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