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