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