view hgext/convert/cvs.py @ 40021:c537144fdbef

wireprotov2: support response caching

One of the things I've learned from managing VCS servers over the
years is that they are hard to scale. It is well known that some
companies have very beefy (read: very expensive) servers to power
their VCS needs. It is also known that specialized servers for
various VCS exist in order to facilitate scaling servers. (Mercurial
is in this boat.)

One of the aspects that make a VCS server hard to scale is the
high CPU load incurred by constant client clone/pull operations.

To alleviate the scaling pain associated with data retrieval
operations, I want to integrate caching into the Mercurial wire
protocol server as robustly as possible such that servers can
aggressively cache responses and defer as much server load as
possible.

This commit represents the initial implementation of a general
caching layer in wire protocol version 2.

We define a new interface and behavior for a wire protocol cacher
in repository.py. (This is probably where a reviewer should look
first to understand what is going on.)

The bulk of the added code is in wireprotov2server.py, where we
define how a command can opt in to being cached and integrate
caching into command dispatching.

From a very high-level:

* A command can declare itself as cacheable by providing a callable
  that can be used to derive a cache key.
* At dispatch time, if a command is cacheable, we attempt to
  construct a cacher and use it for serving the request and/or
  caching the request.
* The dispatch layer handles the bulk of the business logic for
  caching, making cachers mostly "dumb content stores."
* The mechanism for invalidating cached entries (one of the harder
  parts about caching in general) is by varying the cache key when
  state changes. As such, cachers don't need to be concerned with
  cache invalidation.

Initially, we've hooked up support for caching "manifestdata" and
"filedata" commands. These are the simplest to cache, as they should
be immutable over time. Caching of commands related to changeset
data is a bit harder (because cache validation is impacted by
changes to bookmarks, phases, etc). This will be implemented later.
(Strictly speaking, censoring a file should invalidate caches. I've
added an inline TODO to track this edge case.)

To prove it works, this commit implements a test-only extension
providing in-memory caching backed by an lrucachedict. A new test
showing this extension behaving properly is added. FWIW, the
cacher is ~50 lines of code, demonstrating the relative ease with
which a cache can be added to a server.

While the test cacher is not suitable for production workloads, just
for kicks I performed a clone of just the changeset and manifest data
for the mozilla-unified repository. With a fully warmed cache (of just
the manifest data since changeset data is not cached), server-side
CPU usage dropped from ~73s to ~28s. That's pretty significant and
demonstrates the potential that response caching has on server
scalability!

Differential Revision: https://phab.mercurial-scm.org/D4773

# cvs.py: CVS conversion code inspired by hg-cvs-import and git-cvsimport
#
#  Copyright 2005-2009 Matt Mackall <[e-mail]> and others
#
# 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 errno
import os
import re
import socket

from mercurial.i18n import _
from mercurial import (
    encoding,
    error,
    util,
)
from mercurial.utils import (
    dateutil,
    procutil,
)

from . import (
    common,
    cvsps,
)

stringio = util.stringio
checktool = common.checktool
commit = common.commit
converter_source = common.converter_source
makedatetimestamp = common.makedatetimestamp
NoRepo = common.NoRepo

class convert_cvs(converter_source):
    def __init__(self, ui, repotype, path, revs=None):
        super(convert_cvs, self).__init__(ui, repotype, path, revs=revs)

        cvs = os.path.join(path, "CVS")
        if not os.path.exists(cvs):
            raise NoRepo(_("%s does not look like a CVS checkout") % path)

        checktool('cvs')

        self.changeset = None
        self.files = {}
        self.tags = {}
        self.lastbranch = {}
        self.socket = None
        self.cvsroot = open(os.path.join(cvs, "Root"), 'rb').read()[:-1]
        self.cvsrepo = open(os.path.join(cvs, "Repository"), 'rb').read()[:-1]
        self.encoding = encoding.encoding

        self._connect()

    def _parse(self):
        if self.changeset is not None:
            return
        self.changeset = {}

        maxrev = 0
        if self.revs:
            if len(self.revs) > 1:
                raise error.Abort(_('cvs source does not support specifying '
                                   'multiple revs'))
            # TODO: handle tags
            try:
                # patchset number?
                maxrev = int(self.revs[0])
            except ValueError:
                raise error.Abort(_('revision %s is not a patchset number')
                                 % self.revs[0])

        d = encoding.getcwd()
        try:
            os.chdir(self.path)
            id = None

            cache = 'update'
            if not self.ui.configbool('convert', 'cvsps.cache'):
                cache = None
            db = cvsps.createlog(self.ui, cache=cache)
            db = cvsps.createchangeset(self.ui, db,
                fuzz=int(self.ui.config('convert', 'cvsps.fuzz')),
                mergeto=self.ui.config('convert', 'cvsps.mergeto'),
                mergefrom=self.ui.config('convert', 'cvsps.mergefrom'))

            for cs in db:
                if maxrev and cs.id > maxrev:
                    break
                id = (b"%d" % cs.id)
                cs.author = self.recode(cs.author)
                self.lastbranch[cs.branch] = id
                cs.comment = self.recode(cs.comment)
                if self.ui.configbool('convert', 'localtimezone'):
                    cs.date = makedatetimestamp(cs.date[0])
                date = dateutil.datestr(cs.date, '%Y-%m-%d %H:%M:%S %1%2')
                self.tags.update(dict.fromkeys(cs.tags, id))

                files = {}
                for f in cs.entries:
                    files[f.file] = "%s%s" % ('.'.join([(b"%d" % x)
                                                        for x in f.revision]),
                                              ['', '(DEAD)'][f.dead])

                # add current commit to set
                c = commit(author=cs.author, date=date,
                           parents=[(b"%d" % p.id) for p in cs.parents],
                           desc=cs.comment, branch=cs.branch or '')
                self.changeset[id] = c
                self.files[id] = files

            self.heads = self.lastbranch.values()
        finally:
            os.chdir(d)

    def _connect(self):
        root = self.cvsroot
        conntype = None
        user, host = None, None
        cmd = ['cvs', 'server']

        self.ui.status(_("connecting to %s\n") % root)

        if root.startswith(":pserver:"):
            root = root[9:]
            m = re.match(r'(?:(.*?)(?::(.*?))?@)?([^:\/]*)(?::(\d*))?(.*)',
                         root)
            if m:
                conntype = "pserver"
                user, passw, serv, port, root = m.groups()
                if not user:
                    user = "anonymous"
                if not port:
                    port = 2401
                else:
                    port = int(port)
                format0 = ":pserver:%s@%s:%s" % (user, serv, root)
                format1 = ":pserver:%s@%s:%d%s" % (user, serv, port, root)

                if not passw:
                    passw = "A"
                    cvspass = os.path.expanduser("~/.cvspass")
                    try:
                        pf = open(cvspass, 'rb')
                        for line in pf.read().splitlines():
                            part1, part2 = line.split(' ', 1)
                            # /1 :pserver:[e-mail]:2401/cvsroot/foo
                            # Ah<Z
                            if part1 == '/1':
                                part1, part2 = part2.split(' ', 1)
                                format = format1
                            # :pserver:[e-mail]:/cvsroot/foo Ah<Z
                            else:
                                format = format0
                            if part1 == format:
                                passw = part2
                                break
                        pf.close()
                    except IOError as inst:
                        if inst.errno != errno.ENOENT:
                            if not getattr(inst, 'filename', None):
                                inst.filename = cvspass
                            raise

                sck = socket.socket()
                sck.connect((serv, port))
                sck.send("\n".join(["BEGIN AUTH REQUEST", root, user, passw,
                                    "END AUTH REQUEST", ""]))
                if sck.recv(128) != "I LOVE YOU\n":
                    raise error.Abort(_("CVS pserver authentication failed"))

                self.writep = self.readp = sck.makefile('r+')

        if not conntype and root.startswith(":local:"):
            conntype = "local"
            root = root[7:]

        if not conntype:
            # :ext:user@host/home/user/path/to/cvsroot
            if root.startswith(":ext:"):
                root = root[5:]
            m = re.match(br'(?:([^@:/]+)@)?([^:/]+):?(.*)', root)
            # Do not take Windows path "c:\foo\bar" for a connection strings
            if os.path.isdir(root) or not m:
                conntype = "local"
            else:
                conntype = "rsh"
                user, host, root = m.group(1), m.group(2), m.group(3)

        if conntype != "pserver":
            if conntype == "rsh":
                rsh = encoding.environ.get("CVS_RSH") or "ssh"
                if user:
                    cmd = [rsh, '-l', user, host] + cmd
                else:
                    cmd = [rsh, host] + cmd

            # popen2 does not support argument lists under Windows
            cmd = [procutil.shellquote(arg) for arg in cmd]
            cmd = procutil.quotecommand(' '.join(cmd))
            self.writep, self.readp = procutil.popen2(cmd)

        self.realroot = root

        self.writep.write("Root %s\n" % root)
        self.writep.write("Valid-responses ok error Valid-requests Mode"
                          " M Mbinary E Checked-in Created Updated"
                          " Merged Removed\n")
        self.writep.write("valid-requests\n")
        self.writep.flush()
        r = self.readp.readline()
        if not r.startswith("Valid-requests"):
            raise error.Abort(_('unexpected response from CVS server '
                               '(expected "Valid-requests", but got %r)')
                             % r)
        if "UseUnchanged" in r:
            self.writep.write("UseUnchanged\n")
            self.writep.flush()
            r = self.readp.readline()

    def getheads(self):
        self._parse()
        return self.heads

    def getfile(self, name, rev):

        def chunkedread(fp, count):
            # file-objects returned by socket.makefile() do not handle
            # large read() requests very well.
            chunksize = 65536
            output = stringio()
            while count > 0:
                data = fp.read(min(count, chunksize))
                if not data:
                    raise error.Abort(_("%d bytes missing from remote file")
                                     % count)
                count -= len(data)
                output.write(data)
            return output.getvalue()

        self._parse()
        if rev.endswith("(DEAD)"):
            return None, None

        args = ("-N -P -kk -r %s --" % rev).split()
        args.append(self.cvsrepo + '/' + name)
        for x in args:
            self.writep.write("Argument %s\n" % x)
        self.writep.write("Directory .\n%s\nco\n" % self.realroot)
        self.writep.flush()

        data = ""
        mode = None
        while True:
            line = self.readp.readline()
            if line.startswith("Created ") or line.startswith("Updated "):
                self.readp.readline() # path
                self.readp.readline() # entries
                mode = self.readp.readline()[:-1]
                count = int(self.readp.readline()[:-1])
                data = chunkedread(self.readp, count)
            elif line.startswith(" "):
                data += line[1:]
            elif line.startswith("M "):
                pass
            elif line.startswith("Mbinary "):
                count = int(self.readp.readline()[:-1])
                data = chunkedread(self.readp, count)
            else:
                if line == "ok\n":
                    if mode is None:
                        raise error.Abort(_('malformed response from CVS'))
                    return (data, "x" in mode and "x" or "")
                elif line.startswith("E "):
                    self.ui.warn(_("cvs server: %s\n") % line[2:])
                elif line.startswith("Remove"):
                    self.readp.readline()
                else:
                    raise error.Abort(_("unknown CVS response: %s") % line)

    def getchanges(self, rev, full):
        if full:
            raise error.Abort(_("convert from cvs does not support --full"))
        self._parse()
        return sorted(self.files[rev].iteritems()), {}, set()

    def getcommit(self, rev):
        self._parse()
        return self.changeset[rev]

    def gettags(self):
        self._parse()
        return self.tags

    def getchangedfiles(self, rev, i):
        self._parse()
        return sorted(self.files[rev])