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