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