services/fuse/arvados_fuse/command.py

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