##// END OF EJS Templates
wireproto: add streams to frame-based protocol...
wireproto: add streams to frame-based protocol Previously, the frame-based protocol was just a series of frames, with each frame associated with a request ID. In order to scale the protocol, we'll want to enable the use of compression. While it is possible to enable compression at the socket/pipe level, this has its disadvantages. The big one is it undermines the point of frames being standalone, atomic units that can be read and written: if you add compression above the framing protocol, you are back to having a stream-based protocol as opposed to something frame-based. So in order to preserve frames, compression needs to occur at the frame payload level. Compressing each frame's payload individually will limit compression ratios because the window size of the compressor will be limited by the max frame size, which is 32-64kb as currently defined. It will also add CPU overhead, as it is more efficient for compressors to operate on fewer, larger blocks of data than more, smaller blocks. So compressing each frame independently is out. This means we need to compress each frame's payload as if it is part of a larger stream. The simplest approach is to have 1 stream per connection. This could certainly work. However, it has disadvantages (documented below). We could also have 1 stream per RPC/command invocation. (This is the model HTTP/2 goes with.) This also has disadvantages. The main disadvantage to one global stream is that it has the very real potential to create CPU bottlenecks doing compression. Networks are only getting faster and the performance of single CPU cores has been relatively flat. Newer compression formats like zstandard offer better CPU cycle efficiency than predecessors like zlib. But it still all too common to saturate your CPU with compression overhead long before you saturate the network pipe. The main disadvantage with streams per request is that you can't reap the benefits of the compression context for multiple requests. For example, if you send 1000 RPC requests (or HTTP/2 requests for that matter), the response to each would have its own compression context. The overall size of the raw responses would be larger because compression contexts wouldn't be able to reference data from another request or response. The approach for streams as implemented in this commit is to support N streams per connection and for streams to potentially span requests and responses. As explained by the added internals docs, this facilitates servers and clients delegating independent streams and compression to independent threads / CPU cores. This helps alleviate the CPU bottleneck of compression. This design also allows compression contexts to be reused across requests/responses. This can result in improved compression ratios and less overhead for compressors and decompressors having to build new contexts. Another feature that was defined was the ability for individual frames within a stream to declare whether that individual frame's payload uses the content encoding (read: compression) defined by the stream. The idea here is that some servers may serve data from a combination of caches and dynamic resolution. Data coming from caches may be pre-compressed. We want to facilitate servers being able to essentially stream bytes from caches to the wire with minimal overhead. Being able to mix and match with frames are compressed within a stream enables these types of advanced server functionality. This commit defines the new streams mechanism. Basic code for supporting streams in frames has been added. But that code is seriously lacking and doesn't fully conform to the defined protocol. For example, we don't close any streams. And support for content encoding within streams is not yet implemented. The change was rather invasive and I didn't think it would be reasonable to implement the entire feature in a single commit. For the record, I would have loved to reuse an existing multiplexing protocol to build the new wire protocol on top of. However, I couldn't find a protocol that offers the performance and scaling characteristics that I desired. Namely, it should support multiple compression contexts to facilitate scaling out to multiple CPU cores and compression contexts should be able to live longer than single RPC requests. HTTP/2 *almost* fits the bill. But the semantics of HTTP message exchange state that streams can only live for a single request-response. We /could/ tunnel on top of HTTP/2 streams and frames with HEADER and DATA frames. But there's no guarantee that HTTP/2 libraries and proxies would allow us to use HTTP/2 streams and frames without the HTTP message exchange semantics defined in RFC 7540 Section 8. Other RPC protocols like gRPC tunnel are built on top of HTTP/2 and thus preserve its semantics of stream per RPC invocation. Even QUIC does this. We could attempt to invent a higher-level stream that spans HTTP/2 streams. But this would be violating HTTP/2 because there is no guarantee that HTTP/2 streams are routed to the same server. The best we can do - which is what this protocol does - is shoehorn all request and response data into a single HTTP message and create streams within. At that point, we've defined a Content-Type in HTTP parlance. It just so happens our media type can also work as a standalone, stream-based protocol, without leaning on HTTP or similar protocol. Differential Revision: https://phab.mercurial-scm.org/D2907

File last commit:

r36891:16499427 default
r37304:9bfcbe4f default
Show More
archival.py
342 lines | 10.7 KiB | text/x-python | PythonLexer
# archival.py - revision archival for mercurial
#
# Copyright 2006 Vadim Gelfer <vadim.gelfer@gmail.com>
#
# This software may be used and distributed according to the terms of the
# GNU General Public License version 2 or any later version.
from __future__ import absolute_import
import gzip
import os
import struct
import tarfile
import time
import zipfile
import zlib
from .i18n import _
from . import (
error,
formatter,
match as matchmod,
pycompat,
scmutil,
util,
vfs as vfsmod,
)
stringio = util.stringio
# from unzip source code:
_UNX_IFREG = 0x8000
_UNX_IFLNK = 0xa000
def tidyprefix(dest, kind, prefix):
'''choose prefix to use for names in archive. make sure prefix is
safe for consumers.'''
if prefix:
prefix = util.normpath(prefix)
else:
if not isinstance(dest, bytes):
raise ValueError('dest must be string if no prefix')
prefix = os.path.basename(dest)
lower = prefix.lower()
for sfx in exts.get(kind, []):
if lower.endswith(sfx):
prefix = prefix[:-len(sfx)]
break
lpfx = os.path.normpath(util.localpath(prefix))
prefix = util.pconvert(lpfx)
if not prefix.endswith('/'):
prefix += '/'
# Drop the leading '.' path component if present, so Windows can read the
# zip files (issue4634)
if prefix.startswith('./'):
prefix = prefix[2:]
if prefix.startswith('../') or os.path.isabs(lpfx) or '/../' in prefix:
raise error.Abort(_('archive prefix contains illegal components'))
return prefix
exts = {
'tar': ['.tar'],
'tbz2': ['.tbz2', '.tar.bz2'],
'tgz': ['.tgz', '.tar.gz'],
'zip': ['.zip'],
}
def guesskind(dest):
for kind, extensions in exts.iteritems():
if any(dest.endswith(ext) for ext in extensions):
return kind
return None
def _rootctx(repo):
# repo[0] may be hidden
for rev in repo:
return repo[rev]
return repo['null']
# {tags} on ctx includes local tags and 'tip', with no current way to limit
# that to global tags. Therefore, use {latesttag} as a substitute when
# the distance is 0, since that will be the list of global tags on ctx.
_defaultmetatemplate = br'''
repo: {root}
node: {ifcontains(rev, revset("wdir()"), "{p1node}{dirty}", "{node}")}
branch: {branch|utf8}
{ifeq(latesttagdistance, 0, join(latesttag % "tag: {tag}", "\n"),
separate("\n",
join(latesttag % "latesttag: {tag}", "\n"),
"latesttagdistance: {latesttagdistance}",
"changessincelatesttag: {changessincelatesttag}"))}
'''[1:] # drop leading '\n'
def buildmetadata(ctx):
'''build content of .hg_archival.txt'''
repo = ctx.repo()
opts = {
'template': repo.ui.config('experimental', 'archivemetatemplate',
_defaultmetatemplate)
}
out = util.stringio()
fm = formatter.formatter(repo.ui, out, 'archive', opts)
fm.startitem()
fm.context(ctx=ctx)
fm.data(root=_rootctx(repo).hex())
if ctx.rev() is None:
dirty = ''
if ctx.dirty(missing=True):
dirty = '+'
fm.data(dirty=dirty)
fm.end()
return out.getvalue()
class tarit(object):
'''write archive to tar file or stream. can write uncompressed,
or compress with gzip or bzip2.'''
class GzipFileWithTime(gzip.GzipFile):
def __init__(self, *args, **kw):
timestamp = None
if r'timestamp' in kw:
timestamp = kw.pop(r'timestamp')
if timestamp is None:
self.timestamp = time.time()
else:
self.timestamp = timestamp
gzip.GzipFile.__init__(self, *args, **kw)
def _write_gzip_header(self):
self.fileobj.write('\037\213') # magic header
self.fileobj.write('\010') # compression method
fname = self.name
if fname and fname.endswith('.gz'):
fname = fname[:-3]
flags = 0
if fname:
flags = gzip.FNAME
self.fileobj.write(pycompat.bytechr(flags))
gzip.write32u(self.fileobj, int(self.timestamp))
self.fileobj.write('\002')
self.fileobj.write('\377')
if fname:
self.fileobj.write(fname + '\000')
def __init__(self, dest, mtime, kind=''):
self.mtime = mtime
self.fileobj = None
def taropen(mode, name='', fileobj=None):
if kind == 'gz':
mode = mode[0:1]
if not fileobj:
fileobj = open(name, mode + 'b')
gzfileobj = self.GzipFileWithTime(name,
pycompat.sysstr(mode + 'b'),
zlib.Z_BEST_COMPRESSION,
fileobj, timestamp=mtime)
self.fileobj = gzfileobj
return tarfile.TarFile.taropen(
name, pycompat.sysstr(mode), gzfileobj)
else:
return tarfile.open(
name, pycompat.sysstr(mode + kind), fileobj)
if isinstance(dest, bytes):
self.z = taropen('w:', name=dest)
else:
self.z = taropen('w|', fileobj=dest)
def addfile(self, name, mode, islink, data):
name = pycompat.fsdecode(name)
i = tarfile.TarInfo(name)
i.mtime = self.mtime
i.size = len(data)
if islink:
i.type = tarfile.SYMTYPE
i.mode = 0o777
i.linkname = pycompat.fsdecode(data)
data = None
i.size = 0
else:
i.mode = mode
data = stringio(data)
self.z.addfile(i, data)
def done(self):
self.z.close()
if self.fileobj:
self.fileobj.close()
class zipit(object):
'''write archive to zip file or stream. can write uncompressed,
or compressed with deflate.'''
def __init__(self, dest, mtime, compress=True):
self.z = zipfile.ZipFile(pycompat.fsdecode(dest), r'w',
compress and zipfile.ZIP_DEFLATED or
zipfile.ZIP_STORED)
# Python's zipfile module emits deprecation warnings if we try
# to store files with a date before 1980.
epoch = 315532800 # calendar.timegm((1980, 1, 1, 0, 0, 0, 1, 1, 0))
if mtime < epoch:
mtime = epoch
self.mtime = mtime
self.date_time = time.gmtime(mtime)[:6]
def addfile(self, name, mode, islink, data):
i = zipfile.ZipInfo(pycompat.fsdecode(name), self.date_time)
i.compress_type = self.z.compression
# unzip will not honor unix file modes unless file creator is
# set to unix (id 3).
i.create_system = 3
ftype = _UNX_IFREG
if islink:
mode = 0o777
ftype = _UNX_IFLNK
i.external_attr = (mode | ftype) << 16
# add "extended-timestamp" extra block, because zip archives
# without this will be extracted with unexpected timestamp,
# if TZ is not configured as GMT
i.extra += struct.pack('<hhBl',
0x5455, # block type: "extended-timestamp"
1 + 4, # size of this block
1, # "modification time is present"
int(self.mtime)) # last modification (UTC)
self.z.writestr(i, data)
def done(self):
self.z.close()
class fileit(object):
'''write archive as files in directory.'''
def __init__(self, name, mtime):
self.basedir = name
self.opener = vfsmod.vfs(self.basedir)
self.mtime = mtime
def addfile(self, name, mode, islink, data):
if islink:
self.opener.symlink(data, name)
return
f = self.opener(name, "w", atomictemp=False)
f.write(data)
f.close()
destfile = os.path.join(self.basedir, name)
os.chmod(destfile, mode)
if self.mtime is not None:
os.utime(destfile, (self.mtime, self.mtime))
def done(self):
pass
archivers = {
'files': fileit,
'tar': tarit,
'tbz2': lambda name, mtime: tarit(name, mtime, 'bz2'),
'tgz': lambda name, mtime: tarit(name, mtime, 'gz'),
'uzip': lambda name, mtime: zipit(name, mtime, False),
'zip': zipit,
}
def archive(repo, dest, node, kind, decode=True, matchfn=None,
prefix='', mtime=None, subrepos=False):
'''create archive of repo as it was at node.
dest can be name of directory, name of archive file, or file
object to write archive to.
kind is type of archive to create.
decode tells whether to put files through decode filters from
hgrc.
matchfn is function to filter names of files to write to archive.
prefix is name of path to put before every archive member.
mtime is the modified time, in seconds, or None to use the changeset time.
subrepos tells whether to include subrepos.
'''
if kind == 'files':
if prefix:
raise error.Abort(_('cannot give prefix when archiving to files'))
else:
prefix = tidyprefix(dest, kind, prefix)
def write(name, mode, islink, getdata):
data = getdata()
if decode:
data = repo.wwritedata(name, data)
archiver.addfile(prefix + name, mode, islink, data)
if kind not in archivers:
raise error.Abort(_("unknown archive type '%s'") % kind)
ctx = repo[node]
archiver = archivers[kind](dest, mtime or ctx.date()[0])
if repo.ui.configbool("ui", "archivemeta"):
name = '.hg_archival.txt'
if not matchfn or matchfn(name):
write(name, 0o644, False, lambda: buildmetadata(ctx))
if matchfn:
files = [f for f in ctx.manifest().keys() if matchfn(f)]
else:
files = ctx.manifest().keys()
total = len(files)
if total:
files.sort()
scmutil.fileprefetchhooks(repo, ctx, files)
repo.ui.progress(_('archiving'), 0, unit=_('files'), total=total)
for i, f in enumerate(files):
ff = ctx.flags(f)
write(f, 'x' in ff and 0o755 or 0o644, 'l' in ff, ctx[f].data)
repo.ui.progress(_('archiving'), i + 1, item=f,
unit=_('files'), total=total)
repo.ui.progress(_('archiving'), None)
if subrepos:
for subpath in sorted(ctx.substate):
sub = ctx.workingsub(subpath)
submatch = matchmod.subdirmatcher(subpath, matchfn)
total += sub.archive(archiver, prefix, submatch, decode)
if total == 0:
raise error.Abort(_('no files match the archive pattern'))
archiver.done()
return total