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