol compression completely. Previously, commands like "getbundle" always zlib compressed output, adding considerable overhead to generating responses. If you are on a high speed network and your server is under high load, it might be advantageous to trade bandwidth for CPU. Although, zstd at level 1 doesn't use that much CPU, so I'm not convinced that disabling compression wholesale is worthwhile. And, my data seems to indicate a slow down on the client without compression. I suspect this is due to a lack of buffering resulting in an increase in socket read() calls and/or the fact we're transferring an extra 3 GB of data (parsing HTTP chunked transfer and processing extra TCP packets can add up). This is definitely worth investigating and optimizing. But since the "none" compressor isn't enabled by default, I'm inclined to punt on this issue. This commit introduces tons of tests. Some of these should arguably have been implemented on previous commits. But it was difficult to test without the server functionality in place.
2016-12-25 01:29:32 +03:00
# This code for compression should not be streamres specific. It
# is here because we only compress streamres at the moment.
mediatype, engine, engineopts = p.responsetype(rsp.v1compressible)
if mediatype == HGTYPE and rsp.v1compressible:
gen = engine.compressstream(gen, engineopts)
elif mediatype == HGTYPE2:
gen = genversion2(gen, rsp.v1compressible, engine, engineopts)
protocol: send application/mercurial-0.2 responses to capable clients With this commit, the HTTP transport now parses the X-HgProto-<N> header to determine what media type and compression engine to use for responses. So far, we only compress responses that are already being compressed with zlib today (stream response types to specific commands). We can expand things to cover additional response types later. The practical side-effect of this commit is that non-zlib compression engines will be used if both ends support them. This means if both ends have zstd support, zstd - not zlib - will be used to compress data! When cloning the mozilla-unified repository between a local HTTP server and client, the benefits of non-zlib compression are quite noticeable: engine server CPU (s) client CPU (s) bundle size zlib (l=6) 174.1 283.2 1,148,547,026 zstd (l=1) 99.2 267.3 1,127,513,841 zstd (l=3) 103.1 266.9 1,018,861,363 zstd (l=7) 128.3 269.7 919,190,278 zstd (l=10) 162.0 - 894,547,179 none 95.3 277.2 4,097,566,064 The default zstd compression level is 3. So if you deploy zstd capable Mercurial to your clients and servers and CPU time on your server is dominated by "getbundle" requests (clients cloning and pulling) - and my experience at Mozilla tells me this is often the case - this commit could drastically reduce your server-side CPU usage *and* save on bandwidth costs! Another benefit of this change is that server operators can install *any* compression engine. While it isn't enabled by default, the "none" compression engine can now be used to disable wire protocol compression completely. Previously, commands like "getbundle" always zlib compressed output, adding considerable overhead to generating responses. If you are on a high speed network and your server is under high load, it might be advantageous to trade bandwidth for CPU. Although, zstd at level 1 doesn't use that much CPU, so I'm not convinced that disabling compression wholesale is worthwhile. And, my data seems to indicate a slow down on the client without compression. I suspect this is due to a lack of buffering resulting in an increase in socket read() calls and/or the fact we're transferring an extra 3 GB of data (parsing HTTP chunked transfer and processing extra TCP packets can add up). This is definitely worth investigating and optimizing. But since the "none" compressor isn't enabled by default, I'm inclined to punt on this issue. This commit introduces tons of tests. Some of these should arguably have been implemented on previous commits. But it was difficult to test without the server functionality in place.
2016-12-25 01:29:32 +03:00
req.respond(HTTP_OK, mediatype)
return gen
elif isinstance(rsp, wireproto.pushres):
val = p.restore()
rsp = '%d\n%s' % (rsp.res, val)
req.respond(HTTP_OK, HGTYPE, body=rsp)
return []
elif isinstance(rsp, wireproto.pusherr):
# drain the incoming bundle
req.drain()
p.restore()
rsp = '0\n%s\n' % rsp.res
req.respond(HTTP_OK, HGTYPE, body=rsp)
return []
elif isinstance(rsp, wireproto.ooberror):
rsp = rsp.message
req.respond(HTTP_OK, HGERRTYPE, body=rsp)
return []