services/fuse/bin/arv-mount

   1 #!/usr/bin/env python
   2
   3 import argparse
   4 import arvados
   5 import daemon
   6 import logging
   7 import os
   8 import signal
   9 import subprocess
  10 import time
  11
  12 import arvados.commands._util as arv_cmd
  13 from arvados_fuse import *
  14 from arvados.safeapi import ThreadSafeApiCache
  15 import arvados.keep
  16
  17 logger = logging.getLogger('arvados.arv-mount')
  18
  19 class Stat(object):
  20     def __init__(self, prefix, interval,
  21                  egr_name, ing_name,
  22                  egr_func, ing_func):
  23         self.prefix = prefix
  24         self.interval = interval
  25         self.egr_name = egr_name
  26         self.ing_name = ing_name
  27         self.egress = egr_func
  28         self.ingress = ing_func
  29         self.egr_prev = self.egress()
  30         self.ing_prev = self.ingress()
  31
  32     def update(self):
  33         egr = self.egress()
  34         ing = self.ingress()
  35
  36         delta = " -- interval %.4f seconds %d %s %d %s" % (self.interval,
  37                                                            egr - self.egr_prev,
  38                                                            self.egr_name,
  39                                                            ing - self.ing_prev,
  40                                                            self.ing_name)
  41
  42         sys.stderr.write("crunchstat: %s %d %s %d %s%s\n" % (self.prefix,
  43                                                              egr,
  44                                                              self.egr_name,
  45                                                              ing,
  46                                                              self.ing_name,
  47                                                              delta))
  48
  49         self.egr_prev = egr
  50         self.ing_prev = ing
  51
  52
  53 def statlogger(interval, keep, ops):
  54     calls = Stat("keepcalls", interval, "put", "get",
  55                  keep.put_counter.get,
  56                  keep.get_counter.get)
  57     net = Stat("net:keep0", interval, "tx", "rx",
  58                keep.upload_counter.get,
  59                keep.download_counter.get)
  60     cache = Stat("keepcache", interval, "hit", "miss",
  61                keep.hits_counter.get,
  62                keep.misses_counter.get)
  63     fuseops = Stat("fuseops", interval,"write", "read",
  64                    ops.write_ops_counter.get,
  65                    ops.read_ops_counter.get)
  66     blk = Stat("blkio:0:0", interval, "write", "read",
  67                ops.write_counter.get,
  68                ops.read_counter.get)
  69
  70     while True:
  71         time.sleep(interval)
  72         calls.update()
  73         net.update()
  74         cache.update()
  75         fuseops.update()
  76         blk.update()
  77
  78
  79 if __name__ == '__main__':
  80     # Handle command line parameters
  81     parser = argparse.ArgumentParser(
  82         parents=[arv_cmd.retry_opt],
  83         description='''Mount Keep data under the local filesystem.  Default mode is --home''',
  84         epilog="""
  85 Note: When using the --exec feature, you must either specify the
  86 mountpoint before --exec, or mark the end of your --exec arguments
  87 with "--".
  88 """)
  89     parser.add_argument('mountpoint', type=str, help="""Mount point.""")
  90     parser.add_argument('--allow-other', action='store_true',
  91                         help="""Let other users read the mount""")
  92
  93     mount_mode = parser.add_mutually_exclusive_group()
  94
  95     mount_mode.add_argument('--all', action='store_true', help="""Mount a subdirectory for each mode: home, shared, by_tag, by_id (default).""")
  96     mount_mode.add_argument('--home', action='store_true', help="""Mount only the user's home project.""")
  97     mount_mode.add_argument('--shared', action='store_true', help="""Mount only list of projects shared with the user.""")
  98     mount_mode.add_argument('--by-tag', action='store_true',
  99                             help="""Mount subdirectories listed by tag.""")
 100     mount_mode.add_argument('--by-id', action='store_true',
 101                             help="""Mount subdirectories listed by portable data hash or uuid.""")
 102     mount_mode.add_argument('--by-pdh', action='store_true',
 103                             help="""Mount subdirectories listed by portable data hash.""")
 104     mount_mode.add_argument('--project', type=str, help="""Mount a specific project.""")
 105     mount_mode.add_argument('--collection', type=str, help="""Mount only the specified collection.""")
 106
 107     parser.add_argument('--debug', action='store_true', help="""Debug mode""")
 108     parser.add_argument('--logfile', help="""Write debug logs and errors to the specified file (default stderr).""")
 109     parser.add_argument('--foreground', action='store_true', help="""Run in foreground (default is to daemonize unless --exec specified)""", default=False)
 110     parser.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")
 111
 112     parser.add_argument('--file-cache', type=int, help="File data cache size, in bytes (default 256MiB)", default=256*1024*1024)
 113     parser.add_argument('--directory-cache', type=int, help="Directory data cache size, in bytes (default 128MiB)", default=128*1024*1024)
 114
 115     parser.add_argument('--read-only', action='store_false', help="Mount will be read only (default)", dest="enable_write", default=False)
 116     parser.add_argument('--read-write', action='store_true', help="Mount will be read-write", dest="enable_write", default=False)
 117
 118     parser.add_argument('--crunchstat-interval', type=float, help="Write stats to stderr every N seconds (default disabled)", default=0)
 119
 120     parser.add_argument('--exec', type=str, nargs=argparse.REMAINDER,
 121                         dest="exec_args", metavar=('command', 'args', '...', '--'),
 122                         help="""Mount, run a command, then unmount and exit""")
 123
 124     args = parser.parse_args()
 125     args.mountpoint = os.path.realpath(args.mountpoint)
 126     if args.logfile:
 127         args.logfile = os.path.realpath(args.logfile)
 128
 129     # Daemonize as early as possible, so we don't accidentally close
 130     # file descriptors we're using.
 131     if not (args.exec_args or args.foreground):
 132         os.chdir(args.mountpoint)
 133         daemon_ctx = daemon.DaemonContext(working_directory='.')
 134         daemon_ctx.open()
 135     else:
 136         daemon_ctx = None
 137
 138     # Configure a log handler based on command-line switches.
 139     if args.logfile:
 140         log_handler = logging.FileHandler(args.logfile)
 141     elif daemon_ctx:
 142         log_handler = logging.NullHandler()
 143     else:
 144         log_handler = None
 145
 146     if log_handler is not None:
 147         arvados.logger.removeHandler(arvados.log_handler)
 148         arvados.logger.addHandler(log_handler)
 149
 150     if args.debug:
 151         arvados.logger.setLevel(logging.DEBUG)
 152         logger.debug("arv-mount debugging enabled")
 153
 154     logger.info("enable write is %s", args.enable_write)
 155
 156     try:
 157         # Create the request handler
 158         operations = Operations(os.getuid(),
 159                                 os.getgid(),
 160                                 encoding=args.encoding,
 161                                 inode_cache=InodeCache(cap=args.directory_cache),
 162                                 enable_write=args.enable_write)
 163         api = ThreadSafeApiCache(apiconfig=arvados.config.settings(),
 164                                  keep_params={"block_cache": arvados.keep.KeepBlockCache(args.file_cache)})
 165
 166         if args.crunchstat_interval:
 167             statsthread = threading.Thread(target=statlogger, args=(args.crunchstat_interval, api.keep, operations))
 168             statsthread.daemon = True
 169             statsthread.start()
 170
 171         usr = api.users().current().execute(num_retries=args.retries)
 172         now = time.time()
 173         dir_class = None
 174         dir_args = [llfuse.ROOT_INODE, operations.inodes, api, args.retries]
 175         if args.by_id or args.by_pdh:
 176             # Set up the request handler with the 'magic directory' at the root
 177             dir_class = MagicDirectory
 178             dir_args.append(args.by_pdh)
 179         elif args.by_tag:
 180             dir_class = TagsDirectory
 181         elif args.shared:
 182             dir_class = SharedDirectory
 183             dir_args.append(usr)
 184         elif args.home:
 185             dir_class = ProjectDirectory
 186             dir_args.append(usr)
 187             dir_args.append(True)
 188         elif args.collection is not None:
 189             # Set up the request handler with the collection at the root
 190             dir_class = CollectionDirectory
 191             dir_args.append(args.collection)
 192         elif args.project is not None:
 193             dir_class = ProjectDirectory
 194             dir_args.append(api.groups().get(uuid=args.project).execute(
 195                     num_retries=args.retries))
 196
 197         if dir_class is not None:
 198             operations.inodes.add_entry(dir_class(*dir_args))
 199         else:
 200             e = operations.inodes.add_entry(Directory(llfuse.ROOT_INODE, operations.inodes))
 201             dir_args[0] = e.inode
 202
 203             e._entries['by_id'] = operations.inodes.add_entry(MagicDirectory(*dir_args))
 204
 205             e._entries['by_tag'] = operations.inodes.add_entry(TagsDirectory(*dir_args))
 206
 207             dir_args.append(usr)
 208             dir_args.append(True)
 209             e._entries['home'] = operations.inodes.add_entry(ProjectDirectory(*dir_args))
 210             e._entries['shared'] = operations.inodes.add_entry(SharedDirectory(*dir_args))
 211
 212             text = '''
 213 Welcome to Arvados!  This directory provides file system access to files and objects
 214 available on the Arvados installation located at '{}'
 215 using credentials for user '{}'.
 216
 217 From here, the following directories are available:
 218
 219   by_id/     Access to Keep collections by uuid or portable data hash (see by_id/README for details).
 220   by_tag/    Access to Keep collections organized by tag.
 221   home/      The contents of your home project.
 222   shared/    Projects shared with you.
 223 '''.format(arvados.config.get('ARVADOS_API_HOST'), usr['email'])
 224
 225             e._entries["README"] = operations.inodes.add_entry(StringFile(e.inode, text, now))
 226
 227
 228     except Exception:
 229         logger.exception("arv-mount: exception during API setup")
 230         exit(1)
 231
 232     # FUSE options, see mount.fuse(8)
 233     opts = [optname for optname in ['allow_other', 'debug']
 234             if getattr(args, optname)]
 235
 236     # Increase default read/write size from 4KiB to 128KiB
 237     opts += ["big_writes", "max_read=131072"]
 238
 239     if args.exec_args:
 240         # Initialize the fuse connection
 241         llfuse.init(operations, args.mountpoint, opts)
 242
 243         # Subscribe to change events from API server
 244         if not args.by_pdh:
 245             operations.listen_for_events(api)
 246
 247         t = threading.Thread(None, lambda: llfuse.main())
 248         t.start()
 249
 250         # wait until the driver is finished initializing
 251         operations.initlock.wait()
 252
 253         rc = 255
 254         try:
 255             sp = subprocess.Popen(args.exec_args, shell=False)
 256
 257             # forward signals to the process.
 258             signal.signal(signal.SIGINT, lambda signum, frame: sp.send_signal(signum))
 259             signal.signal(signal.SIGTERM, lambda signum, frame: sp.send_signal(signum))
 260             signal.signal(signal.SIGQUIT, lambda signum, frame: sp.send_signal(signum))
 261
 262             # wait for process to complete.
 263             rc = sp.wait()
 264
 265             # restore default signal handlers.
 266             signal.signal(signal.SIGINT, signal.SIG_DFL)
 267             signal.signal(signal.SIGTERM, signal.SIG_DFL)
 268             signal.signal(signal.SIGQUIT, signal.SIG_DFL)
 269         except Exception as e:
 270             logger.exception('arv-mount: exception during exec %s',
 271                              args.exec_args)
 272             try:
 273                 rc = e.errno
 274             except AttributeError:
 275                 pass
 276         finally:
 277             subprocess.call(["fusermount", "-u", "-z", args.mountpoint])
 278             operations.destroy()
 279
 280         exit(rc)
 281     else:
 282         try:
 283             llfuse.init(operations, args.mountpoint, opts)
 284
 285             # Subscribe to change events from API server
 286             operations.listen_for_events(api)
 287
 288             llfuse.main()
 289         except Exception as e:
 290             logger.exception('arv-mount: exception during mount')
 291             exit(getattr(e, 'errno', 1))
 292         finally:
 293             operations.destroy()