src/allmydata/node.py

   1 import datetime, os.path, re, types, ConfigParser, tempfile
   2 from base64 import b32decode, b32encode
   3
   4 from twisted.python import log as twlog
   5 from twisted.application import service
   6 from twisted.internet import defer, reactor
   7 from foolscap.api import Tub, eventually, app_versions
   8 import foolscap.logging.log
   9 from allmydata import get_package_versions, get_package_versions_string
  10 from allmydata.util import log
  11 from allmydata.util import fileutil, iputil, observer
  12 from allmydata.util.assertutil import precondition, _assert
  13 from allmydata.util.fileutil import abspath_expanduser_unicode
  14 from allmydata.util.encodingutil import get_filesystem_encoding, quote_output
  15
  16 # Add our application versions to the data that Foolscap's LogPublisher
  17 # reports.
  18 for thing, things_version in get_package_versions().iteritems():
  19     app_versions.add_version(thing, str(things_version))
  20
  21 # group 1 will be addr (dotted quad string), group 3 if any will be portnum (string)
  22 ADDR_RE=re.compile("^([1-9][0-9]*\.[1-9][0-9]*\.[1-9][0-9]*\.[1-9][0-9]*)(:([1-9][0-9]*))?$")
  23
  24
  25 def formatTimeTahoeStyle(self, when):
  26     # we want UTC timestamps that look like:
  27     #  2007-10-12 00:26:28.566Z [Client] rnp752lz: 'client running'
  28     d = datetime.datetime.utcfromtimestamp(when)
  29     if d.microsecond:
  30         return d.isoformat(" ")[:-3]+"Z"
  31     else:
  32         return d.isoformat(" ") + ".000Z"
  33
  34 PRIV_README="""
  35 This directory contains files which contain private data for the Tahoe node,
  36 such as private keys.  On Unix-like systems, the permissions on this directory
  37 are set to disallow users other than its owner from reading the contents of
  38 the files.   See the 'configuration.rst' documentation file for details."""
  39
  40 class _None: # used as a marker in get_config()
  41     pass
  42
  43 class MissingConfigEntry(Exception):
  44     """ A required config entry was not found. """
  45
  46 class OldConfigError(Exception):
  47     """ An obsolete config file was found. See
  48     docs/historical/configuration.rst. """
  49     def __str__(self):
  50         return ("Found pre-Tahoe-LAFS-v1.3 configuration file(s):\n"
  51                 "%s\n"
  52                 "See docs/historical/configuration.rst."
  53                 % "\n".join([quote_output(fname) for fname in self.args[0]]))
  54
  55 class OldConfigOptionError(Exception):
  56     pass
  57
  58
  59 class Node(service.MultiService):
  60     # this implements common functionality of both Client nodes and Introducer
  61     # nodes.
  62     NODETYPE = "unknown NODETYPE"
  63     PORTNUMFILE = None
  64     CERTFILE = "node.pem"
  65     GENERATED_FILES = []
  66
  67     def __init__(self, basedir=u"."):
  68         service.MultiService.__init__(self)
  69         self.basedir = abspath_expanduser_unicode(unicode(basedir))
  70         self._portnumfile = os.path.join(self.basedir, self.PORTNUMFILE)
  71         self._tub_ready_observerlist = observer.OneShotObserverList()
  72         fileutil.make_dirs(os.path.join(self.basedir, "private"), 0700)
  73         open(os.path.join(self.basedir, "private", "README"), "w").write(PRIV_README)
  74
  75         # creates self.config
  76         self.read_config()
  77         nickname_utf8 = self.get_config("node", "nickname", "<unspecified>")
  78         self.nickname = nickname_utf8.decode("utf-8")
  79         assert type(self.nickname) is unicode
  80
  81         self.init_tempdir()
  82         self.create_tub()
  83         self.logSource="Node"
  84
  85         self.setup_ssh()
  86         self.setup_logging()
  87         self.log("Node constructed. " + get_package_versions_string())
  88         iputil.increase_rlimits()
  89
  90     def init_tempdir(self):
  91         local_tempdir_utf8 = "tmp" # default is NODEDIR/tmp/
  92         tempdir = self.get_config("node", "tempdir", local_tempdir_utf8).decode('utf-8')
  93         tempdir = os.path.join(self.basedir, tempdir)
  94         if not os.path.exists(tempdir):
  95             fileutil.make_dirs(tempdir)
  96         tempfile.tempdir = abspath_expanduser_unicode(tempdir)
  97         # this should cause twisted.web.http (which uses
  98         # tempfile.TemporaryFile) to put large request bodies in the given
  99         # directory. Without this, the default temp dir is usually /tmp/,
 100         # which is frequently too small.
 101         test_name = tempfile.mktemp()
 102         _assert(os.path.dirname(test_name) == tempdir, test_name, tempdir)
 103
 104     def get_config(self, section, option, default=_None, boolean=False):
 105         try:
 106             if boolean:
 107                 return self.config.getboolean(section, option)
 108             return self.config.get(section, option)
 109         except (ConfigParser.NoOptionError, ConfigParser.NoSectionError):
 110             if default is _None:
 111                 fn = os.path.join(self.basedir, u"tahoe.cfg")
 112                 raise MissingConfigEntry("%s is missing the [%s]%s entry"
 113                                          % (quote_output(fn), section, option))
 114             return default
 115
 116     def set_config(self, section, option, value):
 117         if not self.config.has_section(section):
 118             self.config.add_section(section)
 119         self.config.set(section, option, value)
 120         assert self.config.get(section, option) == value
 121
 122     def read_config(self):
 123         self.error_about_old_config_files()
 124         self.config = ConfigParser.SafeConfigParser()
 125
 126         tahoe_cfg = os.path.join(self.basedir, "tahoe.cfg")
 127         try:
 128             f = open(tahoe_cfg, "rb")
 129             try:
 130                 # Skip any initial Byte Order Mark. Since this is an ordinary file, we
 131                 # don't need to handle incomplete reads, and can assume seekability.
 132                 if f.read(3) != '\xEF\xBB\xBF':
 133                     f.seek(0)
 134                 self.config.readfp(f)
 135             finally:
 136                 f.close()
 137         except EnvironmentError:
 138             if os.path.exists(tahoe_cfg):
 139                 raise
 140
 141         cfg_tubport = self.get_config("node", "tub.port", "")
 142         if not cfg_tubport:
 143             # For 'tub.port', tahoe.cfg overrides the individual file on
 144             # disk. So only read self._portnumfile if tahoe.cfg doesn't
 145             # provide a value.
 146             try:
 147                 file_tubport = fileutil.read(self._portnumfile).strip()
 148                 self.set_config("node", "tub.port", file_tubport)
 149             except EnvironmentError:
 150                 if os.path.exists(self._portnumfile):
 151                     raise
 152
 153     def error_about_old_config_files(self):
 154         """ If any old configuration files are detected, raise OldConfigError. """
 155
 156         oldfnames = set()
 157         for name in [
 158             'nickname', 'webport', 'keepalive_timeout', 'log_gatherer.furl',
 159             'disconnect_timeout', 'advertised_ip_addresses', 'introducer.furl',
 160             'helper.furl', 'key_generator.furl', 'stats_gatherer.furl',
 161             'no_storage', 'readonly_storage', 'sizelimit',
 162             'debug_discard_storage', 'run_helper']:
 163             if name not in self.GENERATED_FILES:
 164                 fullfname = os.path.join(self.basedir, name)
 165                 if os.path.exists(fullfname):
 166                     oldfnames.add(fullfname)
 167         if oldfnames:
 168             e = OldConfigError(oldfnames)
 169             twlog.msg(e)
 170             raise e
 171
 172     def create_tub(self):
 173         certfile = os.path.join(self.basedir, "private", self.CERTFILE)
 174         self.tub = Tub(certFile=certfile)
 175         self.tub.setOption("logLocalFailures", True)
 176         self.tub.setOption("logRemoteFailures", True)
 177         self.tub.setOption("expose-remote-exception-types", False)
 178
 179         # see #521 for a discussion of how to pick these timeout values.
 180         keepalive_timeout_s = self.get_config("node", "timeout.keepalive", "")
 181         if keepalive_timeout_s:
 182             self.tub.setOption("keepaliveTimeout", int(keepalive_timeout_s))
 183         disconnect_timeout_s = self.get_config("node", "timeout.disconnect", "")
 184         if disconnect_timeout_s:
 185             # N.B.: this is in seconds, so use "1800" to get 30min
 186             self.tub.setOption("disconnectTimeout", int(disconnect_timeout_s))
 187
 188         self.nodeid = b32decode(self.tub.tubID.upper()) # binary format
 189         self.write_config("my_nodeid", b32encode(self.nodeid).lower() + "\n")
 190         self.short_nodeid = b32encode(self.nodeid).lower()[:8] # ready for printing
 191
 192         tubport = self.get_config("node", "tub.port", "tcp:0")
 193         self.tub.listenOn(tubport)
 194         # we must wait until our service has started before we can find out
 195         # our IP address and thus do tub.setLocation, and we can't register
 196         # any services with the Tub until after that point
 197         self.tub.setServiceParent(self)
 198
 199     def setup_ssh(self):
 200         ssh_port = self.get_config("node", "ssh.port", "")
 201         if ssh_port:
 202             ssh_keyfile = self.get_config("node", "ssh.authorized_keys_file").decode('utf-8')
 203             from allmydata import manhole
 204             m = manhole.AuthorizedKeysManhole(ssh_port, ssh_keyfile.encode(get_filesystem_encoding()))
 205             m.setServiceParent(self)
 206             self.log("AuthorizedKeysManhole listening on %s" % ssh_port)
 207
 208     def get_app_versions(self):
 209         # TODO: merge this with allmydata.get_package_versions
 210         return dict(app_versions.versions)
 211
 212     def get_config_from_file(self, name, required=False):
 213         """Get the (string) contents of a config file, or None if the file
 214         did not exist. If required=True, raise an exception rather than
 215         returning None. Any leading or trailing whitespace will be stripped
 216         from the data."""
 217         fn = os.path.join(self.basedir, name)
 218         try:
 219             return fileutil.read(fn).strip()
 220         except EnvironmentError:
 221             if not required:
 222                 return None
 223             raise
 224
 225     def write_private_config(self, name, value):
 226         """Write the (string) contents of a private config file (which is a
 227         config file that resides within the subdirectory named 'private'), and
 228         return it.
 229         """
 230         privname = os.path.join(self.basedir, "private", name)
 231         open(privname, "w").write(value)
 232
 233     def get_or_create_private_config(self, name, default=_None):
 234         """Try to get the (string) contents of a private config file (which
 235         is a config file that resides within the subdirectory named
 236         'private'), and return it. Any leading or trailing whitespace will be
 237         stripped from the data.
 238
 239         If the file does not exist, and default is not given, report an error.
 240         If the file does not exist and a default is specified, try to create
 241         it using that default, and then return the value that was written.
 242         If 'default' is a string, use it as a default value. If not, treat it
 243         as a zero-argument callable that is expected to return a string.
 244         """
 245         privname = os.path.join(self.basedir, "private", name)
 246         try:
 247             value = fileutil.read(privname)
 248         except EnvironmentError:
 249             if os.path.exists(privname):
 250                 raise
 251             if default is _None:
 252                 raise MissingConfigEntry("The required configuration file %s is missing."
 253                                          % (quote_output(privname),))
 254             if isinstance(default, basestring):
 255                 value = default
 256             else:
 257                 value = default()
 258             fileutil.write(privname, value)
 259         return value.strip()
 260
 261     def write_config(self, name, value, mode="w"):
 262         """Write a string to a config file."""
 263         fn = os.path.join(self.basedir, name)
 264         try:
 265             open(fn, mode).write(value)
 266         except EnvironmentError, e:
 267             self.log("Unable to write config file '%s'" % fn)
 268             self.log(e)
 269
 270     def startService(self):
 271         # Note: this class can be started and stopped at most once.
 272         self.log("Node.startService")
 273         # Record the process id in the twisted log, after startService()
 274         # (__init__ is called before fork(), but startService is called
 275         # after). Note that Foolscap logs handle pid-logging by itself, no
 276         # need to send a pid to the foolscap log here.
 277         twlog.msg("My pid: %s" % os.getpid())
 278         try:
 279             os.chmod("twistd.pid", 0644)
 280         except EnvironmentError:
 281             pass
 282         # Delay until the reactor is running.
 283         eventually(self._startService)
 284
 285     def _startService(self):
 286         precondition(reactor.running)
 287         self.log("Node._startService")
 288
 289         service.MultiService.startService(self)
 290         d = defer.succeed(None)
 291         d.addCallback(lambda res: iputil.get_local_addresses_async())
 292         d.addCallback(self._setup_tub)
 293         def _ready(res):
 294             self.log("%s running" % self.NODETYPE)
 295             self._tub_ready_observerlist.fire(self)
 296             return self
 297         d.addCallback(_ready)
 298         d.addErrback(self._service_startup_failed)
 299
 300     def _service_startup_failed(self, failure):
 301         self.log('_startService() failed')
 302         log.err(failure)
 303         print "Node._startService failed, aborting"
 304         print failure
 305         #reactor.stop() # for unknown reasons, reactor.stop() isn't working.  [ ] TODO
 306         self.log('calling os.abort()')
 307         twlog.msg('calling os.abort()') # make sure it gets into twistd.log
 308         print "calling os.abort()"
 309         os.abort()
 310
 311     def stopService(self):
 312         self.log("Node.stopService")
 313         d = self._tub_ready_observerlist.when_fired()
 314         def _really_stopService(ignored):
 315             self.log("Node._really_stopService")
 316             return service.MultiService.stopService(self)
 317         d.addCallback(_really_stopService)
 318         return d
 319
 320     def shutdown(self):
 321         """Shut down the node. Returns a Deferred that fires (with None) when
 322         it finally stops kicking."""
 323         self.log("Node.shutdown")
 324         return self.stopService()
 325
 326     def setup_logging(self):
 327         # we replace the formatTime() method of the log observer that
 328         # twistd set up for us, with a method that uses our preferred
 329         # timestamp format.
 330         for o in twlog.theLogPublisher.observers:
 331             # o might be a FileLogObserver's .emit method
 332             if type(o) is type(self.setup_logging): # bound method
 333                 ob = o.im_self
 334                 if isinstance(ob, twlog.FileLogObserver):
 335                     newmeth = types.UnboundMethodType(formatTimeTahoeStyle, ob, ob.__class__)
 336                     ob.formatTime = newmeth
 337         # TODO: twisted >2.5.0 offers maxRotatedFiles=50
 338
 339         lgfurl_file = os.path.join(self.basedir, "private", "logport.furl").encode(get_filesystem_encoding())
 340         self.tub.setOption("logport-furlfile", lgfurl_file)
 341         lgfurl = self.get_config("node", "log_gatherer.furl", "")
 342         if lgfurl:
 343             # this is in addition to the contents of log-gatherer-furlfile
 344             self.tub.setOption("log-gatherer-furl", lgfurl)
 345         self.tub.setOption("log-gatherer-furlfile",
 346                            os.path.join(self.basedir, "log_gatherer.furl"))
 347         self.tub.setOption("bridge-twisted-logs", True)
 348         incident_dir = os.path.join(self.basedir, "logs", "incidents")
 349         # this doesn't quite work yet: unit tests fail
 350         foolscap.logging.log.setLogDir(incident_dir.encode(get_filesystem_encoding()))
 351
 352     def log(self, *args, **kwargs):
 353         return log.msg(*args, **kwargs)
 354
 355     def _setup_tub(self, local_addresses):
 356         # we can't get a dynamically-assigned portnum until our Tub is
 357         # running, which means after startService.
 358         l = self.tub.getListeners()[0]
 359         portnum = l.getPortnum()
 360         # record which port we're listening on, so we can grab the same one
 361         # next time
 362         fileutil.write_atomically(self._portnumfile, "%d\n" % portnum, mode="")
 363
 364         base_location = ",".join([ "%s:%d" % (addr, portnum)
 365                                    for addr in local_addresses ])
 366         location = self.get_config("node", "tub.location", base_location)
 367         self.log("Tub location set to %s" % location)
 368         self.tub.setLocation(location)
 369
 370         return self.tub
 371
 372     def when_tub_ready(self):
 373         return self._tub_ready_observerlist.when_fired()
 374
 375     def add_service(self, s):
 376         s.setServiceParent(self)
 377         return s