services/fuse/arvados_fuse/command.py

   1 # Copyright (C) The Arvados Authors. All rights reserved.
   2 #
   3 # SPDX-License-Identifier: AGPL-3.0
   4
   5 import argparse
   6 import arvados
   7 import daemon
   8 import llfuse
   9 import logging
  10 import os
  11 import resource
  12 import signal
  13 import subprocess
  14 import sys
  15 import time
  16
  17 import arvados.commands._util as arv_cmd
  18 from arvados_fuse import crunchstat
  19 from arvados_fuse import *
  20 from arvados_fuse.unmount import unmount
  21 from arvados_fuse._version import __version__
  22
  23 class ArgumentParser(argparse.ArgumentParser):
  24     def __init__(self):
  25         super(ArgumentParser, self).__init__(
  26             parents=[arv_cmd.retry_opt],
  27             description='''Mount Keep data under the local filesystem.  Default mode is --home''',
  28             epilog="""
  29     Note: When using the --exec feature, you must either specify the
  30     mountpoint before --exec, or mark the end of your --exec arguments
  31     with "--".
  32             """)
  33         self.add_argument('--version', action='version',
  34                           version="%s %s" % (sys.argv[0], __version__),
  35                           help='Print version and exit.')
  36         self.add_argument('mountpoint', type=str, help="""Mount point.""")
  37         self.add_argument('--allow-other', action='store_true',
  38                             help="""Let other users read the mount""")
  39         self.add_argument('--subtype', type=str, metavar='STRING',
  40                             help="""Report mounted filesystem type as "fuse.STRING", instead of just "fuse".""")
  41
  42         mode = self.add_mutually_exclusive_group()
  43
  44         mode.add_argument('--all', action='store_const', const='all', dest='mode',
  45                                 help="""Mount a subdirectory for each mode: home, shared, by_tag, by_id (default if no --mount-* arguments are given).""")
  46         mode.add_argument('--custom', action='store_const', const=None, dest='mode',
  47                                 help="""Mount a top level meta-directory with subdirectories as specified by additional --mount-* arguments (default if any --mount-* arguments are given).""")
  48         mode.add_argument('--home', action='store_const', const='home', dest='mode',
  49                                 help="""Mount only the user's home project.""")
  50         mode.add_argument('--shared', action='store_const', const='shared', dest='mode',
  51                                 help="""Mount only list of projects shared with the user.""")
  52         mode.add_argument('--by-tag', action='store_const', const='by_tag', dest='mode',
  53                                 help="""Mount subdirectories listed by tag.""")
  54         mode.add_argument('--by-id', action='store_const', const='by_id', dest='mode',
  55                                 help="""Mount subdirectories listed by portable data hash or uuid.""")
  56         mode.add_argument('--by-pdh', action='store_const', const='by_pdh', dest='mode',
  57                                 help="""Mount subdirectories listed by portable data hash.""")
  58         mode.add_argument('--project', type=str, metavar='UUID',
  59                                 help="""Mount the specified project.""")
  60         mode.add_argument('--collection', type=str, metavar='UUID_or_PDH',
  61                                 help="""Mount only the specified collection.""")
  62
  63         mounts = self.add_argument_group('Custom mount options')
  64         mounts.add_argument('--mount-by-pdh',
  65                             type=str, metavar='PATH', action='append', default=[],
  66                             help="Mount each readable collection at mountpoint/PATH/P where P is the collection's portable data hash.")
  67         mounts.add_argument('--mount-by-id',
  68                             type=str, metavar='PATH', action='append', default=[],
  69                             help="Mount each readable collection at mountpoint/PATH/UUID and mountpoint/PATH/PDH where PDH is the collection's portable data hash and UUID is its UUID.")
  70         mounts.add_argument('--mount-by-tag',
  71                             type=str, metavar='PATH', action='append', default=[],
  72                             help="Mount all collections with tag TAG at mountpoint/PATH/TAG/UUID.")
  73         mounts.add_argument('--mount-home',
  74                             type=str, metavar='PATH', action='append', default=[],
  75                             help="Mount the current user's home project at mountpoint/PATH.")
  76         mounts.add_argument('--mount-shared',
  77                             type=str, metavar='PATH', action='append', default=[],
  78                             help="Mount projects shared with the current user at mountpoint/PATH.")
  79         mounts.add_argument('--mount-tmp',
  80                             type=str, metavar='PATH', action='append', default=[],
  81                             help="Create a new collection, mount it in read/write mode at mountpoint/PATH, and delete it when unmounting.")
  82
  83         self.add_argument('--debug', action='store_true', help="""Debug mode""")
  84         self.add_argument('--logfile', help="""Write debug logs and errors to the specified file (default stderr).""")
  85         self.add_argument('--foreground', action='store_true', help="""Run in foreground (default is to daemonize unless --exec specified)""", default=False)
  86         self.add_argument('--encoding', type=str, help="Character encoding to use for filesystem, default is utf-8 (see Python codec registry for list of available encodings)", default="utf-8")
  87
  88         self.add_argument('--file-cache', type=int, help="File data cache size, in bytes (default 256MiB)", default=256*1024*1024)
  89         self.add_argument('--directory-cache', type=int, help="Directory data cache size, in bytes (default 128MiB)", default=128*1024*1024)
  90
  91         self.add_argument('--disable-event-listening', action='store_true', help="Don't subscribe to events on the API server", dest="disable_event_listening", default=False)
  92
  93         self.add_argument('--read-only', action='store_false', help="Mount will be read only (default)", dest="enable_write", default=False)
  94         self.add_argument('--read-write', action='store_true', help="Mount will be read-write", dest="enable_write", default=False)
  95
  96         self.add_argument('--crunchstat-interval', type=float, help="Write stats to stderr every N seconds (default disabled)", default=0)
  97
  98         unmount = self.add_mutually_exclusive_group()
  99         unmount.add_argument('--unmount', action='store_true', default=False,
 100                              help="Forcefully unmount the specified mountpoint (if it's a fuse mount) and exit. If --subtype is given, unmount only if the mount has the specified subtype. WARNING: This command can affect any kind of fuse mount, not just arv-mount.")
 101         unmount.add_argument('--unmount-all', action='store_true', default=False,
 102                              help="Forcefully unmount every fuse mount at or below the specified path and exit. If --subtype is given, unmount only mounts that have the specified subtype. Exit non-zero if any other types of mounts are found at or below the given path. WARNING: This command can affect any kind of fuse mount, not just arv-mount.")
 103         unmount.add_argument('--replace', action='store_true', default=False,
 104                              help="If a fuse mount is already present at mountpoint, forcefully unmount it before mounting")
 105         self.add_argument('--unmount-timeout',
 106                           type=float, default=2.0,
 107                           help="Time to wait for graceful shutdown after --exec program exits and filesystem is unmounted")
 108
 109         self.add_argument('--exec', type=str, nargs=argparse.REMAINDER,
 110                             dest="exec_args", metavar=('command', 'args', '...', '--'),
 111                             help="""Mount, run a command, then unmount and exit""")
 112
 113
 114 class Mount(object):
 115     def __init__(self, args, logger=logging.getLogger('arvados.arv-mount')):
 116         self.daemon = False
 117         self.logger = logger
 118         self.args = args
 119         self.listen_for_events = False
 120
 121         self.args.mountpoint = os.path.realpath(self.args.mountpoint)
 122         if self.args.logfile:
 123             self.args.logfile = os.path.realpath(self.args.logfile)
 124
 125         try:
 126             self._setup_logging()
 127             self._setup_api()
 128             self._setup_mount()
 129         except Exception as e:
 130             self.logger.exception("arv-mount: exception during setup: %s", e)
 131             exit(1)
 132
 133     def __enter__(self):
 134         if self.args.replace:
 135             unmount(path=self.args.mountpoint,
 136                     timeout=self.args.unmount_timeout)
 137         llfuse.init(self.operations, self.args.mountpoint, self._fuse_options())
 138         if self.daemon:
 139             daemon.DaemonContext(
 140                 working_directory=os.path.dirname(self.args.mountpoint),
 141                 files_preserve=range(
 142                     3, resource.getrlimit(resource.RLIMIT_NOFILE)[1])
 143             ).open()
 144         if self.listen_for_events and not self.args.disable_event_listening:
 145             self.operations.listen_for_events()
 146         self.llfuse_thread = threading.Thread(None, lambda: self._llfuse_main())
 147         self.llfuse_thread.daemon = True
 148         self.llfuse_thread.start()
 149         self.operations.initlock.wait()
 150         return self
 151
 152     def __exit__(self, exc_type, exc_value, traceback):
 153         if self.operations.events:
 154             self.operations.events.close(timeout=self.args.unmount_timeout)
 155         subprocess.call(["fusermount", "-u", "-z", self.args.mountpoint])
 156         self.llfuse_thread.join(timeout=self.args.unmount_timeout)
 157         if self.llfuse_thread.is_alive():
 158             self.logger.warning("Mount.__exit__:"
 159                                 " llfuse thread still alive %fs after umount"
 160                                 " -- abandoning and exiting anyway",
 161                                 self.args.unmount_timeout)
 162
 163     def run(self):
 164         if self.args.unmount or self.args.unmount_all:
 165             unmount(path=self.args.mountpoint,
 166                     subtype=self.args.subtype,
 167                     timeout=self.args.unmount_timeout,
 168                     recursive=self.args.unmount_all)
 169         elif self.args.exec_args:
 170             self._run_exec()
 171         else:
 172             self._run_standalone()
 173
 174     def _fuse_options(self):
 175         """FUSE mount options; see mount.fuse(8)"""
 176         opts = [optname for optname in ['allow_other', 'debug']
 177                 if getattr(self.args, optname)]
 178         # Increase default read/write size from 4KiB to 128KiB
 179         opts += ["big_writes", "max_read=131072"]
 180         if self.args.subtype:
 181             opts += ["subtype="+self.args.subtype]
 182         return opts
 183
 184     def _setup_logging(self):
 185         # Configure a log handler based on command-line switches.
 186         if self.args.logfile:
 187             log_handler = logging.FileHandler(self.args.logfile)
 188             log_handler.setFormatter(logging.Formatter(
 189                 '%(asctime)s %(name)s[%(process)d] %(levelname)s: %(message)s',
 190                 '%Y-%m-%d %H:%M:%S'))
 191         else:
 192             log_handler = None
 193
 194         if log_handler is not None:
 195             arvados.logger.removeHandler(arvados.log_handler)
 196             arvados.logger.addHandler(log_handler)
 197
 198         if self.args.debug:
 199             arvados.logger.setLevel(logging.DEBUG)
 200             logging.getLogger('arvados.keep').setLevel(logging.DEBUG)
 201             logging.getLogger('arvados.api').setLevel(logging.DEBUG)
 202             logging.getLogger('arvados.collection').setLevel(logging.DEBUG)
 203             self.logger.debug("arv-mount debugging enabled")
 204
 205         self.logger.info("%s %s started", sys.argv[0], __version__)
 206         self.logger.info("enable write is %s", self.args.enable_write)
 207
 208     def _setup_api(self):
 209         try:
 210             self.api = arvados.safeapi.ThreadSafeApiCache(
 211                 apiconfig=arvados.config.settings(),
 212                 keep_params={
 213                     'block_cache': arvados.keep.KeepBlockCache(self.args.file_cache),
 214                     'num_retries': self.args.retries,
 215                 })
 216         except KeyError as e:
 217             self.logger.error("Missing environment: %s", e)
 218             exit(1)
 219         # Do a sanity check that we have a working arvados host + token.
 220         self.api.users().current().execute()
 221
 222     def _setup_mount(self):
 223         self.operations = Operations(
 224             os.getuid(),
 225             os.getgid(),
 226             api_client=self.api,
 227             encoding=self.args.encoding,
 228             inode_cache=InodeCache(cap=self.args.directory_cache),
 229             enable_write=self.args.enable_write)
 230
 231         if self.args.crunchstat_interval:
 232             statsthread = threading.Thread(
 233                 target=crunchstat.statlogger,
 234                 args=(self.args.crunchstat_interval,
 235                       self.api.keep,
 236                       self.operations))
 237             statsthread.daemon = True
 238             statsthread.start()
 239
 240         usr = self.api.users().current().execute(num_retries=self.args.retries)
 241         now = time.time()
 242         dir_class = None
 243         dir_args = [llfuse.ROOT_INODE, self.operations.inodes, self.api, self.args.retries]
 244         mount_readme = False
 245
 246         if self.args.collection is not None:
 247             # Set up the request handler with the collection at the root
 248             # First check that the collection is readable
 249             self.api.collections().get(uuid=self.args.collection).execute()
 250             self.args.mode = 'collection'
 251             dir_class = CollectionDirectory
 252             dir_args.append(self.args.collection)
 253         elif self.args.project is not None:
 254             self.args.mode = 'project'
 255             dir_class = ProjectDirectory
 256             dir_args.append(
 257                 self.api.groups().get(uuid=self.args.project).execute(
 258                     num_retries=self.args.retries))
 259
 260         if (self.args.mount_by_id or
 261             self.args.mount_by_pdh or
 262             self.args.mount_by_tag or
 263             self.args.mount_home or
 264             self.args.mount_shared or
 265             self.args.mount_tmp):
 266             if self.args.mode is not None:
 267                 sys.exit(
 268                     "Cannot combine '{}' mode with custom --mount-* options.".
 269                     format(self.args.mode))
 270         elif self.args.mode is None:
 271             # If no --mount-custom or custom mount args, --all is the default
 272             self.args.mode = 'all'
 273
 274         if self.args.mode in ['by_id', 'by_pdh']:
 275             # Set up the request handler with the 'magic directory' at the root
 276             dir_class = MagicDirectory
 277             dir_args.append(self.args.mode == 'by_pdh')
 278         elif self.args.mode == 'by_tag':
 279             dir_class = TagsDirectory
 280         elif self.args.mode == 'shared':
 281             dir_class = SharedDirectory
 282             dir_args.append(usr)
 283         elif self.args.mode == 'home':
 284             dir_class = ProjectDirectory
 285             dir_args.append(usr)
 286             dir_args.append(True)
 287         elif self.args.mode == 'all':
 288             self.args.mount_by_id = ['by_id']
 289             self.args.mount_by_tag = ['by_tag']
 290             self.args.mount_home = ['home']
 291             self.args.mount_shared = ['shared']
 292             mount_readme = True
 293
 294         if dir_class is not None:
 295             ent = dir_class(*dir_args)
 296             self.operations.inodes.add_entry(ent)
 297             self.listen_for_events = ent.want_event_subscribe()
 298             return
 299
 300         e = self.operations.inodes.add_entry(Directory(
 301             llfuse.ROOT_INODE, self.operations.inodes))
 302         dir_args[0] = e.inode
 303
 304         for name in self.args.mount_by_id:
 305             self._add_mount(e, name, MagicDirectory(*dir_args, pdh_only=False))
 306         for name in self.args.mount_by_pdh:
 307             self._add_mount(e, name, MagicDirectory(*dir_args, pdh_only=True))
 308         for name in self.args.mount_by_tag:
 309             self._add_mount(e, name, TagsDirectory(*dir_args))
 310         for name in self.args.mount_home:
 311             self._add_mount(e, name, ProjectDirectory(*dir_args, project_object=usr, poll=True))
 312         for name in self.args.mount_shared:
 313             self._add_mount(e, name, SharedDirectory(*dir_args, exclude=usr, poll=True))
 314         for name in self.args.mount_tmp:
 315             self._add_mount(e, name, TmpCollectionDirectory(*dir_args))
 316
 317         if mount_readme:
 318             text = self._readme_text(
 319                 arvados.config.get('ARVADOS_API_HOST'),
 320                 usr['email'])
 321             self._add_mount(e, 'README', StringFile(e.inode, text, now))
 322
 323     def _add_mount(self, tld, name, ent):
 324         if name in ['', '.', '..'] or '/' in name:
 325             sys.exit("Mount point '{}' is not supported.".format(name))
 326         tld._entries[name] = self.operations.inodes.add_entry(ent)
 327         self.listen_for_events = (self.listen_for_events or ent.want_event_subscribe())
 328
 329     def _readme_text(self, api_host, user_email):
 330         return '''
 331 Welcome to Arvados!  This directory provides file system access to
 332 files and objects available on the Arvados installation located at
 333 '{}' using credentials for user '{}'.
 334
 335 From here, the following directories are available:
 336
 337   by_id/     Access to Keep collections by uuid or portable data hash (see by_id/README for details).
 338   by_tag/    Access to Keep collections organized by tag.
 339   home/      The contents of your home project.
 340   shared/    Projects shared with you.
 341
 342 '''.format(api_host, user_email)
 343
 344     def _run_exec(self):
 345         rc = 255
 346         with self:
 347             try:
 348                 sp = subprocess.Popen(self.args.exec_args, shell=False)
 349
 350                 # forward signals to the process.
 351                 signal.signal(signal.SIGINT, lambda signum, frame: sp.send_signal(signum))
 352                 signal.signal(signal.SIGTERM, lambda signum, frame: sp.send_signal(signum))
 353                 signal.signal(signal.SIGQUIT, lambda signum, frame: sp.send_signal(signum))
 354
 355                 # wait for process to complete.
 356                 rc = sp.wait()
 357
 358                 # restore default signal handlers.
 359                 signal.signal(signal.SIGINT, signal.SIG_DFL)
 360                 signal.signal(signal.SIGTERM, signal.SIG_DFL)
 361                 signal.signal(signal.SIGQUIT, signal.SIG_DFL)
 362             except Exception as e:
 363                 self.logger.exception(
 364                     'arv-mount: exception during exec %s', self.args.exec_args)
 365                 try:
 366                     rc = e.errno
 367                 except AttributeError:
 368                     pass
 369         exit(rc)
 370
 371     def _run_standalone(self):
 372         try:
 373             self.daemon = not self.args.foreground
 374             with self:
 375                 self.llfuse_thread.join(timeout=None)
 376         except Exception as e:
 377             self.logger.exception('arv-mount: exception during mount: %s', e)
 378             exit(getattr(e, 'errno', 1))
 379         exit(0)
 380
 381     def _llfuse_main(self):
 382         try:
 383             llfuse.main()
 384         except:
 385             llfuse.close(unmount=False)
 386             raise
 387         llfuse.close()