src/allmydata/scripts/debug.py

   1
   2 # do not import any allmydata modules at this level. Do that from inside
   3 # individual functions instead.
   4 import sys, struct, time, os
   5 from twisted.python import usage
   6
   7 class DumpOptions(usage.Options):
   8     """tahoe dump-share SHARE_FILENAME"""
   9
  10     def parseArgs(self, filename):
  11         self['filename'] = filename
  12
  13 def dump_share(config, out=sys.stdout, err=sys.stderr):
  14     from allmydata import uri, storage
  15
  16     # check the version, to see if we have a mutable or immutable share
  17     print >>out, "share filename: %s" % config['filename']
  18
  19     f = open(config['filename'], "rb")
  20     prefix = f.read(32)
  21     f.close()
  22     if prefix == storage.MutableShareFile.MAGIC:
  23         return dump_mutable_share(config, out, err)
  24     # otherwise assume it's immutable
  25     f = storage.ShareFile(config['filename'])
  26     # use a ReadBucketProxy to parse the bucket and find the uri extension
  27     bp = storage.ReadBucketProxy(None)
  28     offsets = bp._parse_offsets(f.read_share_data(0, 0x24))
  29     seek = offsets['uri_extension']
  30     length = struct.unpack(">L", f.read_share_data(seek, 4))[0]
  31     seek += 4
  32     UEB_data = f.read_share_data(seek, length)
  33
  34     unpacked = uri.unpack_extension_readable(UEB_data)
  35     keys1 = ("size", "num_segments", "segment_size",
  36              "needed_shares", "total_shares")
  37     keys2 = ("codec_name", "codec_params", "tail_codec_params")
  38     keys3 = ("plaintext_hash", "plaintext_root_hash",
  39              "crypttext_hash", "crypttext_root_hash",
  40              "share_root_hash", "UEB_hash")
  41     display_keys = {"size": "file_size"}
  42     for k in keys1:
  43         if k in unpacked:
  44             dk = display_keys.get(k, k)
  45             print >>out, "%20s: %s" % (dk, unpacked[k])
  46     print >>out
  47     for k in keys2:
  48         if k in unpacked:
  49             dk = display_keys.get(k, k)
  50             print >>out, "%20s: %s" % (dk, unpacked[k])
  51     print >>out
  52     for k in keys3:
  53         if k in unpacked:
  54             dk = display_keys.get(k, k)
  55             print >>out, "%20s: %s" % (dk, unpacked[k])
  56
  57     leftover = set(unpacked.keys()) - set(keys1 + keys2 + keys3)
  58     if leftover:
  59         print >>out
  60         print >>out, "LEFTOVER:"
  61         for k in sorted(leftover):
  62             print >>out, "%20s: %s" % (k, unpacked[k])
  63
  64     sizes = {}
  65     sizes['data'] = bp._data_size
  66     sizes['validation'] = (offsets['uri_extension'] -
  67                            offsets['plaintext_hash_tree'])
  68     sizes['uri-extension'] = len(UEB_data)
  69     print >>out
  70     print >>out, " Size of data within the share:"
  71     for k in sorted(sizes):
  72         print >>out, "%20s: %s" % (k, sizes[k])
  73
  74     # display lease information too
  75     leases = list(f.iter_leases())
  76     if leases:
  77         for i,lease in enumerate(leases):
  78             (owner_num, renew_secret, cancel_secret, expiration_time) = lease
  79             when = format_expiration_time(expiration_time)
  80             print >>out, " Lease #%d: owner=%d, expire in %s" % (i, owner_num,
  81                                                                  when)
  82     else:
  83         print >>out, " No leases."
  84
  85     print >>out
  86     return 0
  87
  88 def format_expiration_time(expiration_time):
  89     now = time.time()
  90     remains = expiration_time - now
  91     when = "%ds" % remains
  92     if remains > 24*3600:
  93         when += " (%d days)" % (remains / (24*3600))
  94     elif remains > 3600:
  95         when += " (%d hours)" % (remains / 3600)
  96     return when
  97
  98
  99 def dump_mutable_share(config, out, err):
 100     from allmydata import storage
 101     from allmydata.util import idlib
 102     m = storage.MutableShareFile(config['filename'])
 103     f = open(config['filename'], "rb")
 104     WE, nodeid = m._read_write_enabler_and_nodeid(f)
 105     num_extra_leases = m._read_num_extra_leases(f)
 106     data_length = m._read_data_length(f)
 107     extra_lease_offset = m._read_extra_lease_offset(f)
 108     container_size = extra_lease_offset - m.DATA_OFFSET
 109     leases = list(m._enumerate_leases(f))
 110
 111     share_type = "unknown"
 112     f.seek(m.DATA_OFFSET)
 113     if f.read(1) == "\x00":
 114         # this slot contains an SMDF share
 115         share_type = "SDMF"
 116     f.close()
 117
 118     print >>out
 119     print >>out, "Mutable slot found:"
 120     print >>out, " share_type: %s" % share_type
 121     print >>out, " write_enabler: %s" % idlib.b2a(WE)
 122     print >>out, " WE for nodeid: %s" % idlib.nodeid_b2a(nodeid)
 123     print >>out, " num_extra_leases: %d" % num_extra_leases
 124     print >>out, " container_size: %d" % container_size
 125     print >>out, " data_length: %d" % data_length
 126     if leases:
 127         for (leasenum, (oid,et,rs,cs,anid)) in leases:
 128             print >>out
 129             print >>out, " Lease #%d:" % leasenum
 130             print >>out, "  ownerid: %d" % oid
 131             when = format_expiration_time(et)
 132             print >>out, "  expires in %s" % when
 133             print >>out, "  renew_secret: %s" % idlib.b2a(rs)
 134             print >>out, "  cancel_secret: %s" % idlib.b2a(cs)
 135             print >>out, "  secrets are for nodeid: %s" % idlib.nodeid_b2a(anid)
 136     else:
 137         print >>out, "No leases."
 138     print >>out
 139
 140     if share_type == "SDMF":
 141         dump_SDMF_share(m.DATA_OFFSET, data_length, config, out, err)
 142
 143     return 0
 144
 145 def dump_SDMF_share(offset, length, config, out, err):
 146     from allmydata import mutable
 147     from allmydata.util import idlib
 148
 149     f = open(config['filename'], "rb")
 150     f.seek(offset)
 151     data = f.read(min(length, 2000))
 152     f.close()
 153
 154     try:
 155         pieces = mutable.unpack_share(data)
 156     except mutable.NeedMoreDataError, e:
 157         # retry once with the larger size
 158         size = e.needed_bytes
 159         f = open(config['filename'], "rb")
 160         f.seek(offset)
 161         data = f.read(min(length, size))
 162         f.close()
 163         pieces = mutable.unpack_share(data)
 164
 165     (seqnum, root_hash, IV, k, N, segsize, datalen,
 166      pubkey, signature, share_hash_chain, block_hash_tree,
 167      share_data, enc_privkey) = pieces
 168
 169     print >>out, " SDMF contents:"
 170     print >>out, "  seqnum: %d" % seqnum
 171     print >>out, "  root_hash: %s" % idlib.b2a(root_hash)
 172     print >>out, "  IV: %s" % idlib.b2a(IV)
 173     print >>out, "  required_shares: %d" % k
 174     print >>out, "  total_shares: %d" % N
 175     print >>out, "  segsize: %d" % segsize
 176     print >>out, "  datalen: %d" % datalen
 177     share_hash_ids = ",".join(sorted([str(hid)
 178                                       for hid in share_hash_chain.keys()]))
 179     print >>out, "  share_hash_chain: %s" % share_hash_ids
 180     print >>out, "  block_hash_tree: %d nodes" % len(block_hash_tree)
 181
 182     print >>out
 183
 184
 185
 186 class DumpCapOptions(usage.Options):
 187     optParameters = [
 188         ["nodeid", "n", None, "storage server nodeid (ascii), to construct WE and secrets."],
 189         ["client-secret", "c", None, "client's base secret (ascii), to construct secrets"],
 190         ["client-dir", "d", None, "client's base directory, from which a -c secret will be read"],
 191         ]
 192     def parseArgs(self, cap):
 193         self.cap = cap
 194
 195 def dump_cap(config, out=sys.stdout, err=sys.stderr):
 196     from allmydata import uri
 197     from allmydata.util.idlib import a2b
 198     from base64 import b32decode
 199     import urlparse, urllib
 200
 201     cap = config.cap
 202     nodeid = None
 203     if config['nodeid']:
 204         nodeid = b32decode(config['nodeid'].upper())
 205     secret = None
 206     if config['client-secret']:
 207         secret = a2b(config['client-secret'])
 208     elif config['client-dir']:
 209         secretfile = os.path.join(config['client-dir'], "private", "secret")
 210         try:
 211             secret = a2b(open(secretfile, "r").read().strip())
 212         except EnvironmentError:
 213             pass
 214
 215     if cap.startswith("http"):
 216         scheme, netloc, path, params, query, fragment = urlparse.urlparse(cap)
 217         assert path.startswith("/uri/")
 218         cap = urllib.unquote(path[len("/uri/"):])
 219
 220     u = uri.from_string(cap)
 221
 222     print >>out
 223     dump_uri_instance(u, nodeid, secret, out, err)
 224
 225 def _dump_secrets(storage_index, secret, nodeid, out):
 226     from allmydata.util import hashutil
 227     from allmydata.util.idlib import b2a
 228
 229     if secret:
 230         crs = hashutil.my_renewal_secret_hash(secret)
 231         print >>out, " client renewal secret:", b2a(crs)
 232         frs = hashutil.file_renewal_secret_hash(crs, storage_index)
 233         print >>out, " file renewal secret:", b2a(frs)
 234         if nodeid:
 235             renew = hashutil.bucket_renewal_secret_hash(frs, nodeid)
 236             print >>out, " lease renewal secret:", b2a(renew)
 237         ccs = hashutil.my_cancel_secret_hash(secret)
 238         print >>out, " client cancel secret:", b2a(ccs)
 239         fcs = hashutil.file_cancel_secret_hash(ccs, storage_index)
 240         print >>out, " file cancel secret:", b2a(fcs)
 241         if nodeid:
 242             cancel = hashutil.bucket_cancel_secret_hash(fcs, nodeid)
 243             print >>out, " lease cancel secret:", b2a(cancel)
 244
 245 def dump_uri_instance(u, nodeid, secret, out, err, show_header=True):
 246     from allmydata import uri
 247     from allmydata.util.idlib import b2a
 248     from allmydata.util import hashutil
 249
 250     if isinstance(u, uri.CHKFileURI):
 251         if show_header:
 252             print >>out, "CHK File:"
 253         print >>out, " key:", b2a(u.key)
 254         print >>out, " UEB hash:", b2a(u.uri_extension_hash)
 255         print >>out, " size:", u.size
 256         print >>out, " k/N: %d/%d" % (u.needed_shares, u.total_shares)
 257         print >>out, " storage index:", b2a(u.storage_index)
 258         _dump_secrets(u.storage_index, secret, nodeid, out)
 259     elif isinstance(u, uri.CHKFileVerifierURI):
 260         if show_header:
 261             print >>out, "CHK Verifier URI:"
 262         print >>out, " UEB hash:", b2a(u.uri_extension_hash)
 263         print >>out, " size:", u.size
 264         print >>out, " k/N: %d/%d" % (u.needed_shares, u.total_shares)
 265         print >>out, " storage index:", b2a(u.storage_index)
 266
 267     elif isinstance(u, uri.LiteralFileURI):
 268         if show_header:
 269             print >>out, "Literal File URI:"
 270         print >>out, " data:", u.data
 271
 272     elif isinstance(u, uri.WriteableSSKFileURI):
 273         if show_header:
 274             print >>out, "SSK Writeable URI:"
 275         print >>out, " writekey:", b2a(u.writekey)
 276         print >>out, " readkey:", b2a(u.readkey)
 277         print >>out, " storage index:", b2a(u.storage_index)
 278         print >>out, " fingerprint:", b2a(u.fingerprint)
 279         print >>out
 280         if nodeid:
 281             we = hashutil.ssk_write_enabler_hash(u.writekey, nodeid)
 282             print >>out, " write_enabler:", b2a(we)
 283             print >>out
 284         _dump_secrets(u.storage_index, secret, nodeid, out)
 285
 286     elif isinstance(u, uri.ReadonlySSKFileURI):
 287         if show_header:
 288             print >>out, "SSK Read-only URI:"
 289         print >>out, " readkey:", b2a(u.readkey)
 290         print >>out, " storage index:", b2a(u.storage_index)
 291         print >>out, " fingerprint:", b2a(u.fingerprint)
 292     elif isinstance(u, uri.SSKVerifierURI):
 293         if show_header:
 294             print >>out, "SSK Verifier URI:"
 295         print >>out, " storage index:", b2a(u.storage_index)
 296         print >>out, " fingerprint:", b2a(u.fingerprint)
 297
 298     elif isinstance(u, uri.NewDirectoryURI):
 299         if show_header:
 300             print >>out, "Directory Writeable URI:"
 301         dump_uri_instance(u._filenode_uri, nodeid, secret, out, err, False)
 302     elif isinstance(u, uri.ReadonlyNewDirectoryURI):
 303         if show_header:
 304             print >>out, "Directory Read-only URI:"
 305         dump_uri_instance(u._filenode_uri, nodeid, secret, out, err, False)
 306     elif isinstance(u, uri.NewDirectoryURIVerifier):
 307         if show_header:
 308             print >>out, "Directory Verifier URI:"
 309         dump_uri_instance(u._filenode_uri, nodeid, secret, out, err, False)
 310     else:
 311         print >>out, "unknown cap type"
 312
 313 class FindSharesOptions(usage.Options):
 314     def parseArgs(self, storage_index_s, *nodedirs):
 315         self.si_s = storage_index_s
 316         self.nodedirs = nodedirs
 317
 318 def find_shares(config, out=sys.stdout, err=sys.stderr):
 319     """Given a storage index and a list of node directories, emit a list of
 320     all matching shares to stdout, one per line. For example:
 321
 322      find-shares.py 44kai1tui348689nrw8fjegc8c ~/testnet/node-*
 323
 324     gives:
 325
 326     /home/warner/testnet/node-1/storage/shares/44k/44kai1tui348689nrw8fjegc8c/5
 327     /home/warner/testnet/node-1/storage/shares/44k/44kai1tui348689nrw8fjegc8c/9
 328     /home/warner/testnet/node-2/storage/shares/44k/44kai1tui348689nrw8fjegc8c/2
 329     """
 330     from allmydata import storage
 331     from allmydata.util import idlib
 332
 333     sharedir = storage.storage_index_to_dir(idlib.a2b(config.si_s))
 334     for d in config.nodedirs:
 335         d = os.path.join(os.path.expanduser(d), "storage/shares", sharedir)
 336         if os.path.exists(d):
 337             for shnum in os.listdir(d):
 338                 print >>out, os.path.join(d, shnum)
 339
 340     return 0
 341
 342
 343 class CatalogSharesOptions(usage.Options):
 344     """
 345     Run this as 'catalog-shares NODEDIRS..', and it will emit a line to stdout
 346     for each share it finds:
 347
 348       CHK $SI $k/$N $filesize $UEB_hash $expiration $abspath_sharefile
 349       SDMF $SI $k/$N $seqnum/$roothash $expiration $abspath_sharefile
 350       UNKNOWN $abspath_sharefile
 351
 352     It may be useful to build up a catalog of shares from many storage servers
 353     and then sort the results. If you see shares with the same SI but different
 354     parameters/filesize/UEB_hash, then something is wrong.
 355
 356     """
 357     def parseArgs(self, *nodedirs):
 358         self.nodedirs = nodedirs
 359
 360 def describe_share(abs_sharefile, si_s, shnum_s, now, out, err):
 361     from allmydata import uri, storage, mutable
 362     from allmydata.util import idlib
 363     import struct
 364
 365     f = open(abs_sharefile, "rb")
 366     prefix = f.read(32)
 367
 368     if prefix == storage.MutableShareFile.MAGIC:
 369         # mutable share
 370         m = storage.MutableShareFile(abs_sharefile)
 371         WE, nodeid = m._read_write_enabler_and_nodeid(f)
 372         num_extra_leases = m._read_num_extra_leases(f)
 373         data_length = m._read_data_length(f)
 374         extra_lease_offset = m._read_extra_lease_offset(f)
 375         container_size = extra_lease_offset - m.DATA_OFFSET
 376         leases = list(m._enumerate_leases(f))
 377         expiration_time = min( [expiration_time
 378                                 for (leasenum,
 379                                      (ownerid, expiration_time, rs, cs, nodeid))
 380                                 in leases] )
 381         expiration = max(0, expiration_time - now)
 382
 383         share_type = "unknown"
 384         f.seek(m.DATA_OFFSET)
 385         if f.read(1) == "\x00":
 386             # this slot contains an SMDF share
 387             share_type = "SDMF"
 388
 389         if share_type == "SDMF":
 390             f.seek(m.DATA_OFFSET)
 391             data = f.read(min(data_length, 2000))
 392
 393             try:
 394                 pieces = mutable.unpack_share(data)
 395             except mutable.NeedMoreDataError, e:
 396                 # retry once with the larger size
 397                 size = e.needed_bytes
 398                 f.seek(m.DATA_OFFSET)
 399                 data = f.read(min(data_length, size))
 400                 pieces = mutable.unpack_share(data)
 401             (seqnum, root_hash, IV, k, N, segsize, datalen,
 402              pubkey, signature, share_hash_chain, block_hash_tree,
 403              share_data, enc_privkey) = pieces
 404
 405             print >>out, "SDMF %s %d/%d #%d:%s %d %s" % (si_s, k, N, seqnum,
 406                                                          idlib.b2a(root_hash),
 407                                                          expiration,
 408                                                          abs_sharefile)
 409         else:
 410             print >>out, "UNKNOWN mutable %s" % (abs_sharefile,)
 411
 412     elif struct.unpack(">L", prefix[:4]) == (1,):
 413         # immutable
 414
 415         sf = storage.ShareFile(abs_sharefile)
 416         # use a ReadBucketProxy to parse the bucket and find the uri extension
 417         bp = storage.ReadBucketProxy(None)
 418         offsets = bp._parse_offsets(sf.read_share_data(0, 0x24))
 419         seek = offsets['uri_extension']
 420         length = struct.unpack(">L", sf.read_share_data(seek, 4))[0]
 421         seek += 4
 422         UEB_data = sf.read_share_data(seek, length)
 423         expiration_time = min( [expiration_time
 424                                 for (ownerid, rs, cs, expiration_time)
 425                                 in sf.iter_leases()] )
 426         expiration = max(0, expiration_time - now)
 427
 428         unpacked = uri.unpack_extension_readable(UEB_data)
 429         k = unpacked["needed_shares"]
 430         N = unpacked["total_shares"]
 431         filesize = unpacked["size"]
 432         ueb_hash = unpacked["UEB_hash"]
 433
 434         print >>out, "CHK %s %d/%d %d %s %d %s" % (si_s, k, N, filesize,
 435                                                    ueb_hash, expiration,
 436                                                    abs_sharefile)
 437
 438     else:
 439         print >>out, "UNKNOWN really-unknown %s" % (abs_sharefile,)
 440
 441     f.close()
 442
 443
 444 def catalog_shares(config, out=sys.stdout, err=sys.stderr):
 445     now = time.time()
 446     for d in config.nodedirs:
 447         d = os.path.join(os.path.expanduser(d), "storage/shares")
 448         if os.path.exists(d):
 449             for abbrevdir in os.listdir(d):
 450                 abbrevdir = os.path.join(d, abbrevdir)
 451                 for si_s in os.listdir(abbrevdir):
 452                     si_dir = os.path.join(abbrevdir, si_s)
 453                     for shnum_s in os.listdir(si_dir):
 454                         abs_sharefile = os.path.join(si_dir, shnum_s)
 455                         abs_sharefile = os.path.abspath(abs_sharefile)
 456                         assert os.path.isfile(abs_sharefile)
 457                         describe_share(abs_sharefile, si_s, shnum_s, now,
 458                                        out, err)
 459     return 0
 460
 461
 462
 463 subCommands = [
 464     ["dump-share", None, DumpOptions,
 465      "Unpack and display the contents of a share (uri_extension and leases)."],
 466     ["dump-cap", None, DumpCapOptions, "Unpack a read-cap or write-cap"],
 467     ["find-shares", None, FindSharesOptions, "Locate sharefiles in node dirs"],
 468     ["catalog-shares", None, CatalogSharesOptions, "Describe shares in node dirs"],
 469     ]
 470
 471 dispatch = {
 472     "dump-share": dump_share,
 473     "dump-cap": dump_cap,
 474     "find-shares": find_shares,
 475     "catalog-shares": catalog_shares,
 476     }