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