view hgext/blackbox.py @ 26380:56a640b0f656

revlog: don't flush data file after every added revision

The current behavior of revlogs is to flush the data file when writing
data to it. Tracing system calls revealed that changegroup processing
incurred numerous write(2) calls for values much smaller than the
default buffer size (Python defaults to 4096, but it can be adjusted
based on detected block size at run time by CPython).

The reason we flush revlogs is so readers have all data available.
For example, the current code in revlog.py will re-open the revlog
file (instead of seeking an existing file handle) to read the text
of a revision. This happens when starting a new delta chain when
adding several revisions from changegroups, for example. Yes, this
is likely sub-optimal (we should probably be sharing file descriptors
between readers and writers to avoid the flushing and associated
overhead of re-opening files).

While flushing revlogs is necessary, it appears all callers are
diligent about flushing files before a read is performed (see
buildtext() in _addrevision()), making the flush in
_writeentry() redundant and unncessary. So, we remove it. In practice,
this means we incur a write(2) a) when the buffer is full (typically
4096 bytes) b) when a new delta chain is created rather than after
every added revision. This applies to every revlog, but by volume
it mostly impacts filelogs.

Removing the redundant flush from _writeentry() significantly
reduces the number of write(2) calls during changegroup processing on
my Linux machine. When applying a changegroup of the hg repo based on
my local repo, the total number of write(2) calls during application
of the mercurial/localrepo.py revlogs dropped from 1,320 to 217 with
this patch applied. Total I/O related system calls dropped from 1,577
to 474.

When unbundling a mozilla-central gzipped bundle (264,403 changesets
with 1,492,215 changes to 222,507 files), total write(2) calls
dropped from 1,252,881 to 827,106 and total system calls dropped from
3,601,259 to 3,178,636 - a reduction of 425,775!

While the system call reduction is significant, it appears
to have no impact on wall time on my Linux and Windows machines. Still,
fewer syscalls is fewer syscalls. Surely this can't hurt. If nothing
else, it makes examining remaining system call usage simpler and opens
the door to experimenting with the performance impact of different
buffer sizes.

# blackbox.py - log repository events to a file for post-mortem debugging
#
# Copyright 2010 Nicolas Dumazet
# Copyright 2013 Facebook, Inc.
#
# This software may be used and distributed according to the terms of the
# GNU General Public License version 2 or any later version.

"""log repository events to a blackbox for debugging

Logs event information to .hg/blackbox.log to help debug and diagnose problems.
The events that get logged can be configured via the blackbox.track config key.
Examples::

  [blackbox]
  track = *

  [blackbox]
  track = command, commandfinish, commandexception, exthook, pythonhook

  [blackbox]
  track = incoming

  [blackbox]
  # limit the size of a log file
  maxsize = 1.5 MB
  # rotate up to N log files when the current one gets too big
  maxfiles = 3

"""

from mercurial import util, cmdutil
from mercurial.i18n import _
import errno, os, re

cmdtable = {}
command = cmdutil.command(cmdtable)
# Note for extension authors: ONLY specify testedwith = 'internal' for
# extensions which SHIP WITH MERCURIAL. Non-mainline extensions should
# be specifying the version(s) of Mercurial they are tested with, or
# leave the attribute unspecified.
testedwith = 'internal'
lastblackbox = None

def wrapui(ui):
    class blackboxui(ui.__class__):
        @util.propertycache
        def track(self):
            return self.configlist('blackbox', 'track', ['*'])

        def _openlogfile(self):
            def rotate(oldpath, newpath):
                try:
                    os.unlink(newpath)
                except OSError as err:
                    if err.errno != errno.ENOENT:
                        self.debug("warning: cannot remove '%s': %s\n" %
                                   (newpath, err.strerror))
                try:
                    if newpath:
                        os.rename(oldpath, newpath)
                except OSError as err:
                    if err.errno != errno.ENOENT:
                        self.debug("warning: cannot rename '%s' to '%s': %s\n" %
                                   (newpath, oldpath, err.strerror))

            fp = self._bbopener('blackbox.log', 'a')
            maxsize = self.configbytes('blackbox', 'maxsize', 1048576)
            if maxsize > 0:
                st = os.fstat(fp.fileno())
                if st.st_size >= maxsize:
                    path = fp.name
                    fp.close()
                    maxfiles = self.configint('blackbox', 'maxfiles', 7)
                    for i in xrange(maxfiles - 1, 1, -1):
                        rotate(oldpath='%s.%d' % (path, i - 1),
                               newpath='%s.%d' % (path, i))
                    rotate(oldpath=path,
                           newpath=maxfiles > 0 and path + '.1')
                    fp = self._bbopener('blackbox.log', 'a')
            return fp

        def log(self, event, *msg, **opts):
            global lastblackbox
            super(blackboxui, self).log(event, *msg, **opts)

            if not '*' in self.track and not event in self.track:
                return

            if util.safehasattr(self, '_blackbox'):
                blackbox = self._blackbox
            elif util.safehasattr(self, '_bbopener'):
                try:
                    self._blackbox = self._openlogfile()
                except (IOError, OSError) as err:
                    self.debug('warning: cannot write to blackbox.log: %s\n' %
                               err.strerror)
                    del self._bbopener
                    self._blackbox = None
                blackbox = self._blackbox
            else:
                # certain ui instances exist outside the context of
                # a repo, so just default to the last blackbox that
                # was seen.
                blackbox = lastblackbox

            if blackbox:
                date = util.datestr(None, '%Y/%m/%d %H:%M:%S')
                user = util.getuser()
                pid = str(os.getpid())
                formattedmsg = msg[0] % msg[1:]
                try:
                    blackbox.write('%s %s (%s)> %s' %
                                   (date, user, pid, formattedmsg))
                except IOError as err:
                    self.debug('warning: cannot write to blackbox.log: %s\n' %
                               err.strerror)
                lastblackbox = blackbox

        def setrepo(self, repo):
            self._bbopener = repo.vfs

    ui.__class__ = blackboxui

def uisetup(ui):
    wrapui(ui)

def reposetup(ui, repo):
    # During 'hg pull' a httppeer repo is created to represent the remote repo.
    # It doesn't have a .hg directory to put a blackbox in, so we don't do
    # the blackbox setup for it.
    if not repo.local():
        return

    if util.safehasattr(ui, 'setrepo'):
        ui.setrepo(repo)

@command('^blackbox',
    [('l', 'limit', 10, _('the number of events to show')),
    ],
    _('hg blackbox [OPTION]...'))
def blackbox(ui, repo, *revs, **opts):
    '''view the recent repository events
    '''

    if not os.path.exists(repo.join('blackbox.log')):
        return

    limit = opts.get('limit')
    blackbox = repo.vfs('blackbox.log', 'r')
    lines = blackbox.read().split('\n')

    count = 0
    output = []
    for line in reversed(lines):
        if count >= limit:
            break

        # count the commands by matching lines like: 2013/01/23 19:13:36 root>
        if re.match('^\d{4}/\d{2}/\d{2} \d{2}:\d{2}:\d{2} .*> .*', line):
            count += 1
        output.append(line)

    ui.status('\n'.join(reversed(output)))