##// END OF EJS Templates
procutil: avoid using os.fork() to implement runbgcommand...
procutil: avoid using os.fork() to implement runbgcommand We ran into the following deadlock: - some command creates an ssh peer, then raises without explicitly closing the peer (hg id + extension in our case) - dispatch catches the exception, calls ui.log('commandfinish', ..) (the sshpeer is still not closed), which calls logtoprocess, which calls procutil.runbgcommand. - in the child of runbgcommand's fork(), between the fork and the exec, the opening of file descriptors triggers a gc which runs the destructor for sshpeer, which waits on ssh's stderr being closed, which never happens since ssh's stderr is held open by the parent of the fork where said destructor hasn't run Remotefilelog appears to have a hack around this deadlock as well. I don't know if there's more subtlety to it, because even though the problem is determistic, it is very fragile, so I didn't manage to reduce it. I can imagine three ways of tackling this problem: 1. don't run any python between fork and exec in runbgcommand 2. make the finalizer harmless after the fork 3. close the peer without relying on gc behavior This commit goes with 1, as forking without exec'ing is tricky in general in a language with gc finalizers. And maybe it's better in the presence of rust threads. A future commit will try 2 or 3. Performance wise: at low memory usage, it's an improvement. At higher memory usage, it's about 2x faster than before when ensurestart=True, but 2x slower when ensurestart=False. Not sure if that matters. The reason for that last bit is that the subprocess.Popen always waits for the execve to finish, and at high memory usage, execve is slow because it deallocates the large page table. Numbers and script: before after mem=1.0GB, ensurestart=True 52.1ms 26.0ms mem=1.0GB, ensurestart=False 14.7ms 26.0ms mem=0.5GB, ensurestart=True 23.2ms 11.2ms mem=0.5GB, ensurestart=False 6.2ms 11.3ms mem=0.2GB, ensurestart=True 15.7ms 7.4ms mem=0.2GB, ensurestart=False 4.3ms 8.1ms mem=0.0GB, ensurestart=True 2.3ms 0.7ms mem=0.0GB, ensurestart=False 0.8ms 0.8ms import time for memsize in [1_000_000_000, 500_000_000, 250_000_000, 0]: mem = 'a' * memsize for ensurestart in [True, False]: now = time.time() n = 100 for i in range(n): procutil.runbgcommand([b'true'], {}, ensurestart=ensurestart) after = time.time() ms = (after - now) / float(n) * 1000 print(f'mem={memsize / 1e9:.1f}GB, ensurestart={ensurestart} -> {ms:.1f}ms') Differential Revision: https://phab.mercurial-scm.org/D9019

File last commit:

r47552:e2f7b269 merge default
r47651:8759e22f default
Show More
engine.py
587 lines | 18.2 KiB | text/x-python | PythonLexer
# upgrade.py - functions for in place upgrade of Mercurial repository
#
# Copyright (c) 2016-present, Gregory Szorc
#
# 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 stat
from ..i18n import _
from ..pycompat import getattr
from .. import (
changelog,
error,
filelog,
manifest,
metadata,
pycompat,
requirements,
revlog,
scmutil,
util,
vfs as vfsmod,
)
from ..revlogutils import nodemap
def _revlogfrompath(repo, path):
"""Obtain a revlog from a repo path.
An instance of the appropriate class is returned.
"""
if path == b'00changelog.i':
return changelog.changelog(repo.svfs)
elif path.endswith(b'00manifest.i'):
mandir = path[: -len(b'00manifest.i')]
return manifest.manifestrevlog(
repo.nodeconstants, repo.svfs, tree=mandir
)
else:
# reverse of "/".join(("data", path + ".i"))
return filelog.filelog(repo.svfs, path[5:-2])
def _copyrevlog(tr, destrepo, oldrl, unencodedname):
"""copy all relevant files for `oldrl` into `destrepo` store
Files are copied "as is" without any transformation. The copy is performed
without extra checks. Callers are responsible for making sure the copied
content is compatible with format of the destination repository.
"""
oldrl = getattr(oldrl, '_revlog', oldrl)
newrl = _revlogfrompath(destrepo, unencodedname)
newrl = getattr(newrl, '_revlog', newrl)
oldvfs = oldrl.opener
newvfs = newrl.opener
oldindex = oldvfs.join(oldrl.indexfile)
newindex = newvfs.join(newrl.indexfile)
olddata = oldvfs.join(oldrl.datafile)
newdata = newvfs.join(newrl.datafile)
with newvfs(newrl.indexfile, b'w'):
pass # create all the directories
util.copyfile(oldindex, newindex)
copydata = oldrl.opener.exists(oldrl.datafile)
if copydata:
util.copyfile(olddata, newdata)
if not (
unencodedname.endswith(b'00changelog.i')
or unencodedname.endswith(b'00manifest.i')
):
destrepo.svfs.fncache.add(unencodedname)
if copydata:
destrepo.svfs.fncache.add(unencodedname[:-2] + b'.d')
UPGRADE_CHANGELOG = b"changelog"
UPGRADE_MANIFEST = b"manifest"
UPGRADE_FILELOGS = b"all-filelogs"
UPGRADE_ALL_REVLOGS = frozenset(
[UPGRADE_CHANGELOG, UPGRADE_MANIFEST, UPGRADE_FILELOGS]
)
def getsidedatacompanion(srcrepo, dstrepo):
sidedatacompanion = None
removedreqs = srcrepo.requirements - dstrepo.requirements
addedreqs = dstrepo.requirements - srcrepo.requirements
if requirements.SIDEDATA_REQUIREMENT in removedreqs:
def sidedatacompanion(rl, rev):
rl = getattr(rl, '_revlog', rl)
if rl.flags(rev) & revlog.REVIDX_SIDEDATA:
return True, (), {}, 0, 0
return False, (), {}, 0, 0
elif requirements.COPIESSDC_REQUIREMENT in addedreqs:
sidedatacompanion = metadata.getsidedataadder(srcrepo, dstrepo)
elif requirements.COPIESSDC_REQUIREMENT in removedreqs:
sidedatacompanion = metadata.getsidedataremover(srcrepo, dstrepo)
return sidedatacompanion
def matchrevlog(revlogfilter, entry):
"""check if a revlog is selected for cloning.
In other words, are there any updates which need to be done on revlog
or it can be blindly copied.
The store entry is checked against the passed filter"""
if entry.endswith(b'00changelog.i'):
return UPGRADE_CHANGELOG in revlogfilter
elif entry.endswith(b'00manifest.i'):
return UPGRADE_MANIFEST in revlogfilter
return UPGRADE_FILELOGS in revlogfilter
def _perform_clone(
ui,
dstrepo,
tr,
old_revlog,
unencoded,
upgrade_op,
sidedatacompanion,
oncopiedrevision,
):
""" returns the new revlog object created"""
newrl = None
if matchrevlog(upgrade_op.revlogs_to_process, unencoded):
ui.note(
_(b'cloning %d revisions from %s\n') % (len(old_revlog), unencoded)
)
newrl = _revlogfrompath(dstrepo, unencoded)
old_revlog.clone(
tr,
newrl,
addrevisioncb=oncopiedrevision,
deltareuse=upgrade_op.delta_reuse_mode,
forcedeltabothparents=upgrade_op.force_re_delta_both_parents,
sidedatacompanion=sidedatacompanion,
)
else:
msg = _(b'blindly copying %s containing %i revisions\n')
ui.note(msg % (unencoded, len(old_revlog)))
_copyrevlog(tr, dstrepo, old_revlog, unencoded)
newrl = _revlogfrompath(dstrepo, unencoded)
return newrl
def _clonerevlogs(
ui,
srcrepo,
dstrepo,
tr,
upgrade_op,
):
"""Copy revlogs between 2 repos."""
revcount = 0
srcsize = 0
srcrawsize = 0
dstsize = 0
fcount = 0
frevcount = 0
fsrcsize = 0
frawsize = 0
fdstsize = 0
mcount = 0
mrevcount = 0
msrcsize = 0
mrawsize = 0
mdstsize = 0
crevcount = 0
csrcsize = 0
crawsize = 0
cdstsize = 0
alldatafiles = list(srcrepo.store.walk())
# mapping of data files which needs to be cloned
# key is unencoded filename
# value is revlog_object_from_srcrepo
manifests = {}
changelogs = {}
filelogs = {}
# Perform a pass to collect metadata. This validates we can open all
# source files and allows a unified progress bar to be displayed.
for unencoded, encoded, size in alldatafiles:
if not unencoded.endswith(b'.i'):
continue
rl = _revlogfrompath(srcrepo, unencoded)
info = rl.storageinfo(
exclusivefiles=True,
revisionscount=True,
trackedsize=True,
storedsize=True,
)
revcount += info[b'revisionscount'] or 0
datasize = info[b'storedsize'] or 0
rawsize = info[b'trackedsize'] or 0
srcsize += datasize
srcrawsize += rawsize
# This is for the separate progress bars.
if isinstance(rl, changelog.changelog):
changelogs[unencoded] = rl
crevcount += len(rl)
csrcsize += datasize
crawsize += rawsize
elif isinstance(rl, manifest.manifestrevlog):
manifests[unencoded] = rl
mcount += 1
mrevcount += len(rl)
msrcsize += datasize
mrawsize += rawsize
elif isinstance(rl, filelog.filelog):
filelogs[unencoded] = rl
fcount += 1
frevcount += len(rl)
fsrcsize += datasize
frawsize += rawsize
else:
error.ProgrammingError(b'unknown revlog type')
if not revcount:
return
ui.status(
_(
b'migrating %d total revisions (%d in filelogs, %d in manifests, '
b'%d in changelog)\n'
)
% (revcount, frevcount, mrevcount, crevcount)
)
ui.status(
_(b'migrating %s in store; %s tracked data\n')
% ((util.bytecount(srcsize), util.bytecount(srcrawsize)))
)
# Used to keep track of progress.
progress = None
def oncopiedrevision(rl, rev, node):
progress.increment()
sidedatacompanion = getsidedatacompanion(srcrepo, dstrepo)
# Migrating filelogs
ui.status(
_(
b'migrating %d filelogs containing %d revisions '
b'(%s in store; %s tracked data)\n'
)
% (
fcount,
frevcount,
util.bytecount(fsrcsize),
util.bytecount(frawsize),
)
)
progress = srcrepo.ui.makeprogress(_(b'file revisions'), total=frevcount)
for unencoded, oldrl in sorted(filelogs.items()):
newrl = _perform_clone(
ui,
dstrepo,
tr,
oldrl,
unencoded,
upgrade_op,
sidedatacompanion,
oncopiedrevision,
)
info = newrl.storageinfo(storedsize=True)
fdstsize += info[b'storedsize'] or 0
ui.status(
_(
b'finished migrating %d filelog revisions across %d '
b'filelogs; change in size: %s\n'
)
% (frevcount, fcount, util.bytecount(fdstsize - fsrcsize))
)
# Migrating manifests
ui.status(
_(
b'migrating %d manifests containing %d revisions '
b'(%s in store; %s tracked data)\n'
)
% (
mcount,
mrevcount,
util.bytecount(msrcsize),
util.bytecount(mrawsize),
)
)
if progress:
progress.complete()
progress = srcrepo.ui.makeprogress(
_(b'manifest revisions'), total=mrevcount
)
for unencoded, oldrl in sorted(manifests.items()):
newrl = _perform_clone(
ui,
dstrepo,
tr,
oldrl,
unencoded,
upgrade_op,
sidedatacompanion,
oncopiedrevision,
)
info = newrl.storageinfo(storedsize=True)
mdstsize += info[b'storedsize'] or 0
ui.status(
_(
b'finished migrating %d manifest revisions across %d '
b'manifests; change in size: %s\n'
)
% (mrevcount, mcount, util.bytecount(mdstsize - msrcsize))
)
# Migrating changelog
ui.status(
_(
b'migrating changelog containing %d revisions '
b'(%s in store; %s tracked data)\n'
)
% (
crevcount,
util.bytecount(csrcsize),
util.bytecount(crawsize),
)
)
if progress:
progress.complete()
progress = srcrepo.ui.makeprogress(
_(b'changelog revisions'), total=crevcount
)
for unencoded, oldrl in sorted(changelogs.items()):
newrl = _perform_clone(
ui,
dstrepo,
tr,
oldrl,
unencoded,
upgrade_op,
sidedatacompanion,
oncopiedrevision,
)
info = newrl.storageinfo(storedsize=True)
cdstsize += info[b'storedsize'] or 0
progress.complete()
ui.status(
_(
b'finished migrating %d changelog revisions; change in size: '
b'%s\n'
)
% (crevcount, util.bytecount(cdstsize - csrcsize))
)
dstsize = fdstsize + mdstsize + cdstsize
ui.status(
_(
b'finished migrating %d total revisions; total change in store '
b'size: %s\n'
)
% (revcount, util.bytecount(dstsize - srcsize))
)
def _files_to_copy_post_revlog_clone(srcrepo):
"""yields files which should be copied to destination after revlogs
are cloned"""
for path, kind, st in sorted(srcrepo.store.vfs.readdir(b'', stat=True)):
# don't copy revlogs as they are already cloned
if path.endswith((b'.i', b'.d', b'.n', b'.nd')):
continue
# Skip transaction related files.
if path.startswith(b'undo'):
continue
# Only copy regular files.
if kind != stat.S_IFREG:
continue
# Skip other skipped files.
if path in (b'lock', b'fncache'):
continue
# TODO: should we skip cache too?
yield path
def _replacestores(currentrepo, upgradedrepo, backupvfs, upgrade_op):
"""Replace the stores after current repository is upgraded
Creates a backup of current repository store at backup path
Replaces upgraded store files in current repo from upgraded one
Arguments:
currentrepo: repo object of current repository
upgradedrepo: repo object of the upgraded data
backupvfs: vfs object for the backup path
upgrade_op: upgrade operation object
to be used to decide what all is upgraded
"""
# TODO: don't blindly rename everything in store
# There can be upgrades where store is not touched at all
if upgrade_op.backup_store:
util.rename(currentrepo.spath, backupvfs.join(b'store'))
else:
currentrepo.vfs.rmtree(b'store', forcibly=True)
util.rename(upgradedrepo.spath, currentrepo.spath)
def finishdatamigration(ui, srcrepo, dstrepo, requirements):
"""Hook point for extensions to perform additional actions during upgrade.
This function is called after revlogs and store files have been copied but
before the new store is swapped into the original location.
"""
def upgrade(ui, srcrepo, dstrepo, upgrade_op):
"""Do the low-level work of upgrading a repository.
The upgrade is effectively performed as a copy between a source
repository and a temporary destination repository.
The source repository is unmodified for as long as possible so the
upgrade can abort at any time without causing loss of service for
readers and without corrupting the source repository.
"""
assert srcrepo.currentwlock()
assert dstrepo.currentwlock()
backuppath = None
backupvfs = None
ui.status(
_(
b'(it is safe to interrupt this process any time before '
b'data migration completes)\n'
)
)
if upgrade_op.requirements_only:
ui.status(_(b'upgrading repository requirements\n'))
scmutil.writereporequirements(srcrepo, upgrade_op.new_requirements)
# if there is only one action and that is persistent nodemap upgrade
# directly write the nodemap file and update requirements instead of going
# through the whole cloning process
elif (
len(upgrade_op.upgrade_actions) == 1
and b'persistent-nodemap' in upgrade_op._upgrade_actions_names
and not upgrade_op.removed_actions
):
ui.status(
_(b'upgrading repository to use persistent nodemap feature\n')
)
with srcrepo.transaction(b'upgrade') as tr:
unfi = srcrepo.unfiltered()
cl = unfi.changelog
nodemap.persist_nodemap(tr, cl, force=True)
# we want to directly operate on the underlying revlog to force
# create a nodemap file. This is fine since this is upgrade code
# and it heavily relies on repository being revlog based
# hence accessing private attributes can be justified
nodemap.persist_nodemap(
tr, unfi.manifestlog._rootstore._revlog, force=True
)
scmutil.writereporequirements(srcrepo, upgrade_op.new_requirements)
elif (
len(upgrade_op.removed_actions) == 1
and [
x
for x in upgrade_op.removed_actions
if x.name == b'persistent-nodemap'
]
and not upgrade_op.upgrade_actions
):
ui.status(
_(b'downgrading repository to not use persistent nodemap feature\n')
)
with srcrepo.transaction(b'upgrade') as tr:
unfi = srcrepo.unfiltered()
cl = unfi.changelog
nodemap.delete_nodemap(tr, srcrepo, cl)
# check comment 20 lines above for accessing private attributes
nodemap.delete_nodemap(
tr, srcrepo, unfi.manifestlog._rootstore._revlog
)
scmutil.writereporequirements(srcrepo, upgrade_op.new_requirements)
else:
with dstrepo.transaction(b'upgrade') as tr:
_clonerevlogs(
ui,
srcrepo,
dstrepo,
tr,
upgrade_op,
)
# Now copy other files in the store directory.
for p in _files_to_copy_post_revlog_clone(srcrepo):
srcrepo.ui.status(_(b'copying %s\n') % p)
src = srcrepo.store.rawvfs.join(p)
dst = dstrepo.store.rawvfs.join(p)
util.copyfile(src, dst, copystat=True)
finishdatamigration(ui, srcrepo, dstrepo, requirements)
ui.status(_(b'data fully upgraded in a temporary repository\n'))
if upgrade_op.backup_store:
backuppath = pycompat.mkdtemp(
prefix=b'upgradebackup.', dir=srcrepo.path
)
backupvfs = vfsmod.vfs(backuppath)
# Make a backup of requires file first, as it is the first to be modified.
util.copyfile(
srcrepo.vfs.join(b'requires'), backupvfs.join(b'requires')
)
# We install an arbitrary requirement that clients must not support
# as a mechanism to lock out new clients during the data swap. This is
# better than allowing a client to continue while the repository is in
# an inconsistent state.
ui.status(
_(
b'marking source repository as being upgraded; clients will be '
b'unable to read from repository\n'
)
)
scmutil.writereporequirements(
srcrepo, srcrepo.requirements | {b'upgradeinprogress'}
)
ui.status(_(b'starting in-place swap of repository data\n'))
if upgrade_op.backup_store:
ui.status(
_(b'replaced files will be backed up at %s\n') % backuppath
)
# Now swap in the new store directory. Doing it as a rename should make
# the operation nearly instantaneous and atomic (at least in well-behaved
# environments).
ui.status(_(b'replacing store...\n'))
tstart = util.timer()
_replacestores(srcrepo, dstrepo, backupvfs, upgrade_op)
elapsed = util.timer() - tstart
ui.status(
_(
b'store replacement complete; repository was inconsistent for '
b'%0.1fs\n'
)
% elapsed
)
# We first write the requirements file. Any new requirements will lock
# out legacy clients.
ui.status(
_(
b'finalizing requirements file and making repository readable '
b'again\n'
)
)
scmutil.writereporequirements(srcrepo, upgrade_op.new_requirements)
if upgrade_op.backup_store:
# The lock file from the old store won't be removed because nothing has a
# reference to its new location. So clean it up manually. Alternatively, we
# could update srcrepo.svfs and other variables to point to the new
# location. This is simpler.
assert backupvfs is not None # help pytype
backupvfs.unlink(b'store/lock')
return backuppath