1 # Copyright (C) The Arvados Authors. All rights reserved.
3 # SPDX-License-Identifier: AGPL-3.0
5 from future.utils import native_str
6 from builtins import range
7 from builtins import object
21 import arvados.commands._util as arv_cmd
22 from arvados_fuse import crunchstat
23 from arvados_fuse import *
24 from arvados_fuse.unmount import unmount
25 from arvados_fuse._version import __version__
27 class ArgumentParser(argparse.ArgumentParser):
29 super(ArgumentParser, self).__init__(
30 parents=[arv_cmd.retry_opt],
31 description='''Mount Keep data under the local filesystem. Default mode is --home''',
33 Note: When using the --exec feature, you must either specify the
34 mountpoint before --exec, or mark the end of your --exec arguments
37 self.add_argument('--version', action='version',
38 version=u"%s %s" % (sys.argv[0], __version__),
39 help='Print version and exit.')
40 self.add_argument('mountpoint', type=str, help="""Mount point.""")
41 self.add_argument('--allow-other', action='store_true',
42 help="""Let other users read the mount""")
43 self.add_argument('--subtype', type=str, metavar='STRING',
44 help="""Report mounted filesystem type as "fuse.STRING", instead of just "fuse".""")
46 mode = self.add_mutually_exclusive_group()
48 mode.add_argument('--all', action='store_const', const='all', dest='mode',
49 help="""Mount a subdirectory for each mode: home, shared, by_tag, by_id (default if no --mount-* arguments are given).""")
50 mode.add_argument('--custom', action='store_const', const=None, dest='mode',
51 help="""Mount a top level meta-directory with subdirectories as specified by additional --mount-* arguments (default if any --mount-* arguments are given).""")
52 mode.add_argument('--home', action='store_const', const='home', dest='mode',
53 help="""Mount only the user's home project.""")
54 mode.add_argument('--shared', action='store_const', const='shared', dest='mode',
55 help="""Mount only list of projects shared with the user.""")
56 mode.add_argument('--by-tag', action='store_const', const='by_tag', dest='mode',
57 help="""Mount subdirectories listed by tag.""")
58 mode.add_argument('--by-id', action='store_const', const='by_id', dest='mode',
59 help="""Mount subdirectories listed by portable data hash or uuid.""")
60 mode.add_argument('--by-pdh', action='store_const', const='by_pdh', dest='mode',
61 help="""Mount subdirectories listed by portable data hash.""")
62 mode.add_argument('--project', type=str, metavar='UUID',
63 help="""Mount the specified project.""")
64 mode.add_argument('--collection', type=str, metavar='UUID_or_PDH',
65 help="""Mount only the specified collection.""")
67 mounts = self.add_argument_group('Custom mount options')
68 mounts.add_argument('--mount-by-pdh',
69 type=str, metavar='PATH', action='append', default=[],
70 help="Mount each readable collection at mountpoint/PATH/P where P is the collection's portable data hash.")
71 mounts.add_argument('--mount-by-id',
72 type=str, metavar='PATH', action='append', default=[],
73 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.")
74 mounts.add_argument('--mount-by-tag',
75 type=str, metavar='PATH', action='append', default=[],
76 help="Mount all collections with tag TAG at mountpoint/PATH/TAG/UUID.")
77 mounts.add_argument('--mount-home',
78 type=str, metavar='PATH', action='append', default=[],
79 help="Mount the current user's home project at mountpoint/PATH.")
80 mounts.add_argument('--mount-shared',
81 type=str, metavar='PATH', action='append', default=[],
82 help="Mount projects shared with the current user at mountpoint/PATH.")
83 mounts.add_argument('--mount-tmp',
84 type=str, metavar='PATH', action='append', default=[],
85 help="Create a new collection, mount it in read/write mode at mountpoint/PATH, and delete it when unmounting.")
88 self.add_argument('--debug', action='store_true', help="""Debug mode""")
89 self.add_argument('--logfile', help="""Write debug logs and errors to the specified file (default stderr).""")
90 self.add_argument('--foreground', action='store_true', help="""Run in foreground (default is to daemonize unless --exec specified)""", default=False)
91 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")
93 self.add_argument('--file-cache', type=int, help="File data cache size, in bytes (default 8 GiB for disk-based cache or 256 MiB with RAM-only cache)", default=0)
94 self.add_argument('--directory-cache', type=int, help="Directory data cache size, in bytes (default 128 MiB)", default=128*1024*1024)
96 cachetype = self.add_mutually_exclusive_group()
97 cachetype.add_argument('--ram-cache', action='store_false', dest='disk_cache', help="Use in-memory caching only", default=True)
98 cachetype.add_argument('--disk-cache', action='store_true', dest='disk_cache', help="Use disk based caching (default)", default=True)
100 self.add_argument('--disk-cache-dir', type=str, help="Disk cache location (default ~/.cache/arvados/keep)", default=None)
102 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)
104 self.add_argument('--read-only', action='store_false', help="Mount will be read only (default)", dest="enable_write", default=False)
105 self.add_argument('--read-write', action='store_true', help="Mount will be read-write", dest="enable_write", default=False)
106 self.add_argument('--storage-classes', type=str, metavar='CLASSES', help="Specify comma separated list of storage classes to be used when saving data of new collections", default=None)
108 self.add_argument('--crunchstat-interval', type=float, help="Write stats to stderr every N seconds (default disabled)", default=0)
110 unmount = self.add_mutually_exclusive_group()
111 unmount.add_argument('--unmount', action='store_true', default=False,
112 help="Forcefully unmount the specified mountpoint (if it's a fuse mount) and exit. If --subtype is given, unmount only if the mount has the specified subtype. WARNING: This command can affect any kind of fuse mount, not just arv-mount.")
113 unmount.add_argument('--unmount-all', action='store_true', default=False,
114 help="Forcefully unmount every fuse mount at or below the specified path and exit. If --subtype is given, unmount only mounts that have the specified subtype. Exit non-zero if any other types of mounts are found at or below the given path. WARNING: This command can affect any kind of fuse mount, not just arv-mount.")
115 unmount.add_argument('--replace', action='store_true', default=False,
116 help="If a fuse mount is already present at mountpoint, forcefully unmount it before mounting")
117 self.add_argument('--unmount-timeout',
118 type=float, default=2.0,
119 help="Time to wait for graceful shutdown after --exec program exits and filesystem is unmounted")
121 self.add_argument('--exec', type=str, nargs=argparse.REMAINDER,
122 dest="exec_args", metavar=('command', 'args', '...', '--'),
123 help="""Mount, run a command, then unmount and exit""")
127 def __init__(self, args, logger=logging.getLogger('arvados.arv-mount')):
131 self.listen_for_events = False
133 self.args.mountpoint = os.path.realpath(self.args.mountpoint)
134 if self.args.logfile:
135 self.args.logfile = os.path.realpath(self.args.logfile)
138 nofile_limit = resource.getrlimit(resource.RLIMIT_NOFILE)
139 if nofile_limit[0] < 10240:
140 resource.setrlimit(resource.RLIMIT_NOFILE, (min(10240, nofile_limit[1]), nofile_limit[1]))
141 except Exception as e:
142 self.logger.warning("arv-mount: unable to adjust file handle limit: %s", e)
144 self.logger.debug("arv-mount: file handle limit is %s", resource.getrlimit(resource.RLIMIT_NOFILE))
147 self._setup_logging()
150 except Exception as e:
151 self.logger.exception("arv-mount: exception during setup: %s", e)
155 if self.args.replace:
156 unmount(path=self.args.mountpoint,
157 timeout=self.args.unmount_timeout)
158 llfuse.init(self.operations, native_str(self.args.mountpoint), self._fuse_options())
160 daemon.DaemonContext(
161 working_directory=os.path.dirname(self.args.mountpoint),
162 files_preserve=list(range(
163 3, resource.getrlimit(resource.RLIMIT_NOFILE)[1]))
165 if self.listen_for_events and not self.args.disable_event_listening:
166 self.operations.listen_for_events()
167 self.llfuse_thread = threading.Thread(None, lambda: self._llfuse_main())
168 self.llfuse_thread.daemon = True
169 self.llfuse_thread.start()
170 self.operations.initlock.wait()
173 def __exit__(self, exc_type, exc_value, traceback):
174 if self.operations.events:
175 self.operations.events.close(timeout=self.args.unmount_timeout)
176 subprocess.call(["fusermount", "-u", "-z", self.args.mountpoint])
177 self.llfuse_thread.join(timeout=self.args.unmount_timeout)
178 if self.llfuse_thread.is_alive():
179 self.logger.warning("Mount.__exit__:"
180 " llfuse thread still alive %fs after umount"
181 " -- abandoning and exiting anyway",
182 self.args.unmount_timeout)
185 if self.args.unmount or self.args.unmount_all:
186 unmount(path=self.args.mountpoint,
187 subtype=self.args.subtype,
188 timeout=self.args.unmount_timeout,
189 recursive=self.args.unmount_all)
190 elif self.args.exec_args:
193 self._run_standalone()
195 def _fuse_options(self):
196 """FUSE mount options; see mount.fuse(8)"""
197 opts = [optname for optname in ['allow_other', 'debug']
198 if getattr(self.args, optname)]
199 # Increase default read/write size from 4KiB to 128KiB
200 opts += ["big_writes", "max_read=131072"]
201 if self.args.subtype:
202 opts += ["subtype="+self.args.subtype]
205 def _setup_logging(self):
206 # Configure a log handler based on command-line switches.
207 if self.args.logfile:
208 log_handler = logging.FileHandler(self.args.logfile)
209 log_handler.setFormatter(logging.Formatter(
210 '%(asctime)s %(name)s[%(process)d] %(levelname)s: %(message)s',
211 '%Y-%m-%d %H:%M:%S'))
215 if log_handler is not None:
216 arvados.logger.removeHandler(arvados.log_handler)
217 arvados.logger.addHandler(log_handler)
220 arvados.logger.setLevel(logging.DEBUG)
221 logging.getLogger('arvados.keep').setLevel(logging.DEBUG)
222 logging.getLogger('arvados.api').setLevel(logging.DEBUG)
223 logging.getLogger('arvados.collection').setLevel(logging.DEBUG)
224 self.logger.debug("arv-mount debugging enabled")
226 self.logger.info("%s %s started", sys.argv[0], __version__)
227 self.logger.info("enable write is %s", self.args.enable_write)
229 def _setup_api(self):
231 self.api = arvados.safeapi.ThreadSafeApiCache(
232 apiconfig=arvados.config.settings(),
233 # default value of file_cache is 0, this tells KeepBlockCache to
234 # choose a default based on whether disk_cache is enabled or not.
236 'block_cache': arvados.keep.KeepBlockCache(cache_max=self.args.file_cache,
237 disk_cache=self.args.disk_cache,
238 disk_cache_dir=self.args.disk_cache_dir),
239 'num_retries': self.args.retries,
243 except KeyError as e:
244 self.logger.error("Missing environment: %s", e)
246 # Do a sanity check that we have a working arvados host + token.
247 self.api.users().current().execute()
249 def _setup_mount(self):
250 self.operations = Operations(
254 encoding=self.args.encoding,
255 inode_cache=InodeCache(cap=self.args.directory_cache),
256 enable_write=self.args.enable_write)
258 if self.args.crunchstat_interval:
259 statsthread = threading.Thread(
260 target=crunchstat.statlogger,
261 args=(self.args.crunchstat_interval,
264 statsthread.daemon = True
267 usr = self.api.users().current().execute(num_retries=self.args.retries)
270 dir_args = [llfuse.ROOT_INODE, self.operations.inodes, self.api, self.args.retries, self.args.enable_write]
273 storage_classes = None
274 if self.args.storage_classes is not None:
275 storage_classes = self.args.storage_classes.replace(' ', '').split(',')
276 self.logger.info("Storage classes requested for new collections: {}".format(', '.join(storage_classes)))
278 if self.args.collection is not None:
279 # Set up the request handler with the collection at the root
280 # First check that the collection is readable
281 self.api.collections().get(uuid=self.args.collection).execute()
282 self.args.mode = 'collection'
283 dir_class = CollectionDirectory
284 dir_args.append(self.args.collection)
285 elif self.args.project is not None:
286 self.args.mode = 'project'
287 dir_class = ProjectDirectory
289 self.api.groups().get(uuid=self.args.project).execute(
290 num_retries=self.args.retries))
292 if (self.args.mount_by_id or
293 self.args.mount_by_pdh or
294 self.args.mount_by_tag or
295 self.args.mount_home or
296 self.args.mount_shared or
297 self.args.mount_tmp):
298 if self.args.mode is not None:
300 "Cannot combine '{}' mode with custom --mount-* options.".
301 format(self.args.mode))
302 elif self.args.mode is None:
303 # If no --mount-custom or custom mount args, --all is the default
304 self.args.mode = 'all'
306 if self.args.mode in ['by_id', 'by_pdh']:
307 # Set up the request handler with the 'magic directory' at the root
308 dir_class = MagicDirectory
309 dir_args.append(self.args.mode == 'by_pdh')
310 elif self.args.mode == 'by_tag':
311 dir_class = TagsDirectory
312 elif self.args.mode == 'shared':
313 dir_class = SharedDirectory
315 elif self.args.mode == 'home':
316 dir_class = ProjectDirectory
318 dir_args.append(True)
319 elif self.args.mode == 'all':
320 self.args.mount_by_id = ['by_id']
321 self.args.mount_by_tag = ['by_tag']
322 self.args.mount_home = ['home']
323 self.args.mount_shared = ['shared']
326 if dir_class is not None:
327 if dir_class in [TagsDirectory, CollectionDirectory]:
328 ent = dir_class(*dir_args)
330 ent = dir_class(*dir_args, storage_classes=storage_classes)
331 self.operations.inodes.add_entry(ent)
332 self.listen_for_events = ent.want_event_subscribe()
335 e = self.operations.inodes.add_entry(Directory(
336 llfuse.ROOT_INODE, self.operations.inodes, self.api.config, self.args.enable_write))
337 dir_args[0] = e.inode
339 for name in self.args.mount_by_id:
340 self._add_mount(e, name, MagicDirectory(*dir_args, pdh_only=False, storage_classes=storage_classes))
341 for name in self.args.mount_by_pdh:
342 self._add_mount(e, name, MagicDirectory(*dir_args, pdh_only=True))
343 for name in self.args.mount_by_tag:
344 self._add_mount(e, name, TagsDirectory(*dir_args))
345 for name in self.args.mount_home:
346 self._add_mount(e, name, ProjectDirectory(*dir_args, project_object=usr, poll=True, storage_classes=storage_classes))
347 for name in self.args.mount_shared:
348 self._add_mount(e, name, SharedDirectory(*dir_args, exclude=usr, poll=True, storage_classes=storage_classes))
349 for name in self.args.mount_tmp:
350 self._add_mount(e, name, TmpCollectionDirectory(*dir_args, storage_classes=storage_classes))
353 text = self._readme_text(
354 arvados.config.get('ARVADOS_API_HOST'),
356 self._add_mount(e, 'README', StringFile(e.inode, text, now))
358 def _add_mount(self, tld, name, ent):
359 if name in ['', '.', '..'] or '/' in name:
360 sys.exit("Mount point '{}' is not supported.".format(name))
361 tld._entries[name] = self.operations.inodes.add_entry(ent)
362 self.listen_for_events = (self.listen_for_events or ent.want_event_subscribe())
364 def _readme_text(self, api_host, user_email):
366 Welcome to Arvados! This directory provides file system access to
367 files and objects available on the Arvados installation located at
368 '{}' using credentials for user '{}'.
370 From here, the following directories are available:
372 by_id/ Access to Keep collections by uuid or portable data hash (see by_id/README for details).
373 by_tag/ Access to Keep collections organized by tag.
374 home/ The contents of your home project.
375 shared/ Projects shared with you.
377 '''.format(api_host, user_email)
383 sp = subprocess.Popen(self.args.exec_args, shell=False)
385 # forward signals to the process.
386 signal.signal(signal.SIGINT, lambda signum, frame: sp.send_signal(signum))
387 signal.signal(signal.SIGTERM, lambda signum, frame: sp.send_signal(signum))
388 signal.signal(signal.SIGQUIT, lambda signum, frame: sp.send_signal(signum))
390 # wait for process to complete.
393 # restore default signal handlers.
394 signal.signal(signal.SIGINT, signal.SIG_DFL)
395 signal.signal(signal.SIGTERM, signal.SIG_DFL)
396 signal.signal(signal.SIGQUIT, signal.SIG_DFL)
397 except Exception as e:
398 self.logger.exception(
399 'arv-mount: exception during exec %s', self.args.exec_args)
402 except AttributeError:
406 def _run_standalone(self):
408 self.daemon = not self.args.foreground
410 self.llfuse_thread.join(timeout=None)
411 except Exception as e:
412 self.logger.exception('arv-mount: exception during mount: %s', e)
413 exit(getattr(e, 'errno', 1))
416 def _llfuse_main(self):
420 llfuse.close(unmount=False)