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