This document describes the current stable version of Celery (5.2). For development docs, go here.
Source code for celery.platforms
"""Platforms.
Utilities dealing with platform specifics: signals, daemonization,
users, groups, and so on.
"""
import atexit
import errno
import math
import numbers
import os
import platform as _platform
import signal as _signal
import sys
import warnings
from collections import namedtuple
from contextlib import contextmanager
from billiard.compat import close_open_fds, get_fdmax
from billiard.util import set_pdeathsig as _set_pdeathsig
# fileno used to be in this module
from kombu.utils.compat import maybe_fileno
from kombu.utils.encoding import safe_str
from .exceptions import SecurityError, SecurityWarning, reraise
from .local import try_import
try:
from billiard.process import current_process
except ImportError: # pragma: no cover
current_process = None
_setproctitle = try_import('setproctitle')
resource = try_import('resource')
pwd = try_import('pwd')
grp = try_import('grp')
mputil = try_import('multiprocessing.util')
__all__ = (
'EX_OK', 'EX_FAILURE', 'EX_UNAVAILABLE', 'EX_USAGE', 'SYSTEM',
'IS_macOS', 'IS_WINDOWS', 'SIGMAP', 'pyimplementation', 'LockFailed',
'get_fdmax', 'Pidfile', 'create_pidlock', 'close_open_fds',
'DaemonContext', 'detached', 'parse_uid', 'parse_gid', 'setgroups',
'initgroups', 'setgid', 'setuid', 'maybe_drop_privileges', 'signals',
'signal_name', 'set_process_title', 'set_mp_process_title',
'get_errno_name', 'ignore_errno', 'fd_by_path', 'isatty',
)
# exitcodes
EX_OK = getattr(os, 'EX_OK', 0)
EX_FAILURE = 1
EX_UNAVAILABLE = getattr(os, 'EX_UNAVAILABLE', 69)
EX_USAGE = getattr(os, 'EX_USAGE', 64)
EX_CANTCREAT = getattr(os, 'EX_CANTCREAT', 73)
SYSTEM = _platform.system()
IS_macOS = SYSTEM == 'Darwin'
IS_WINDOWS = SYSTEM == 'Windows'
DAEMON_WORKDIR = '/'
PIDFILE_FLAGS = os.O_CREAT | os.O_EXCL | os.O_WRONLY
PIDFILE_MODE = ((os.R_OK | os.W_OK) << 6) | ((os.R_OK) << 3) | (os.R_OK)
PIDLOCKED = """ERROR: Pidfile ({0}) already exists.
Seems we're already running? (pid: {1})"""
_range = namedtuple('_range', ('start', 'stop'))
ROOT_DISALLOWED = """\
Running a worker with superuser privileges when the
worker accepts messages serialized with pickle is a very bad idea!
If you really want to continue then you have to set the C_FORCE_ROOT
environment variable (but please think about this before you do).
User information: uid={uid} euid={euid} gid={gid} egid={egid}
"""
ROOT_DISCOURAGED = """\
You're running the worker with superuser privileges: this is
absolutely not recommended!
Please specify a different user using the --uid option.
User information: uid={uid} euid={euid} gid={gid} egid={egid}
"""
ASSUMING_ROOT = """\
An entry for the specified gid or egid was not found.
We're assuming this is a potential security issue.
"""
SIGNAMES = {
sig for sig in dir(_signal)
if sig.startswith('SIG') and '_' not in sig
}
SIGMAP = {getattr(_signal, name): name for name in SIGNAMES}
[docs]def isatty(fh):
"""Return true if the process has a controlling terminal."""
try:
return fh.isatty()
except AttributeError:
pass
[docs]def pyimplementation():
"""Return string identifying the current Python implementation."""
if hasattr(_platform, 'python_implementation'):
return _platform.python_implementation()
elif sys.platform.startswith('java'):
return 'Jython ' + sys.platform
elif hasattr(sys, 'pypy_version_info'):
v = '.'.join(str(p) for p in sys.pypy_version_info[:3])
if sys.pypy_version_info[3:]:
v += '-' + ''.join(str(p) for p in sys.pypy_version_info[3:])
return 'PyPy ' + v
else:
return 'CPython'
[docs]class Pidfile:
"""Pidfile.
This is the type returned by :func:`create_pidlock`.
See Also:
Best practice is to not use this directly but rather use
the :func:`create_pidlock` function instead:
more convenient and also removes stale pidfiles (when
the process holding the lock is no longer running).
"""
#: Path to the pid lock file.
path = None
def __init__(self, path):
self.path = os.path.abspath(path)
[docs] def acquire(self):
"""Acquire lock."""
try:
self.write_pid()
except OSError as exc:
reraise(LockFailed, LockFailed(str(exc)), sys.exc_info()[2])
return self
__enter__ = acquire
[docs] def is_locked(self):
"""Return true if the pid lock exists."""
return os.path.exists(self.path)
__exit__ = release
[docs] def read_pid(self):
"""Read and return the current pid."""
with ignore_errno('ENOENT'):
with open(self.path) as fh:
line = fh.readline()
if line.strip() == line: # must contain '\n'
raise ValueError(
f'Partial or invalid pidfile {self.path}')
try:
return int(line.strip())
except ValueError:
raise ValueError(
f'pidfile {self.path} contents invalid.')
[docs] def remove(self):
"""Remove the lock."""
with ignore_errno(errno.ENOENT, errno.EACCES):
os.unlink(self.path)
[docs] def remove_if_stale(self):
"""Remove the lock if the process isn't running.
I.e. process does not respons to signal.
"""
try:
pid = self.read_pid()
except ValueError:
print('Broken pidfile found - Removing it.', file=sys.stderr)
self.remove()
return True
if not pid:
self.remove()
return True
try:
os.kill(pid, 0)
except os.error as exc:
if exc.errno == errno.ESRCH or exc.errno == errno.EPERM:
print('Stale pidfile exists - Removing it.', file=sys.stderr)
self.remove()
return True
except SystemError:
print('Stale pidfile exists - Removing it.', file=sys.stderr)
self.remove()
return True
return False
[docs] def write_pid(self):
pid = os.getpid()
content = f'{pid}\n'
pidfile_fd = os.open(self.path, PIDFILE_FLAGS, PIDFILE_MODE)
pidfile = os.fdopen(pidfile_fd, 'w')
try:
pidfile.write(content)
# flush and sync so that the re-read below works.
pidfile.flush()
try:
os.fsync(pidfile_fd)
except AttributeError: # pragma: no cover
pass
finally:
pidfile.close()
rfh = open(self.path)
try:
if rfh.read() != content:
raise LockFailed(
"Inconsistency: Pidfile content doesn't match at re-read")
finally:
rfh.close()
PIDFile = Pidfile # XXX compat alias
[docs]def create_pidlock(pidfile):
"""Create and verify pidfile.
If the pidfile already exists the program exits with an error message,
however if the process it refers to isn't running anymore, the pidfile
is deleted and the program continues.
This function will automatically install an :mod:`atexit` handler
to release the lock at exit, you can skip this by calling
:func:`_create_pidlock` instead.
Returns:
Pidfile: used to manage the lock.
Example:
>>> pidlock = create_pidlock('/var/run/app.pid')
"""
pidlock = _create_pidlock(pidfile)
atexit.register(pidlock.release)
return pidlock
def _create_pidlock(pidfile):
pidlock = Pidfile(pidfile)
if pidlock.is_locked() and not pidlock.remove_if_stale():
print(PIDLOCKED.format(pidfile, pidlock.read_pid()), file=sys.stderr)
raise SystemExit(EX_CANTCREAT)
pidlock.acquire()
return pidlock
[docs]def fd_by_path(paths):
"""Return a list of file descriptors.
This method returns list of file descriptors corresponding to
file paths passed in paths variable.
Arguments:
paths: List[str]: List of file paths.
Returns:
List[int]: List of file descriptors.
Example:
>>> keep = fd_by_path(['/dev/urandom', '/my/precious/'])
"""
stats = set()
for path in paths:
try:
fd = os.open(path, os.O_RDONLY)
except OSError:
continue
try:
stats.add(os.fstat(fd)[1:3])
finally:
os.close(fd)
def fd_in_stats(fd):
try:
return os.fstat(fd)[1:3] in stats
except OSError:
return False
return [_fd for _fd in range(get_fdmax(2048)) if fd_in_stats(_fd)]
[docs]class DaemonContext:
"""Context manager daemonizing the process."""
_is_open = False
def __init__(self, pidfile=None, workdir=None, umask=None,
fake=False, after_chdir=None, after_forkers=True,
**kwargs):
if isinstance(umask, str):
# octal or decimal, depending on initial zero.
umask = int(umask, 8 if umask.startswith('0') else 10)
self.workdir = workdir or DAEMON_WORKDIR
self.umask = umask
self.fake = fake
self.after_chdir = after_chdir
self.after_forkers = after_forkers
self.stdfds = (sys.stdin, sys.stdout, sys.stderr)
[docs] def redirect_to_null(self, fd):
if fd is not None:
dest = os.open(os.devnull, os.O_RDWR)
os.dup2(dest, fd)
[docs] def open(self):
if not self._is_open:
if not self.fake:
self._detach()
os.chdir(self.workdir)
if self.umask is not None:
os.umask(self.umask)
if self.after_chdir:
self.after_chdir()
if not self.fake:
# We need to keep /dev/urandom from closing because
# shelve needs it, and Beat needs shelve to start.
keep = list(self.stdfds) + fd_by_path(['/dev/urandom'])
close_open_fds(keep)
for fd in self.stdfds:
self.redirect_to_null(maybe_fileno(fd))
if self.after_forkers and mputil is not None:
mputil._run_after_forkers()
self._is_open = True
__enter__ = open
__exit__ = close
def _detach(self):
if os.fork() == 0: # first child
os.setsid() # create new session
if os.fork() > 0: # pragma: no cover
# second child
os._exit(0)
else:
os._exit(0)
return self
[docs]def detached(logfile=None, pidfile=None, uid=None, gid=None, umask=0,
workdir=None, fake=False, **opts):
"""Detach the current process in the background (daemonize).
Arguments:
logfile (str): Optional log file.
The ability to write to this file
will be verified before the process is detached.
pidfile (str): Optional pid file.
The pidfile won't be created,
as this is the responsibility of the child. But the process will
exit if the pid lock exists and the pid written is still running.
uid (int, str): Optional user id or user name to change
effective privileges to.
gid (int, str): Optional group id or group name to change
effective privileges to.
umask (str, int): Optional umask that'll be effective in
the child process.
workdir (str): Optional new working directory.
fake (bool): Don't actually detach, intended for debugging purposes.
**opts (Any): Ignored.
Example:
>>> from celery.platforms import detached, create_pidlock
>>> with detached(
... logfile='/var/log/app.log',
... pidfile='/var/run/app.pid',
... uid='nobody'):
... # Now in detached child process with effective user set to nobody,
... # and we know that our logfile can be written to, and that
... # the pidfile isn't locked.
... pidlock = create_pidlock('/var/run/app.pid')
...
... # Run the program
... program.run(logfile='/var/log/app.log')
"""
if not resource:
raise RuntimeError('This platform does not support detach.')
workdir = os.getcwd() if workdir is None else workdir
signals.reset('SIGCLD') # Make sure SIGCLD is using the default handler.
maybe_drop_privileges(uid=uid, gid=gid)
def after_chdir_do():
# Since without stderr any errors will be silently suppressed,
# we need to know that we have access to the logfile.
logfile and open(logfile, 'a').close()
# Doesn't actually create the pidfile, but makes sure it's not stale.
if pidfile:
_create_pidlock(pidfile).release()
return DaemonContext(
umask=umask, workdir=workdir, fake=fake, after_chdir=after_chdir_do,
)
[docs]def parse_uid(uid):
"""Parse user id.
Arguments:
uid (str, int): Actual uid, or the username of a user.
Returns:
int: The actual uid.
"""
try:
return int(uid)
except ValueError:
try:
return pwd.getpwnam(uid).pw_uid
except (AttributeError, KeyError):
raise KeyError(f'User does not exist: {uid}')
[docs]def parse_gid(gid):
"""Parse group id.
Arguments:
gid (str, int): Actual gid, or the name of a group.
Returns:
int: The actual gid of the group.
"""
try:
return int(gid)
except ValueError:
try:
return grp.getgrnam(gid).gr_gid
except (AttributeError, KeyError):
raise KeyError(f'Group does not exist: {gid}')
def _setgroups_hack(groups):
# :fun:`setgroups` may have a platform-dependent limit,
# and it's not always possible to know in advance what this limit
# is, so we use this ugly hack stolen from glibc.
groups = groups[:]
while 1:
try:
return os.setgroups(groups)
except ValueError: # error from Python's check.
if len(groups) <= 1:
raise
groups[:] = groups[:-1]
except OSError as exc: # error from the OS.
if exc.errno != errno.EINVAL or len(groups) <= 1:
raise
groups[:] = groups[:-1]
[docs]def setgroups(groups):
"""Set active groups from a list of group ids."""
max_groups = None
try:
max_groups = os.sysconf('SC_NGROUPS_MAX')
except Exception: # pylint: disable=broad-except
pass
try:
return _setgroups_hack(groups[:max_groups])
except OSError as exc:
if exc.errno != errno.EPERM:
raise
if any(group not in groups for group in os.getgroups()):
# we shouldn't be allowed to change to this group.
raise
[docs]def initgroups(uid, gid):
"""Init process group permissions.
Compat version of :func:`os.initgroups` that was first
added to Python 2.7.
"""
if not pwd: # pragma: no cover
return
username = pwd.getpwuid(uid)[0]
if hasattr(os, 'initgroups'): # Python 2.7+
return os.initgroups(username, gid)
groups = [gr.gr_gid for gr in grp.getgrall()
if username in gr.gr_mem]
setgroups(groups)
[docs]def setgid(gid):
"""Version of :func:`os.setgid` supporting group names."""
os.setgid(parse_gid(gid))
[docs]def setuid(uid):
"""Version of :func:`os.setuid` supporting usernames."""
os.setuid(parse_uid(uid))
[docs]def maybe_drop_privileges(uid=None, gid=None):
"""Change process privileges to new user/group.
If UID and GID is specified, the real user/group is changed.
If only UID is specified, the real user is changed, and the group is
changed to the users primary group.
If only GID is specified, only the group is changed.
"""
if sys.platform == 'win32':
return
if os.geteuid():
# no point trying to setuid unless we're root.
if not os.getuid():
raise SecurityError('contact support')
uid = uid and parse_uid(uid)
gid = gid and parse_gid(gid)
if uid:
_setuid(uid, gid)
else:
gid and setgid(gid)
if uid and not os.getuid() and not os.geteuid():
raise SecurityError('Still root uid after drop privileges!')
if gid and not os.getgid() and not os.getegid():
raise SecurityError('Still root gid after drop privileges!')
def _setuid(uid, gid):
# If GID isn't defined, get the primary GID of the user.
if not gid and pwd:
gid = pwd.getpwuid(uid).pw_gid
# Must set the GID before initgroups(), as setgid()
# is known to zap the group list on some platforms.
# setgid must happen before setuid (otherwise the setgid operation
# may fail because of insufficient privileges and possibly stay
# in a privileged group).
setgid(gid)
initgroups(uid, gid)
# at last:
setuid(uid)
# ... and make sure privileges cannot be restored:
try:
setuid(0)
except OSError as exc:
if exc.errno != errno.EPERM:
raise
# we should get here: cannot restore privileges,
# everything was fine.
else:
raise SecurityError(
'non-root user able to restore privileges after setuid.')
if hasattr(_signal, 'setitimer'):
def _arm_alarm(seconds):
_signal.setitimer(_signal.ITIMER_REAL, seconds)
else:
def _arm_alarm(seconds):
_signal.alarm(math.ceil(seconds))
class Signals:
"""Convenience interface to :mod:`signals`.
If the requested signal isn't supported on the current platform,
the operation will be ignored.
Example:
>>> from celery.platforms import signals
>>> from proj.handlers import my_handler
>>> signals['INT'] = my_handler
>>> signals['INT']
my_handler
>>> signals.supported('INT')
True
>>> signals.signum('INT')
2
>>> signals.ignore('USR1')
>>> signals['USR1'] == signals.ignored
True
>>> signals.reset('USR1')
>>> signals['USR1'] == signals.default
True
>>> from proj.handlers import exit_handler, hup_handler
>>> signals.update(INT=exit_handler,
... TERM=exit_handler,
... HUP=hup_handler)
"""
ignored = _signal.SIG_IGN
default = _signal.SIG_DFL
def arm_alarm(self, seconds):
return _arm_alarm(seconds)
def reset_alarm(self):
return _signal.alarm(0)
def supported(self, name):
"""Return true value if signal by ``name`` exists on this platform."""
try:
self.signum(name)
except AttributeError:
return False
else:
return True
def signum(self, name):
"""Get signal number by name."""
if isinstance(name, numbers.Integral):
return name
if not isinstance(name, str) \
or not name.isupper():
raise TypeError('signal name must be uppercase string.')
if not name.startswith('SIG'):
name = 'SIG' + name
return getattr(_signal, name)
def reset(self, *signal_names):
"""Reset signals to the default signal handler.
Does nothing if the platform has no support for signals,
or the specified signal in particular.
"""
self.update((sig, self.default) for sig in signal_names)
def ignore(self, *names):
"""Ignore signal using :const:`SIG_IGN`.
Does nothing if the platform has no support for signals,
or the specified signal in particular.
"""
self.update((sig, self.ignored) for sig in names)
def __getitem__(self, name):
return _signal.getsignal(self.signum(name))
def __setitem__(self, name, handler):
"""Install signal handler.
Does nothing if the current platform has no support for signals,
or the specified signal in particular.
"""
try:
_signal.signal(self.signum(name), handler)
except (AttributeError, ValueError):
pass
def update(self, _d_=None, **sigmap):
"""Set signal handlers from a mapping."""
for name, handler in dict(_d_ or {}, **sigmap).items():
self[name] = handler
signals = Signals()
get_signal = signals.signum # compat
install_signal_handler = signals.__setitem__ # compat
reset_signal = signals.reset # compat
ignore_signal = signals.ignore # compat
[docs]def signal_name(signum):
"""Return name of signal from signal number."""
return SIGMAP[signum][3:]
def strargv(argv):
arg_start = 2 if 'manage' in argv[0] else 1
if len(argv) > arg_start:
return ' '.join(argv[arg_start:])
return ''
def set_pdeathsig(name):
"""Sends signal ``name`` to process when parent process terminates."""
if signals.supported('SIGKILL'):
try:
_set_pdeathsig(signals.signum('SIGKILL'))
except OSError:
# We ignore when OS does not support set_pdeathsig
pass
[docs]def set_process_title(progname, info=None):
"""Set the :command:`ps` name for the currently running process.
Only works if :pypi:`setproctitle` is installed.
"""
proctitle = f'[{progname}]'
proctitle = f'{proctitle} {info}' if info else proctitle
if _setproctitle:
_setproctitle.setproctitle(safe_str(proctitle))
return proctitle
if os.environ.get('NOSETPS'): # pragma: no cover
def set_mp_process_title(*a, **k):
"""Disabled feature."""
else:
[docs] def set_mp_process_title(progname, info=None, hostname=None):
"""Set the :command:`ps` name from the current process name.
Only works if :pypi:`setproctitle` is installed.
"""
if hostname:
progname = f'{progname}: {hostname}'
name = current_process().name if current_process else 'MainProcess'
return set_process_title(f'{progname}:{name}', info=info)
[docs]def get_errno_name(n):
"""Get errno for string (e.g., ``ENOENT``)."""
if isinstance(n, str):
return getattr(errno, n)
return n
[docs]@contextmanager
def ignore_errno(*errnos, **kwargs):
"""Context manager to ignore specific POSIX error codes.
Takes a list of error codes to ignore: this can be either
the name of the code, or the code integer itself::
>>> with ignore_errno('ENOENT'):
... with open('foo', 'r') as fh:
... return fh.read()
>>> with ignore_errno(errno.ENOENT, errno.EPERM):
... pass
Arguments:
types (Tuple[Exception]): A tuple of exceptions to ignore
(when the errno matches). Defaults to :exc:`Exception`.
"""
types = kwargs.get('types') or (Exception,)
errnos = [get_errno_name(errno) for errno in errnos]
try:
yield
except types as exc:
if not hasattr(exc, 'errno'):
raise
if exc.errno not in errnos:
raise
def check_privileges(accept_content):
if grp is None or pwd is None:
return
pickle_or_serialize = ('pickle' in accept_content
or 'application/group-python-serialize' in accept_content)
uid = os.getuid() if hasattr(os, 'getuid') else 65535
gid = os.getgid() if hasattr(os, 'getgid') else 65535
euid = os.geteuid() if hasattr(os, 'geteuid') else 65535
egid = os.getegid() if hasattr(os, 'getegid') else 65535
if hasattr(os, 'fchown'):
if not all(hasattr(os, attr)
for attr in ('getuid', 'getgid', 'geteuid', 'getegid')):
raise SecurityError('suspicious platform, contact support')
# Get the group database entry for the current user's group and effective
# group id using grp.getgrgid() method
# We must handle the case where either the gid or the egid are not found.
try:
gid_entry = grp.getgrgid(gid)
egid_entry = grp.getgrgid(egid)
except KeyError:
warnings.warn(SecurityWarning(ASSUMING_ROOT))
_warn_or_raise_security_error(egid, euid, gid, uid,
pickle_or_serialize)
return
# Get the group and effective group name based on gid
gid_grp_name = gid_entry[0]
egid_grp_name = egid_entry[0]
# Create lists to use in validation step later.
gids_in_use = (gid_grp_name, egid_grp_name)
groups_with_security_risk = ('sudo', 'wheel')
is_root = uid == 0 or euid == 0
# Confirm that the gid and egid are not one that
# can be used to escalate privileges.
if is_root or any(group in gids_in_use
for group in groups_with_security_risk):
_warn_or_raise_security_error(egid, euid, gid, uid,
pickle_or_serialize)
def _warn_or_raise_security_error(egid, euid, gid, uid, pickle_or_serialize):
c_force_root = os.environ.get('C_FORCE_ROOT', False)
if pickle_or_serialize and not c_force_root:
raise SecurityError(ROOT_DISALLOWED.format(
uid=uid, euid=euid, gid=gid, egid=egid,
))
warnings.warn(SecurityWarning(ROOT_DISCOURAGED.format(
uid=uid, euid=euid, gid=gid, egid=egid,
)))
