src/allmydata/test/test_repairer.py

   1 # -*- coding: utf-8 -*-
   2 from allmydata.test import common
   3 from allmydata.monitor import Monitor
   4 from allmydata import check_results
   5 from allmydata.interfaces import NotEnoughSharesError
   6 from allmydata.immutable import upload
   7 from allmydata.util.consumer import download_to_data
   8 from twisted.internet import defer
   9 from twisted.trial import unittest
  10 import random
  11 from allmydata.test.no_network import GridTestMixin
  12
  13 # We'll allow you to pass this test even if you trigger eighteen times as
  14 # many disk reads and block fetches as would be optimal.
  15 READ_LEEWAY = 18
  16 MAX_DELTA_READS = 10 * READ_LEEWAY # N = 10
  17
  18 timeout=240 # François's ARM box timed out after 120 seconds of Verifier.test_corrupt_crypttext_hashtree
  19
  20 class RepairTestMixin:
  21     def failUnlessIsInstance(self, x, xtype):
  22         self.failUnless(isinstance(x, xtype), x)
  23
  24     def _count_reads(self):
  25         sum_of_read_counts = 0
  26         for (i, ss, storedir) in self.iterate_servers():
  27             counters = ss.stats_provider.get_stats()['counters']
  28             sum_of_read_counts += counters.get('storage_server.read', 0)
  29         return sum_of_read_counts
  30
  31     def _count_allocates(self):
  32         sum_of_allocate_counts = 0
  33         for (i, ss, storedir) in self.iterate_servers():
  34             counters = ss.stats_provider.get_stats()['counters']
  35             sum_of_allocate_counts += counters.get('storage_server.allocate', 0)
  36         return sum_of_allocate_counts
  37
  38     def _count_writes(self):
  39         sum_of_write_counts = 0
  40         for (i, ss, storedir) in self.iterate_servers():
  41             counters = ss.stats_provider.get_stats()['counters']
  42             sum_of_write_counts += counters.get('storage_server.write', 0)
  43         return sum_of_write_counts
  44
  45     def _stash_counts(self):
  46         self.before_repair_reads = self._count_reads()
  47         self.before_repair_allocates = self._count_allocates()
  48         self.before_repair_writes = self._count_writes()
  49
  50     def _get_delta_counts(self):
  51         delta_reads = self._count_reads() - self.before_repair_reads
  52         delta_allocates = self._count_allocates() - self.before_repair_allocates
  53         delta_writes = self._count_writes() - self.before_repair_writes
  54         return (delta_reads, delta_allocates, delta_writes)
  55
  56     def failIfBigger(self, x, y):
  57         self.failIf(x > y, "%s > %s" % (x, y))
  58
  59     def upload_and_stash(self):
  60         c0 = self.g.clients[0]
  61         c1 = self.g.clients[1]
  62         c0.DEFAULT_ENCODING_PARAMETERS['max_segment_size'] = 12
  63         d = c0.upload(upload.Data(common.TEST_DATA, convergence=""))
  64         def _stash_uri(ur):
  65             self.uri = ur.uri
  66             self.c0_filenode = c0.create_node_from_uri(ur.uri)
  67             self.c1_filenode = c1.create_node_from_uri(ur.uri)
  68         d.addCallback(_stash_uri)
  69         return d
  70
  71 class Verifier(GridTestMixin, unittest.TestCase, RepairTestMixin):
  72     def test_check_without_verify(self):
  73         """Check says the file is healthy when none of the shares have been
  74         touched. It says that the file is unhealthy when all of them have
  75         been removed. It doesn't use any reads.
  76         """
  77         self.basedir = "repairer/Verifier/check_without_verify"
  78         self.set_up_grid(num_clients=2)
  79         d = self.upload_and_stash()
  80         d.addCallback(lambda ignored: self._stash_counts())
  81         d.addCallback(lambda ignored:
  82                       self.c0_filenode.check(Monitor(), verify=False))
  83         def _check(cr):
  84             self.failUnless(cr.is_healthy())
  85             delta_reads, delta_allocates, delta_writes = self._get_delta_counts()
  86             self.failIfBigger(delta_reads, 0)
  87         d.addCallback(_check)
  88
  89         def _remove_all(ignored):
  90             for sh in self.find_uri_shares(self.uri):
  91                 self.delete_share(sh)
  92         d.addCallback(_remove_all)
  93
  94         d.addCallback(lambda ignored: self._stash_counts())
  95         d.addCallback(lambda ignored:
  96                       self.c0_filenode.check(Monitor(), verify=False))
  97         def _check2(cr):
  98             self.failIf(cr.is_healthy())
  99             delta_reads, delta_allocates, delta_writes = self._get_delta_counts()
 100             self.failIfBigger(delta_reads, 0)
 101         d.addCallback(_check2)
 102         return d
 103
 104     def _help_test_verify(self, corruptor, judgement, shnum=0, debug=False):
 105         self.set_up_grid(num_clients=2)
 106         d = self.upload_and_stash()
 107         d.addCallback(lambda ignored: self._stash_counts())
 108
 109         d.addCallback(lambda ignored:
 110                       self.corrupt_shares_numbered(self.uri, [shnum],corruptor,debug=debug))
 111         d.addCallback(lambda ignored:
 112                       self.c1_filenode.check(Monitor(), verify=True))
 113         def _check(vr):
 114             delta_reads, delta_allocates, delta_writes = self._get_delta_counts()
 115             self.failIfBigger(delta_reads, MAX_DELTA_READS)
 116             try:
 117                 judgement(vr)
 118             except unittest.FailTest, e:
 119                 # FailTest just uses e.args[0] == str
 120                 new_arg = str(e.args[0]) + "\nvr.data is: " + str(vr.get_data())
 121                 e.args = (new_arg,)
 122                 raise
 123         d.addCallback(_check)
 124         return d
 125
 126     def judge_no_problem(self, vr):
 127         """ Verify says the file is healthy when none of the shares have been
 128         touched in a way that matters. It doesn't use more than seven times
 129         as many reads as it needs."""
 130         self.failUnless(vr.is_healthy(), (vr, vr.is_healthy(), vr.get_data()))
 131         data = vr.get_data()
 132         self.failUnless(data['count-shares-good'] == 10, data)
 133         self.failUnless(len(data['sharemap']) == 10, data)
 134         self.failUnless(data['count-shares-needed'] == 3, data)
 135         self.failUnless(data['count-shares-expected'] == 10, data)
 136         self.failUnless(data['count-good-share-hosts'] == 10, data)
 137         self.failUnless(len(data['servers-responding']) == 10, data)
 138         self.failUnless(len(data['list-corrupt-shares']) == 0, data)
 139
 140     def test_ok_no_corruption(self):
 141         self.basedir = "repairer/Verifier/ok_no_corruption"
 142         return self._help_test_verify(common._corrupt_nothing,
 143                                       self.judge_no_problem)
 144
 145     def test_ok_filedata_size(self):
 146         self.basedir = "repairer/Verifier/ok_filedatasize"
 147         return self._help_test_verify(common._corrupt_size_of_file_data,
 148                                       self.judge_no_problem)
 149
 150     def test_ok_sharedata_size(self):
 151         self.basedir = "repairer/Verifier/ok_sharedata_size"
 152         return self._help_test_verify(common._corrupt_size_of_sharedata,
 153                                       self.judge_no_problem)
 154
 155     def test_ok_segment_size(self):
 156         self.basedir = "repairer/Verifier/test_ok_segment_size"
 157         return self._help_test_verify(common._corrupt_segment_size,
 158                                       self.judge_no_problem)
 159
 160     def judge_visible_corruption(self, vr):
 161         """Corruption which is detected by the server means that the server
 162         will send you back a Failure in response to get_bucket instead of
 163         giving you the share data. Test that verifier handles these answers
 164         correctly. It doesn't use more than seven times as many reads as it
 165         needs."""
 166         self.failIf(vr.is_healthy(), (vr, vr.is_healthy(), vr.get_data()))
 167         data = vr.get_data()
 168         self.failUnless(data['count-shares-good'] == 9, data)
 169         self.failUnless(len(data['sharemap']) == 9, data)
 170         self.failUnless(data['count-shares-needed'] == 3, data)
 171         self.failUnless(data['count-shares-expected'] == 10, data)
 172         self.failUnless(data['count-good-share-hosts'] == 9, data)
 173         self.failUnless(len(data['servers-responding']) == 10, data)
 174         self.failUnless(len(data['list-corrupt-shares']) == 0, data)
 175
 176     def test_corrupt_file_verno(self):
 177         self.basedir = "repairer/Verifier/corrupt_file_verno"
 178         return self._help_test_verify(common._corrupt_file_version_number,
 179                                       self.judge_visible_corruption)
 180
 181     def judge_share_version_incompatibility(self, vr):
 182         # corruption of the share version (inside the container, the 1/2
 183         # value that determines whether we've got 4-byte offsets or 8-byte
 184         # offsets) to something larger than 2 will trigger a
 185         # ShareVersionIncompatible exception, which should be counted in
 186         # list-incompatible-shares, rather than list-corrupt-shares.
 187         self.failIf(vr.is_healthy(), (vr, vr.is_healthy(), vr.get_data()))
 188         data = vr.get_data()
 189         self.failUnlessEqual(data['count-shares-good'], 9)
 190         self.failUnlessEqual(len(data['sharemap']), 9)
 191         self.failUnlessEqual(data['count-shares-needed'], 3)
 192         self.failUnlessEqual(data['count-shares-expected'], 10)
 193         self.failUnlessEqual(data['count-good-share-hosts'], 9)
 194         self.failUnlessEqual(len(data['servers-responding']), 10)
 195         self.failUnlessEqual(len(data['list-corrupt-shares']), 0)
 196         self.failUnlessEqual(data['count-corrupt-shares'], 0)
 197         self.failUnlessEqual(len(data['list-incompatible-shares']), 1)
 198         self.failUnlessEqual(data['count-incompatible-shares'], 1)
 199
 200     def test_corrupt_share_verno(self):
 201         self.basedir = "repairer/Verifier/corrupt_share_verno"
 202         return self._help_test_verify(common._corrupt_sharedata_version_number,
 203                                       self.judge_share_version_incompatibility)
 204
 205     def judge_invisible_corruption(self, vr):
 206         # corruption of fields that the server does not check (which is most
 207         # of them), which will be detected by the client as it downloads
 208         # those shares.
 209         self.failIf(vr.is_healthy(), (vr, vr.is_healthy(), vr.get_data()))
 210         data = vr.get_data()
 211         self.failUnlessEqual(data['count-shares-good'], 9)
 212         self.failUnlessEqual(data['count-shares-needed'], 3)
 213         self.failUnlessEqual(data['count-shares-expected'], 10)
 214         self.failUnlessEqual(data['count-good-share-hosts'], 9)
 215         self.failUnlessEqual(data['count-corrupt-shares'], 1)
 216         self.failUnlessEqual(len(data['list-corrupt-shares']), 1)
 217         self.failUnlessEqual(data['count-incompatible-shares'], 0)
 218         self.failUnlessEqual(len(data['list-incompatible-shares']), 0)
 219         self.failUnlessEqual(len(data['servers-responding']), 10)
 220         self.failUnlessEqual(len(data['sharemap']), 9)
 221
 222     def test_corrupt_sharedata_offset(self):
 223         self.basedir = "repairer/Verifier/corrupt_sharedata_offset"
 224         return self._help_test_verify(common._corrupt_offset_of_sharedata,
 225                                       self.judge_invisible_corruption)
 226
 227     def test_corrupt_ueb_offset(self):
 228         self.basedir = "repairer/Verifier/corrupt_ueb_offset"
 229         return self._help_test_verify(common._corrupt_offset_of_uri_extension,
 230                                       self.judge_invisible_corruption)
 231
 232     def test_corrupt_ueb_offset_shortread(self):
 233         self.basedir = "repairer/Verifier/corrupt_ueb_offset_shortread"
 234         return self._help_test_verify(common._corrupt_offset_of_uri_extension_to_force_short_read,
 235                                       self.judge_invisible_corruption)
 236
 237     def test_corrupt_sharedata(self):
 238         self.basedir = "repairer/Verifier/corrupt_sharedata"
 239         return self._help_test_verify(common._corrupt_share_data,
 240                                       self.judge_invisible_corruption)
 241
 242     def test_corrupt_ueb_length(self):
 243         self.basedir = "repairer/Verifier/corrupt_ueb_length"
 244         return self._help_test_verify(common._corrupt_length_of_uri_extension,
 245                                       self.judge_invisible_corruption)
 246
 247     def test_corrupt_ueb(self):
 248         self.basedir = "repairer/Verifier/corrupt_ueb"
 249         return self._help_test_verify(common._corrupt_uri_extension,
 250                                       self.judge_invisible_corruption)
 251
 252     def test_truncate_crypttext_hashtree(self):
 253         # change the start of the block hashtree, to truncate the preceding
 254         # crypttext hashtree
 255         self.basedir = "repairer/Verifier/truncate_crypttext_hashtree"
 256         return self._help_test_verify(common._corrupt_offset_of_block_hashes_to_truncate_crypttext_hashes,
 257                                       self.judge_invisible_corruption)
 258
 259     def test_corrupt_block_hashtree_offset(self):
 260         self.basedir = "repairer/Verifier/corrupt_block_hashtree_offset"
 261         return self._help_test_verify(common._corrupt_offset_of_block_hashes,
 262                                       self.judge_invisible_corruption)
 263
 264     def test_wrong_share_verno(self):
 265         self.basedir = "repairer/Verifier/wrong_share_verno"
 266         return self._help_test_verify(common._corrupt_sharedata_version_number_to_plausible_version,
 267                                       self.judge_invisible_corruption)
 268
 269     def test_corrupt_share_hashtree_offset(self):
 270         self.basedir = "repairer/Verifier/corrupt_share_hashtree_offset"
 271         return self._help_test_verify(common._corrupt_offset_of_share_hashes,
 272                                       self.judge_invisible_corruption)
 273
 274     def test_corrupt_crypttext_hashtree_offset(self):
 275         self.basedir = "repairer/Verifier/corrupt_crypttext_hashtree_offset"
 276         return self._help_test_verify(common._corrupt_offset_of_ciphertext_hash_tree,
 277                                       self.judge_invisible_corruption)
 278
 279     def test_corrupt_crypttext_hashtree(self):
 280         self.basedir = "repairer/Verifier/corrupt_crypttext_hashtree"
 281         return self._help_test_verify(common._corrupt_crypttext_hash_tree,
 282                                       self.judge_invisible_corruption)
 283
 284     def test_corrupt_crypttext_hashtree_byte_x221(self):
 285         self.basedir = "repairer/Verifier/corrupt_crypttext_hashtree_byte_9_bit_7"
 286         return self._help_test_verify(common._corrupt_crypttext_hash_tree_byte_x221,
 287                                       self.judge_invisible_corruption, debug=True)
 288
 289     def test_corrupt_block_hashtree(self):
 290         self.basedir = "repairer/Verifier/corrupt_block_hashtree"
 291         return self._help_test_verify(common._corrupt_block_hashes,
 292                                       self.judge_invisible_corruption)
 293
 294     def test_corrupt_share_hashtree(self):
 295         self.basedir = "repairer/Verifier/corrupt_share_hashtree"
 296         return self._help_test_verify(common._corrupt_share_hashes,
 297                                       self.judge_invisible_corruption)
 298
 299     # TODO: the Verifier should decode to ciphertext and check it against the
 300     # crypttext-hash-tree. Check this by constructing a bogus file, in which
 301     # the crypttext-hash-tree is modified after encoding is done, but before
 302     # the UEB is finalized. The Verifier should see a valid
 303     # crypttext-hash-tree but then the ciphertext should show up as invalid.
 304     # Normally this could only be triggered by a bug in FEC decode.
 305
 306     def OFF_test_each_byte(self):
 307         # this test takes 140s to run on my laptop, and doesn't have any
 308         # actual asserts, so it's commented out. It corrupts each byte of the
 309         # share in sequence, and checks to see which ones the Verifier
 310         # catches and which it misses. Ticket #819 contains details: there
 311         # are several portions of the share that are unused, for which
 312         # corruption is not supposed to be caught.
 313         #
 314         # If the test ran quickly, we could use the share size to compute the
 315         # offsets of these unused portions and assert that everything outside
 316         # of them was detected. We could then replace the rest of
 317         # Verifier.test_* (which takes 16s to run on my laptop) with this
 318         # one.
 319         self.basedir = "repairer/Verifier/each_byte"
 320         self.set_up_grid(num_clients=2)
 321         d = self.upload_and_stash()
 322         def _grab_sh0(res):
 323             self.sh0_file = [sharefile
 324                              for (shnum, serverid, sharefile)
 325                              in self.find_uri_shares(self.uri)
 326                              if shnum == 0][0]
 327             self.sh0_orig = open(self.sh0_file, "rb").read()
 328         d.addCallback(_grab_sh0)
 329         def _fix_sh0(res):
 330             f = open(self.sh0_file, "wb")
 331             f.write(self.sh0_orig)
 332             f.close()
 333         def _corrupt(ign, which):
 334             def _corruptor(s, debug=False):
 335                 return s[:which] + chr(ord(s[which])^0x01) + s[which+1:]
 336             self.corrupt_shares_numbered(self.uri, [0], _corruptor)
 337         results = {}
 338         def _did_check(vr, i):
 339             #print "corrupt %d: healthy=%s" % (i, vr.is_healthy())
 340             results[i] = vr.is_healthy()
 341         def _start(ign):
 342             d = defer.succeed(None)
 343             for i in range(len(self.sh0_orig)):
 344                 d.addCallback(_corrupt, i)
 345                 d.addCallback(lambda ign:
 346                               self.c1_filenode.check(Monitor(), verify=True))
 347                 d.addCallback(_did_check, i)
 348                 d.addCallback(_fix_sh0)
 349             return d
 350         d.addCallback(_start)
 351         def _show_results(ign):
 352             f = open("test_each_byte_output", "w")
 353             for i in sorted(results.keys()):
 354                 print >>f, "%d: %s" % (i, results[i])
 355             f.close()
 356             print "Please look in _trial_temp/test_each_byte_output for results"
 357         d.addCallback(_show_results)
 358         return d
 359
 360 # We'll allow you to pass this test even if you trigger thirty-five times as
 361 # many block sends and disk writes as would be optimal.
 362 WRITE_LEEWAY = 35
 363 # Optimally, you could repair one of these (small) files in a single write.
 364 DELTA_WRITES_PER_SHARE = 1 * WRITE_LEEWAY
 365
 366 class Repairer(GridTestMixin, unittest.TestCase, RepairTestMixin,
 367                common.ShouldFailMixin):
 368
 369     def test_harness(self):
 370         # This test is actually to make sure our test harness works, rather
 371         # than testing anything about Tahoe code itself.
 372
 373         self.basedir = "repairer/Repairer/test_code"
 374         self.set_up_grid(num_clients=2)
 375         d = self.upload_and_stash()
 376
 377         d.addCallback(lambda ignored: self.find_uri_shares(self.uri))
 378         def _stash_shares(oldshares):
 379             self.oldshares = oldshares
 380         d.addCallback(_stash_shares)
 381         d.addCallback(lambda ignored: self.find_uri_shares(self.uri))
 382         def _compare(newshares):
 383             self.failUnlessEqual(newshares, self.oldshares)
 384         d.addCallback(_compare)
 385
 386         def _delete_8(ignored):
 387             shnum = self.oldshares[0][0]
 388             self.delete_shares_numbered(self.uri, [shnum])
 389             for sh in self.oldshares[1:8]:
 390                 self.delete_share(sh)
 391         d.addCallback(_delete_8)
 392         d.addCallback(lambda ignored: self.find_uri_shares(self.uri))
 393         d.addCallback(lambda shares: self.failUnlessEqual(len(shares), 2))
 394
 395         d.addCallback(lambda ignored:
 396                       self.shouldFail(NotEnoughSharesError, "then_download",
 397                                       None,
 398                                       download_to_data, self.c1_filenode))
 399
 400         d.addCallback(lambda ignored:
 401                       self.shouldFail(NotEnoughSharesError, "then_repair",
 402                                       None,
 403                                       self.c1_filenode.check_and_repair,
 404                                       Monitor(), verify=False))
 405
 406         # test share corruption
 407         def _test_corrupt(ignored):
 408             olddata = {}
 409             shares = self.find_uri_shares(self.uri)
 410             for (shnum, serverid, sharefile) in shares:
 411                 olddata[ (shnum, serverid) ] = open(sharefile, "rb").read()
 412             for sh in shares:
 413                 self.corrupt_share(sh, common._corrupt_uri_extension)
 414             for (shnum, serverid, sharefile) in shares:
 415                 newdata = open(sharefile, "rb").read()
 416                 self.failIfEqual(olddata[ (shnum, serverid) ], newdata)
 417         d.addCallback(_test_corrupt)
 418
 419         def _remove_all(ignored):
 420             for sh in self.find_uri_shares(self.uri):
 421                 self.delete_share(sh)
 422         d.addCallback(_remove_all)
 423         d.addCallback(lambda ignored: self.find_uri_shares(self.uri))
 424         d.addCallback(lambda shares: self.failUnlessEqual(shares, []))
 425
 426         return d
 427
 428     def test_repair_from_deletion_of_1(self):
 429         """ Repair replaces a share that got deleted. """
 430         self.basedir = "repairer/Repairer/repair_from_deletion_of_1"
 431         self.set_up_grid(num_clients=2)
 432         d = self.upload_and_stash()
 433
 434         d.addCallback(lambda ignored:
 435                       self.delete_shares_numbered(self.uri, [2]))
 436         d.addCallback(lambda ignored: self._stash_counts())
 437         d.addCallback(lambda ignored:
 438                       self.c0_filenode.check_and_repair(Monitor(),
 439                                                         verify=False))
 440         def _check_results(crr):
 441             self.failUnlessIsInstance(crr, check_results.CheckAndRepairResults)
 442             pre = crr.get_pre_repair_results()
 443             self.failUnlessIsInstance(pre, check_results.CheckResults)
 444             post = crr.get_post_repair_results()
 445             self.failUnlessIsInstance(post, check_results.CheckResults)
 446             delta_reads, delta_allocates, delta_writes = self._get_delta_counts()
 447             self.failIfBigger(delta_reads, MAX_DELTA_READS)
 448             self.failIfBigger(delta_allocates, DELTA_WRITES_PER_SHARE)
 449             self.failIf(pre.is_healthy())
 450             self.failUnless(post.is_healthy())
 451
 452             # Now we inspect the filesystem to make sure that it has 10
 453             # shares.
 454             shares = self.find_uri_shares(self.uri)
 455             self.failIf(len(shares) < 10)
 456         d.addCallback(_check_results)
 457
 458         d.addCallback(lambda ignored:
 459                       self.c0_filenode.check(Monitor(), verify=True))
 460         d.addCallback(lambda vr: self.failUnless(vr.is_healthy()))
 461
 462         # Now we delete seven of the other shares, then try to download the
 463         # file and assert that it succeeds at downloading and has the right
 464         # contents. This can't work unless it has already repaired the
 465         # previously-deleted share #2.
 466
 467         d.addCallback(lambda ignored:
 468                       self.delete_shares_numbered(self.uri, range(3, 10+1)))
 469         d.addCallback(lambda ignored: download_to_data(self.c1_filenode))
 470         d.addCallback(lambda newdata:
 471                       self.failUnlessEqual(newdata, common.TEST_DATA))
 472         return d
 473
 474     def test_repair_from_deletion_of_7(self):
 475         """ Repair replaces seven shares that got deleted. """
 476         self.basedir = "repairer/Repairer/repair_from_deletion_of_7"
 477         self.set_up_grid(num_clients=2)
 478         d = self.upload_and_stash()
 479         d.addCallback(lambda ignored:
 480                       self.delete_shares_numbered(self.uri, range(7)))
 481         d.addCallback(lambda ignored: self._stash_counts())
 482         d.addCallback(lambda ignored:
 483                       self.c0_filenode.check_and_repair(Monitor(),
 484                                                         verify=False))
 485         def _check_results(crr):
 486             self.failUnlessIsInstance(crr, check_results.CheckAndRepairResults)
 487             pre = crr.get_pre_repair_results()
 488             self.failUnlessIsInstance(pre, check_results.CheckResults)
 489             post = crr.get_post_repair_results()
 490             self.failUnlessIsInstance(post, check_results.CheckResults)
 491             delta_reads, delta_allocates, delta_writes = self._get_delta_counts()
 492
 493             self.failIfBigger(delta_reads, MAX_DELTA_READS)
 494             self.failIfBigger(delta_allocates, (DELTA_WRITES_PER_SHARE * 7))
 495             self.failIf(pre.is_healthy())
 496             self.failUnless(post.is_healthy(), post.data)
 497
 498             # Make sure we really have 10 shares.
 499             shares = self.find_uri_shares(self.uri)
 500             self.failIf(len(shares) < 10)
 501         d.addCallback(_check_results)
 502
 503         d.addCallback(lambda ignored:
 504                       self.c0_filenode.check(Monitor(), verify=True))
 505         d.addCallback(lambda vr: self.failUnless(vr.is_healthy()))
 506
 507         # Now we delete seven of the other shares, then try to download the
 508         # file and assert that it succeeds at downloading and has the right
 509         # contents. This can't work unless it has already repaired the
 510         # previously-deleted share #2.
 511
 512         d.addCallback(lambda ignored:
 513                       self.delete_shares_numbered(self.uri, range(3, 10+1)))
 514         d.addCallback(lambda ignored: download_to_data(self.c1_filenode))
 515         d.addCallback(lambda newdata:
 516                       self.failUnlessEqual(newdata, common.TEST_DATA))
 517         return d
 518
 519     def test_repairer_servers_of_happiness(self):
 520         # The repairer is supposed to generate and place as many of the
 521         # missing shares as possible without caring about how they are
 522         # distributed.
 523         self.basedir = "repairer/Repairer/repairer_servers_of_happiness"
 524         self.set_up_grid(num_clients=2, num_servers=10)
 525         d = self.upload_and_stash()
 526         # Now delete some servers. We want to leave 3 servers, which
 527         # will allow us to restore the file to a healthy state without
 528         # distributing the shares widely enough to satisfy the default
 529         # happiness setting.
 530         def _delete_some_servers(ignored):
 531             for i in xrange(7):
 532                 self.g.remove_server(self.g.servers_by_number[i].my_nodeid)
 533
 534             assert len(self.g.servers_by_number) == 3
 535
 536         d.addCallback(_delete_some_servers)
 537         # Now try to repair the file.
 538         d.addCallback(lambda ignored:
 539             self.c0_filenode.check_and_repair(Monitor(), verify=False))
 540         def _check_results(crr):
 541             self.failUnlessIsInstance(crr,
 542                                       check_results.CheckAndRepairResults)
 543             pre = crr.get_pre_repair_results()
 544             post = crr.get_post_repair_results()
 545             for p in (pre, post):
 546                 self.failUnlessIsInstance(p, check_results.CheckResults)
 547
 548             self.failIf(pre.is_healthy())
 549             self.failUnless(post.is_healthy())
 550
 551         d.addCallback(_check_results)
 552         return d
 553
 554     # why is test_repair_from_corruption_of_1 disabled? Read on:
 555     #
 556     # As recently documented in NEWS.rst for the 1.3.0 release, the current
 557     # immutable repairer suffers from several limitations:
 558     #
 559     #  * minimalistic verifier: it's just download without decryption, so we
 560     #    don't look for corruption in N-k shares, and for many fields (those
 561     #    which are the same in all shares) we only look for corruption in a
 562     #    single share
 563     #
 564     #  * some kinds of corruption cause download to fail (when it ought to
 565     #    just switch to a different share), so repair will fail on these too
 566     #
 567     #  * RIStorageServer doesn't offer a way to delete old corrupt immutable
 568     #    shares (the authority model is not at all clear), so the best the
 569     #    repairer can do is to put replacement shares on new servers,
 570     #    unfortunately leaving the corrupt shares in place
 571     #
 572     # This test is pretty strenuous: it asserts that the repairer does the
 573     # ideal thing in 8 distinct situations, with randomized corruption in
 574     # each. Because of the aforementioned limitations, it is highly unlikely
 575     # to pass any of these. We're also concerned that the download-fails case
 576     # can provoke a lost-progress bug (one was fixed, but there might be more
 577     # lurking), which will cause the test to fail despite a ".todo" marker,
 578     # and will probably cause subsequent unrelated tests to fail too (due to
 579     # "unclean reactor" problems).
 580     #
 581     # In addition, I (warner) have recently refactored the rest of this class
 582     # to use the much-faster no_network.GridTestMixin, so this tests needs to
 583     # be updated before it will be able to run again.
 584     #
 585     # So we're turning this test off until we've done one or more of the
 586     # following:
 587     #  * remove some of these limitations
 588     #  * break the test up into smaller, more functionally-oriented pieces
 589     #  * simplify the repairer enough to let us be confident that it is free
 590     #    of lost-progress bugs
 591
 592     def OFF_test_repair_from_corruption_of_1(self):
 593         d = defer.succeed(None)
 594
 595         d.addCallback(self.find_all_shares)
 596         stash = [None]
 597         def _stash_it(res):
 598             stash[0] = res
 599             return res
 600         d.addCallback(_stash_it)
 601         def _put_it_all_back(ignored):
 602             self.replace_shares(stash[0], storage_index=self.uri.get_storage_index())
 603             return ignored
 604
 605         def _repair_from_corruption(shnum, corruptor_func):
 606             before_repair_reads = self._count_reads()
 607             before_repair_allocates = self._count_writes()
 608
 609             d2 = self.filenode.check_and_repair(Monitor(), verify=True)
 610             def _after_repair(checkandrepairresults):
 611                 prerepairres = checkandrepairresults.get_pre_repair_results()
 612                 postrepairres = checkandrepairresults.get_post_repair_results()
 613                 after_repair_reads = self._count_reads()
 614                 after_repair_allocates = self._count_writes()
 615
 616                 # The "* 2" in reads is because you might read a whole share
 617                 # before figuring out that it is corrupted. It might be
 618                 # possible to make this delta reads number a little tighter.
 619                 self.failIf(after_repair_reads - before_repair_reads > (MAX_DELTA_READS * 2), (after_repair_reads, before_repair_reads))
 620                 # The "* 2" in writes is because each server has two shares,
 621                 # and it is reasonable for repairer to conclude that there
 622                 # are two shares that it should upload, if the server fails
 623                 # to serve the first share.
 624                 self.failIf(after_repair_allocates - before_repair_allocates > (DELTA_WRITES_PER_SHARE * 2), (after_repair_allocates, before_repair_allocates))
 625                 self.failIf(prerepairres.is_healthy(), (prerepairres.data, corruptor_func))
 626                 self.failUnless(postrepairres.is_healthy(), (postrepairres.data, corruptor_func))
 627
 628                 # Now we inspect the filesystem to make sure that it has 10
 629                 # shares.
 630                 shares = self.find_all_shares()
 631                 self.failIf(len(shares) < 10)
 632
 633                 # Now we assert that the verifier reports the file as healthy.
 634                 d3 = self.filenode.check(Monitor(), verify=True)
 635                 def _after_verify(verifyresults):
 636                     self.failUnless(verifyresults.is_healthy())
 637                 d3.addCallback(_after_verify)
 638
 639                 # Now we delete seven of the other shares, then try to
 640                 # download the file and assert that it succeeds at
 641                 # downloading and has the right contents. This can't work
 642                 # unless it has already repaired the previously-corrupted share.
 643                 def _then_delete_7_and_try_a_download(unused=None):
 644                     shnums = range(10)
 645                     shnums.remove(shnum)
 646                     random.shuffle(shnums)
 647                     for sharenum in shnums[:7]:
 648                         self._delete_a_share(sharenum=sharenum)
 649
 650                     return self._download_and_check_plaintext()
 651                 d3.addCallback(_then_delete_7_and_try_a_download)
 652                 return d3
 653
 654             d2.addCallback(_after_repair)
 655             return d2
 656
 657         for corruptor_func in (
 658             common._corrupt_file_version_number,
 659             common._corrupt_sharedata_version_number,
 660             common._corrupt_offset_of_sharedata,
 661             common._corrupt_offset_of_uri_extension,
 662             common._corrupt_offset_of_uri_extension_to_force_short_read,
 663             common._corrupt_share_data,
 664             common._corrupt_length_of_uri_extension,
 665             common._corrupt_uri_extension,
 666             ):
 667             # Now we corrupt a share...
 668             d.addCallback(self._corrupt_a_random_share, corruptor_func)
 669             # And repair...
 670             d.addCallback(_repair_from_corruption, corruptor_func)
 671
 672         return d
 673     #test_repair_from_corruption_of_1.todo = "Repairer doesn't properly replace corrupted shares yet."
 674
 675     def test_tiny_reads(self):
 676         # ticket #1223 points out three problems:
 677         #   repairer reads beyond end of input file
 678         #   new-downloader does not tolerate overreads
 679         #   uploader does lots of tiny reads, inefficient
 680         self.basedir = "repairer/Repairer/test_tiny_reads"
 681         self.set_up_grid()
 682         c0 = self.g.clients[0]
 683         DATA = "a"*135
 684         c0.DEFAULT_ENCODING_PARAMETERS['k'] = 22
 685         c0.DEFAULT_ENCODING_PARAMETERS['n'] = 66
 686         d = c0.upload(upload.Data(DATA, convergence=""))
 687         def _then(ur):
 688             self.uri = ur.uri
 689             self.delete_shares_numbered(self.uri, [0])
 690             self.c0_filenode = c0.create_node_from_uri(ur.uri)
 691             self._stash_counts()
 692             return self.c0_filenode.check_and_repair(Monitor())
 693         d.addCallback(_then)
 694         def _check(ign):
 695             (r,a,w) = self._get_delta_counts()
 696             # when the uploader (driven by the repairer) does full-segment
 697             # reads, this makes 44 server read calls (2*k). Before, when it
 698             # was doing input_chunk_size reads (7 bytes), it was doing over
 699             # 400.
 700             self.failIf(r > 100, "too many reads: %d>100" % r)
 701         d.addCallback(_check)
 702         return d
 703
 704
 705 # XXX extend these tests to show that the checker detects which specific
 706 # share on which specific server is broken -- this is necessary so that the
 707 # checker results can be passed to the repairer and the repairer can go ahead
 708 # and upload fixes without first doing what is effectively a check (/verify)
 709 # run
 710
 711 # XXX extend these tests to show bad behavior of various kinds from servers:
 712 # raising exception from each remove_foo() method, for example
 713
 714 # XXX test disconnect DeadReferenceError from get_buckets and get_block_whatsit
 715
 716 # XXX test corruption that truncates other hash trees than just the crypttext
 717 # hash tree
 718
 719 # XXX test the notify-someone-about-corruption feature (also implement that
 720 # feature)
 721
 722 # XXX test whether repairer (downloader) correctly downloads a file even if
 723 # to do so it has to acquire shares from a server that has already tried to
 724 # serve it a corrupted share. (I don't think the current downloader would
 725 # pass this test, depending on the kind of corruption.)