apt_dht_Khashmir/DHT.py

   1
   2 import os, sha, random
   3
   4 from twisted.internet import defer, reactor
   5 from twisted.internet.abstract import isIPAddress
   6 from twisted.python import log
   7 from twisted.trial import unittest
   8 from zope.interface import implements
   9
  10 from apt_dht.interfaces import IDHT
  11 from khashmir import Khashmir
  12
  13 class DHTError(Exception):
  14     """Represents errors that occur in the DHT."""
  15
  16 class DHT:
  17
  18     implements(IDHT)
  19
  20     def __init__(self):
  21         self.config = None
  22         self.cache_dir = ''
  23         self.bootstrap = []
  24         self.bootstrap_node = False
  25         self.joining = None
  26         self.joined = False
  27         self.outstandingJoins = 0
  28         self.foundAddrs = []
  29         self.storing = {}
  30         self.retrieving = {}
  31         self.retrieved = {}
  32
  33     def loadConfig(self, config, section):
  34         """See L{apt_dht.interfaces.IDHT}."""
  35         self.config_parser = config
  36         self.section = section
  37         self.config = {}
  38         self.cache_dir = self.config_parser.get(section, 'cache_dir')
  39         self.bootstrap = self.config_parser.getstringlist(section, 'BOOTSTRAP')
  40         self.bootstrap_node = self.config_parser.getboolean(section, 'BOOTSTRAP_NODE')
  41         for k in self.config_parser.options(section):
  42             if k in ['K', 'HASH_LENGTH', 'CONCURRENT_REQS', 'STORE_REDUNDANCY',
  43                      'MAX_FAILURES', 'PORT']:
  44                 self.config[k] = self.config_parser.getint(section, k)
  45             elif k in ['CHECKPOINT_INTERVAL', 'MIN_PING_INTERVAL',
  46                        'BUCKET_STALENESS', 'KEINITIAL_DELAY', 'KE_DELAY', 'KE_AGE']:
  47                 self.config[k] = self.config_parser.gettime(section, k)
  48             elif k in ['SPEW']:
  49                 self.config[k] = self.config_parser.getboolean(section, k)
  50             else:
  51                 self.config[k] = self.config_parser.get(section, k)
  52
  53     def join(self):
  54         """See L{apt_dht.interfaces.IDHT}."""
  55         if self.config is None:
  56             raise DHTError, "configuration not loaded"
  57         if self.joining:
  58             raise DHTError, "a join is already in progress"
  59
  60         self.khashmir = Khashmir(self.config, self.cache_dir)
  61
  62         self.joining = defer.Deferred()
  63         for node in self.bootstrap:
  64             host, port = node.rsplit(':', 1)
  65             port = int(port)
  66             if isIPAddress(host):
  67                 self._join_gotIP(host, port)
  68             else:
  69                 reactor.resolve(host).addCallback(self._join_gotIP, port)
  70
  71         return self.joining
  72
  73     def _join_gotIP(self, ip, port):
  74         """Called after an IP address has been found for a single bootstrap node."""
  75         self.outstandingJoins += 1
  76         self.khashmir.addContact(ip, port, self._join_single, self._join_error)
  77
  78     def _join_single(self, addr):
  79         """Called when a single bootstrap node has been added."""
  80         self.outstandingJoins -= 1
  81         if addr:
  82             self.foundAddrs.append(addr)
  83         if addr or self.outstandingJoins <= 0:
  84             self.khashmir.findCloseNodes(self._join_complete, self._join_complete)
  85         log.msg('Got back from bootstrap node: %r' % (addr,))
  86
  87     def _join_error(self, failure = None):
  88         """Called when a single bootstrap node has failed."""
  89         self.outstandingJoins -= 1
  90         log.msg("bootstrap node could not be reached")
  91         if self.outstandingJoins <= 0:
  92             self.khashmir.findCloseNodes(self._join_complete, self._join_complete)
  93
  94     def _join_complete(self, result):
  95         """Called when the tables have been initialized with nodes."""
  96         if not self.joined and len(result) > 0:
  97             self.joined = True
  98         if self.joining and self.outstandingJoins <= 0:
  99             df = self.joining
 100             self.joining = None
 101             if self.joined or self.bootstrap_node:
 102                 self.joined = True
 103                 df.callback(self.foundAddrs)
 104             else:
 105                 df.errback(DHTError('could not find any nodes to bootstrap to'))
 106
 107     def getAddrs(self):
 108         return self.foundAddrs
 109
 110     def leave(self):
 111         """See L{apt_dht.interfaces.IDHT}."""
 112         if self.config is None:
 113             raise DHTError, "configuration not loaded"
 114
 115         if self.joined or self.joining:
 116             if self.joining:
 117                 self.joining.errback(DHTError('still joining when leave was called'))
 118                 self.joining = None
 119             self.joined = False
 120             self.khashmir.shutdown()
 121
 122     def getValue(self, key):
 123         """See L{apt_dht.interfaces.IDHT}."""
 124         if self.config is None:
 125             raise DHTError, "configuration not loaded"
 126         if not self.joined:
 127             raise DHTError, "have not joined a network yet"
 128
 129         d = defer.Deferred()
 130         if key not in self.retrieving:
 131             self.khashmir.valueForKey(key, self._getValue)
 132         self.retrieving.setdefault(key, []).append(d)
 133         return d
 134
 135     def _getValue(self, key, result):
 136         if result:
 137             self.retrieved.setdefault(key, []).extend(result)
 138         else:
 139             final_result = []
 140             if key in self.retrieved:
 141                 final_result = self.retrieved[key]
 142                 del self.retrieved[key]
 143             for i in range(len(self.retrieving[key])):
 144                 d = self.retrieving[key].pop(0)
 145                 d.callback(final_result)
 146             del self.retrieving[key]
 147
 148     def storeValue(self, key, value):
 149         """See L{apt_dht.interfaces.IDHT}."""
 150         if self.config is None:
 151             raise DHTError, "configuration not loaded"
 152         if not self.joined:
 153             raise DHTError, "have not joined a network yet"
 154
 155         if key in self.storing and value in self.storing[key]:
 156             raise DHTError, "already storing that key with the same value"
 157
 158         d = defer.Deferred()
 159         self.khashmir.storeValueForKey(key, value, self._storeValue)
 160         self.storing.setdefault(key, {})[value] = d
 161         return d
 162
 163     def _storeValue(self, key, value, result):
 164         if key in self.storing and value in self.storing[key]:
 165             if len(result) > 0:
 166                 self.storing[key][value].callback(result)
 167             else:
 168                 self.storing[key][value].errback(DHTError('could not store value %s in key %s' % (value, key)))
 169             del self.storing[key][value]
 170             if len(self.storing[key].keys()) == 0:
 171                 del self.storing[key]
 172
 173 class TestSimpleDHT(unittest.TestCase):
 174     """Unit tests for the DHT."""
 175
 176     timeout = 2
 177     DHT_DEFAULTS = {'PORT': 9977, 'K': 8, 'HASH_LENGTH': 160,
 178                     'CHECKPOINT_INTERVAL': 900, 'CONCURRENT_REQS': 4,
 179                     'STORE_REDUNDANCY': 3, 'MAX_FAILURES': 3,
 180                     'MIN_PING_INTERVAL': 900,'BUCKET_STALENESS': 3600,
 181                     'KEINITIAL_DELAY': 15, 'KE_DELAY': 1200,
 182                     'KE_AGE': 3600, 'SPEW': False, }
 183
 184     def setUp(self):
 185         self.a = DHT()
 186         self.b = DHT()
 187         self.a.config = self.DHT_DEFAULTS.copy()
 188         self.a.config['PORT'] = 4044
 189         self.a.bootstrap = ["127.0.0.1:4044"]
 190         self.a.bootstrap_node = True
 191         self.a.cache_dir = '/tmp'
 192         self.b.config = self.DHT_DEFAULTS.copy()
 193         self.b.config['PORT'] = 4045
 194         self.b.bootstrap = ["127.0.0.1:4044"]
 195         self.b.cache_dir = '/tmp'
 196
 197     def test_bootstrap_join(self):
 198         d = self.a.join()
 199         return d
 200
 201     def node_join(self, result):
 202         d = self.b.join()
 203         return d
 204
 205     def test_join(self):
 206         self.lastDefer = defer.Deferred()
 207         d = self.a.join()
 208         d.addCallback(self.node_join)
 209         d.addCallback(self.lastDefer.callback)
 210         return self.lastDefer
 211
 212     def value_stored(self, result, value):
 213         self.stored -= 1
 214         if self.stored == 0:
 215             self.get_values()
 216
 217     def store_values(self, result):
 218         self.stored = 3
 219         d = self.a.storeValue(sha.new('4045').digest(), str(4045*3))
 220         d.addCallback(self.value_stored, 4045)
 221         d = self.a.storeValue(sha.new('4044').digest(), str(4044*2))
 222         d.addCallback(self.value_stored, 4044)
 223         d = self.b.storeValue(sha.new('4045').digest(), str(4045*2))
 224         d.addCallback(self.value_stored, 4045)
 225
 226     def check_values(self, result, values):
 227         self.checked -= 1
 228         self.failUnless(len(result) == len(values))
 229         for v in result:
 230             self.failUnless(v in values)
 231         if self.checked == 0:
 232             self.lastDefer.callback(1)
 233
 234     def get_values(self):
 235         self.checked = 4
 236         d = self.a.getValue(sha.new('4044').digest())
 237         d.addCallback(self.check_values, [str(4044*2)])
 238         d = self.b.getValue(sha.new('4044').digest())
 239         d.addCallback(self.check_values, [str(4044*2)])
 240         d = self.a.getValue(sha.new('4045').digest())
 241         d.addCallback(self.check_values, [str(4045*2), str(4045*3)])
 242         d = self.b.getValue(sha.new('4045').digest())
 243         d.addCallback(self.check_values, [str(4045*2), str(4045*3)])
 244
 245     def test_store(self):
 246         from twisted.internet.base import DelayedCall
 247         DelayedCall.debug = True
 248         self.lastDefer = defer.Deferred()
 249         d = self.a.join()
 250         d.addCallback(self.node_join)
 251         d.addCallback(self.store_values)
 252         return self.lastDefer
 253
 254     def tearDown(self):
 255         self.a.leave()
 256         try:
 257             os.unlink(self.a.khashmir.store.db)
 258         except:
 259             pass
 260         self.b.leave()
 261         try:
 262             os.unlink(self.b.khashmir.store.db)
 263         except:
 264             pass
 265
 266 class TestMultiDHT(unittest.TestCase):
 267
 268     timeout = 60
 269     num = 20
 270     DHT_DEFAULTS = {'PORT': 9977, 'K': 8, 'HASH_LENGTH': 160,
 271                     'CHECKPOINT_INTERVAL': 900, 'CONCURRENT_REQS': 4,
 272                     'STORE_REDUNDANCY': 3, 'MAX_FAILURES': 3,
 273                     'MIN_PING_INTERVAL': 900,'BUCKET_STALENESS': 3600,
 274                     'KEINITIAL_DELAY': 15, 'KE_DELAY': 1200,
 275                     'KE_AGE': 3600, 'SPEW': False, }
 276
 277     def setUp(self):
 278         self.l = []
 279         self.startport = 4081
 280         for i in range(self.num):
 281             self.l.append(DHT())
 282             self.l[i].config = self.DHT_DEFAULTS.copy()
 283             self.l[i].config['PORT'] = self.startport + i
 284             self.l[i].bootstrap = ["127.0.0.1:" + str(self.startport)]
 285             self.l[i].cache_dir = '/tmp'
 286         self.l[0].bootstrap_node = True
 287
 288     def node_join(self, result, next_node):
 289         d = self.l[next_node].join()
 290         if next_node + 1 < len(self.l):
 291             d.addCallback(self.node_join, next_node + 1)
 292         else:
 293             d.addCallback(self.lastDefer.callback)
 294
 295     def test_join(self):
 296         self.timeout = 2
 297         self.lastDefer = defer.Deferred()
 298         d = self.l[0].join()
 299         d.addCallback(self.node_join, 1)
 300         return self.lastDefer
 301
 302     def store_values(self, result, i = 0, j = 0):
 303         if j > i:
 304             j -= i+1
 305             i += 1
 306         if i == len(self.l):
 307             self.get_values()
 308         else:
 309             d = self.l[j].storeValue(sha.new(str(self.startport+i)).digest(), str((self.startport+i)*(j+1)))
 310             d.addCallback(self.store_values, i, j+1)
 311
 312     def get_values(self, result = None, check = None, i = 0, j = 0):
 313         if result is not None:
 314             self.failUnless(len(result) == len(check))
 315             for v in result:
 316                 self.failUnless(v in check)
 317         if j >= len(self.l):
 318             j -= len(self.l)
 319             i += 1
 320         if i == len(self.l):
 321             self.lastDefer.callback(1)
 322         else:
 323             d = self.l[i].getValue(sha.new(str(self.startport+j)).digest())
 324             check = []
 325             for k in range(self.startport+j, (self.startport+j)*(j+1)+1, self.startport+j):
 326                 check.append(str(k))
 327             d.addCallback(self.get_values, check, i, j + random.randrange(1, min(len(self.l), 10)))
 328
 329     def store_join(self, result, next_node):
 330         d = self.l[next_node].join()
 331         if next_node + 1 < len(self.l):
 332             d.addCallback(self.store_join, next_node + 1)
 333         else:
 334             d.addCallback(self.store_values)
 335
 336     def test_store(self):
 337         from twisted.internet.base import DelayedCall
 338         DelayedCall.debug = True
 339         self.lastDefer = defer.Deferred()
 340         d = self.l[0].join()
 341         d.addCallback(self.store_join, 1)
 342         return self.lastDefer
 343
 344     def tearDown(self):
 345         for i in self.l:
 346             try:
 347                 i.leave()
 348                 os.unlink(i.khashmir.store.db)
 349             except:
 350                 pass