| 1 | """ |
|---|
| 2 | Ported to Python 3. |
|---|
| 3 | """ |
|---|
| 4 | |
|---|
| 5 | from six import ensure_text |
|---|
| 6 | |
|---|
| 7 | from urllib.parse import quote as url_quote |
|---|
| 8 | import json |
|---|
| 9 | |
|---|
| 10 | from twisted.protocols.basic import LineOnlyReceiver |
|---|
| 11 | |
|---|
| 12 | from allmydata.scripts.common import get_alias, DEFAULT_ALIAS, escape_path, \ |
|---|
| 13 | UnknownAliasError |
|---|
| 14 | from allmydata.scripts.common_http import do_http, format_http_error |
|---|
| 15 | from allmydata.util.encodingutil import quote_output, quote_path, get_io_encoding |
|---|
| 16 | |
|---|
| 17 | class Checker(object): |
|---|
| 18 | pass |
|---|
| 19 | |
|---|
| 20 | def _quote_serverid_index_share(serverid, storage_index, sharenum): |
|---|
| 21 | return "server %s, SI %s, shnum %r" % (quote_output(serverid, quotemarks=False), |
|---|
| 22 | quote_output(storage_index, quotemarks=False), |
|---|
| 23 | sharenum) |
|---|
| 24 | |
|---|
| 25 | def check_location(options, where): |
|---|
| 26 | stdout = options.stdout |
|---|
| 27 | stderr = options.stderr |
|---|
| 28 | nodeurl = options['node-url'] |
|---|
| 29 | if not nodeurl.endswith("/"): |
|---|
| 30 | nodeurl += "/" |
|---|
| 31 | try: |
|---|
| 32 | rootcap, path = get_alias(options.aliases, where, DEFAULT_ALIAS) |
|---|
| 33 | except UnknownAliasError as e: |
|---|
| 34 | e.display(stderr) |
|---|
| 35 | return 1 |
|---|
| 36 | path = str(path, "utf-8") |
|---|
| 37 | if path == '/': |
|---|
| 38 | path = '' |
|---|
| 39 | url = nodeurl + "uri/%s" % url_quote(rootcap) |
|---|
| 40 | if path: |
|---|
| 41 | url += "/" + escape_path(path) |
|---|
| 42 | # todo: should it end with a slash? |
|---|
| 43 | url += "?t=check&output=JSON" |
|---|
| 44 | if options["verify"]: |
|---|
| 45 | url += "&verify=true" |
|---|
| 46 | if options["repair"]: |
|---|
| 47 | url += "&repair=true" |
|---|
| 48 | if options["add-lease"]: |
|---|
| 49 | url += "&add-lease=true" |
|---|
| 50 | |
|---|
| 51 | resp = do_http("POST", url) |
|---|
| 52 | if resp.status != 200: |
|---|
| 53 | print(format_http_error("ERROR", resp), file=stderr) |
|---|
| 54 | return 1 |
|---|
| 55 | jdata = resp.read().decode() |
|---|
| 56 | |
|---|
| 57 | if options.get("raw"): |
|---|
| 58 | stdout.write(jdata) |
|---|
| 59 | stdout.write("\n") |
|---|
| 60 | return 0 |
|---|
| 61 | data = json.loads(jdata) |
|---|
| 62 | |
|---|
| 63 | if options["repair"]: |
|---|
| 64 | # show repair status |
|---|
| 65 | if data["pre-repair-results"]["results"]["healthy"]: |
|---|
| 66 | summary = "healthy" |
|---|
| 67 | else: |
|---|
| 68 | summary = "not healthy" |
|---|
| 69 | stdout.write("Summary: %s\n" % summary) |
|---|
| 70 | cr = data["pre-repair-results"]["results"] |
|---|
| 71 | stdout.write(" storage index: %s\n" % quote_output(data["storage-index"], quotemarks=False)) |
|---|
| 72 | stdout.write(" good-shares: %r (encoding is %r-of-%r)\n" |
|---|
| 73 | % (cr["count-shares-good"], |
|---|
| 74 | cr["count-shares-needed"], |
|---|
| 75 | cr["count-shares-expected"])) |
|---|
| 76 | stdout.write(" wrong-shares: %r\n" % cr["count-wrong-shares"]) |
|---|
| 77 | corrupt = cr["list-corrupt-shares"] |
|---|
| 78 | if corrupt: |
|---|
| 79 | stdout.write(" corrupt shares:\n") |
|---|
| 80 | for (serverid, storage_index, sharenum) in corrupt: |
|---|
| 81 | stdout.write(" %s\n" % _quote_serverid_index_share(serverid, storage_index, sharenum)) |
|---|
| 82 | if data["repair-attempted"]: |
|---|
| 83 | if data["repair-successful"]: |
|---|
| 84 | stdout.write(" repair successful\n") |
|---|
| 85 | else: |
|---|
| 86 | stdout.write(" repair failed\n") |
|---|
| 87 | else: |
|---|
| 88 | # LIT files and directories do not have a "summary" field. |
|---|
| 89 | summary = data.get("summary", "Healthy (LIT)") |
|---|
| 90 | stdout.write("Summary: %s\n" % quote_output(summary, quotemarks=False)) |
|---|
| 91 | cr = data["results"] |
|---|
| 92 | stdout.write(" storage index: %s\n" % quote_output(data["storage-index"], quotemarks=False)) |
|---|
| 93 | |
|---|
| 94 | if all([field in cr for field in ("count-shares-good", "count-shares-needed", |
|---|
| 95 | "count-shares-expected", "count-wrong-shares")]): |
|---|
| 96 | stdout.write(" good-shares: %r (encoding is %r-of-%r)\n" |
|---|
| 97 | % (cr["count-shares-good"], |
|---|
| 98 | cr["count-shares-needed"], |
|---|
| 99 | cr["count-shares-expected"])) |
|---|
| 100 | stdout.write(" wrong-shares: %r\n" % cr["count-wrong-shares"]) |
|---|
| 101 | |
|---|
| 102 | corrupt = cr.get("list-corrupt-shares", []) |
|---|
| 103 | if corrupt: |
|---|
| 104 | stdout.write(" corrupt shares:\n") |
|---|
| 105 | for (serverid, storage_index, sharenum) in corrupt: |
|---|
| 106 | stdout.write(" %s\n" % _quote_serverid_index_share(serverid, storage_index, sharenum)) |
|---|
| 107 | |
|---|
| 108 | return 0; |
|---|
| 109 | |
|---|
| 110 | def check(options): |
|---|
| 111 | if len(options.locations) == 0: |
|---|
| 112 | errno = check_location(options, str()) |
|---|
| 113 | if errno != 0: |
|---|
| 114 | return errno |
|---|
| 115 | return 0 |
|---|
| 116 | for location in options.locations: |
|---|
| 117 | errno = check_location(options, location) |
|---|
| 118 | if errno != 0: |
|---|
| 119 | return errno |
|---|
| 120 | return 0 |
|---|
| 121 | |
|---|
| 122 | class FakeTransport(object): |
|---|
| 123 | disconnecting = False |
|---|
| 124 | |
|---|
| 125 | class DeepCheckOutput(LineOnlyReceiver, object): |
|---|
| 126 | delimiter = b"\n" |
|---|
| 127 | def __init__(self, streamer, options): |
|---|
| 128 | self.streamer = streamer |
|---|
| 129 | self.transport = FakeTransport() |
|---|
| 130 | |
|---|
| 131 | self.verbose = bool(options["verbose"]) |
|---|
| 132 | self.stdout = options.stdout |
|---|
| 133 | self.stderr = options.stderr |
|---|
| 134 | self.num_objects = 0 |
|---|
| 135 | self.files_healthy = 0 |
|---|
| 136 | self.files_unhealthy = 0 |
|---|
| 137 | self.in_error = False |
|---|
| 138 | |
|---|
| 139 | def lineReceived(self, line): |
|---|
| 140 | if self.in_error: |
|---|
| 141 | print(quote_output(line, quotemarks=False), file=self.stderr) |
|---|
| 142 | return |
|---|
| 143 | if line.startswith(b"ERROR:"): |
|---|
| 144 | self.in_error = True |
|---|
| 145 | self.streamer.rc = 1 |
|---|
| 146 | print(quote_output(line, quotemarks=False), file=self.stderr) |
|---|
| 147 | return |
|---|
| 148 | |
|---|
| 149 | d = json.loads(line) |
|---|
| 150 | stdout = self.stdout |
|---|
| 151 | if d["type"] not in ("file", "directory"): |
|---|
| 152 | return |
|---|
| 153 | self.num_objects += 1 |
|---|
| 154 | # non-verbose means print a progress marker every 100 files |
|---|
| 155 | if self.num_objects % 100 == 0: |
|---|
| 156 | print("%d objects checked.." % self.num_objects, file=stdout) |
|---|
| 157 | cr = d["check-results"] |
|---|
| 158 | if cr["results"]["healthy"]: |
|---|
| 159 | self.files_healthy += 1 |
|---|
| 160 | else: |
|---|
| 161 | self.files_unhealthy += 1 |
|---|
| 162 | if self.verbose: |
|---|
| 163 | # verbose means also print one line per file |
|---|
| 164 | path = d["path"] |
|---|
| 165 | if not path: |
|---|
| 166 | path = ["<root>"] |
|---|
| 167 | |
|---|
| 168 | # LIT files and directories do not have a "summary" field. |
|---|
| 169 | summary = cr.get("summary", "Healthy (LIT)") |
|---|
| 170 | # When Python 2 is dropped the ensure_text()/ensure_str() will be unnecessary. |
|---|
| 171 | print(ensure_text("%s: %s" % (quote_path(path), quote_output(summary, quotemarks=False)), |
|---|
| 172 | encoding=get_io_encoding()), file=stdout) |
|---|
| 173 | |
|---|
| 174 | # always print out corrupt shares |
|---|
| 175 | for shareloc in cr["results"].get("list-corrupt-shares", []): |
|---|
| 176 | (serverid, storage_index, sharenum) = shareloc |
|---|
| 177 | print(" corrupt: %s" % _quote_serverid_index_share(serverid, storage_index, sharenum), file=stdout) |
|---|
| 178 | |
|---|
| 179 | def done(self): |
|---|
| 180 | if self.in_error: |
|---|
| 181 | return |
|---|
| 182 | stdout = self.stdout |
|---|
| 183 | print("done: %d objects checked, %d healthy, %d unhealthy" \ |
|---|
| 184 | % (self.num_objects, self.files_healthy, self.files_unhealthy), file=stdout) |
|---|
| 185 | |
|---|
| 186 | class DeepCheckAndRepairOutput(LineOnlyReceiver, object): |
|---|
| 187 | delimiter = b"\n" |
|---|
| 188 | def __init__(self, streamer, options): |
|---|
| 189 | self.streamer = streamer |
|---|
| 190 | self.transport = FakeTransport() |
|---|
| 191 | |
|---|
| 192 | self.verbose = bool(options["verbose"]) |
|---|
| 193 | self.stdout = options.stdout |
|---|
| 194 | self.stderr = options.stderr |
|---|
| 195 | self.num_objects = 0 |
|---|
| 196 | self.pre_repair_files_healthy = 0 |
|---|
| 197 | self.pre_repair_files_unhealthy = 0 |
|---|
| 198 | self.repairs_attempted = 0 |
|---|
| 199 | self.repairs_successful = 0 |
|---|
| 200 | self.post_repair_files_healthy = 0 |
|---|
| 201 | self.post_repair_files_unhealthy = 0 |
|---|
| 202 | self.in_error = False |
|---|
| 203 | |
|---|
| 204 | def lineReceived(self, line): |
|---|
| 205 | if self.in_error: |
|---|
| 206 | print(quote_output(line, quotemarks=False), file=self.stderr) |
|---|
| 207 | return |
|---|
| 208 | if line.startswith(b"ERROR:"): |
|---|
| 209 | self.in_error = True |
|---|
| 210 | self.streamer.rc = 1 |
|---|
| 211 | print(quote_output(line, quotemarks=False), file=self.stderr) |
|---|
| 212 | return |
|---|
| 213 | |
|---|
| 214 | d = json.loads(line) |
|---|
| 215 | stdout = self.stdout |
|---|
| 216 | if d["type"] not in ("file", "directory"): |
|---|
| 217 | return |
|---|
| 218 | self.num_objects += 1 |
|---|
| 219 | # non-verbose means print a progress marker every 100 files |
|---|
| 220 | if self.num_objects % 100 == 0: |
|---|
| 221 | print("%d objects checked.." % self.num_objects, file=stdout) |
|---|
| 222 | crr = d["check-and-repair-results"] |
|---|
| 223 | if d["storage-index"]: |
|---|
| 224 | if crr["pre-repair-results"]["results"]["healthy"]: |
|---|
| 225 | was_healthy = True |
|---|
| 226 | self.pre_repair_files_healthy += 1 |
|---|
| 227 | else: |
|---|
| 228 | was_healthy = False |
|---|
| 229 | self.pre_repair_files_unhealthy += 1 |
|---|
| 230 | if crr["post-repair-results"]["results"]["healthy"]: |
|---|
| 231 | self.post_repair_files_healthy += 1 |
|---|
| 232 | else: |
|---|
| 233 | self.post_repair_files_unhealthy += 1 |
|---|
| 234 | else: |
|---|
| 235 | # LIT file |
|---|
| 236 | was_healthy = True |
|---|
| 237 | self.pre_repair_files_healthy += 1 |
|---|
| 238 | self.post_repair_files_healthy += 1 |
|---|
| 239 | if crr["repair-attempted"]: |
|---|
| 240 | self.repairs_attempted += 1 |
|---|
| 241 | if crr["repair-successful"]: |
|---|
| 242 | self.repairs_successful += 1 |
|---|
| 243 | if self.verbose: |
|---|
| 244 | # verbose means also print one line per file |
|---|
| 245 | path = d["path"] |
|---|
| 246 | if not path: |
|---|
| 247 | path = ["<root>"] |
|---|
| 248 | # we don't seem to have a summary available, so build one |
|---|
| 249 | if was_healthy: |
|---|
| 250 | summary = "healthy" |
|---|
| 251 | else: |
|---|
| 252 | summary = "not healthy" |
|---|
| 253 | print(ensure_text("%s: %s" % (quote_path(path), summary), |
|---|
| 254 | encoding=get_io_encoding()), file=stdout) |
|---|
| 255 | |
|---|
| 256 | # always print out corrupt shares |
|---|
| 257 | prr = crr.get("pre-repair-results", {}) |
|---|
| 258 | for shareloc in prr.get("results", {}).get("list-corrupt-shares", []): |
|---|
| 259 | (serverid, storage_index, sharenum) = shareloc |
|---|
| 260 | print(" corrupt: %s" % _quote_serverid_index_share(serverid, storage_index, sharenum), file=stdout) |
|---|
| 261 | |
|---|
| 262 | # always print out repairs |
|---|
| 263 | if crr["repair-attempted"]: |
|---|
| 264 | if crr["repair-successful"]: |
|---|
| 265 | print(" repair successful", file=stdout) |
|---|
| 266 | else: |
|---|
| 267 | print(" repair failed", file=stdout) |
|---|
| 268 | |
|---|
| 269 | def done(self): |
|---|
| 270 | if self.in_error: |
|---|
| 271 | return |
|---|
| 272 | stdout = self.stdout |
|---|
| 273 | print("done: %d objects checked" % self.num_objects, file=stdout) |
|---|
| 274 | print(" pre-repair: %d healthy, %d unhealthy" \ |
|---|
| 275 | % (self.pre_repair_files_healthy, |
|---|
| 276 | self.pre_repair_files_unhealthy), file=stdout) |
|---|
| 277 | print(" %d repairs attempted, %d successful, %d failed" \ |
|---|
| 278 | % (self.repairs_attempted, |
|---|
| 279 | self.repairs_successful, |
|---|
| 280 | (self.repairs_attempted - self.repairs_successful)), file=stdout) |
|---|
| 281 | print(" post-repair: %d healthy, %d unhealthy" \ |
|---|
| 282 | % (self.post_repair_files_healthy, |
|---|
| 283 | self.post_repair_files_unhealthy), file=stdout) |
|---|
| 284 | |
|---|
| 285 | class DeepCheckStreamer(LineOnlyReceiver, object): |
|---|
| 286 | |
|---|
| 287 | def deepcheck_location(self, options, where): |
|---|
| 288 | stdout = options.stdout |
|---|
| 289 | stderr = options.stderr |
|---|
| 290 | self.rc = 0 |
|---|
| 291 | self.options = options |
|---|
| 292 | nodeurl = options['node-url'] |
|---|
| 293 | if not nodeurl.endswith("/"): |
|---|
| 294 | nodeurl += "/" |
|---|
| 295 | self.nodeurl = nodeurl |
|---|
| 296 | |
|---|
| 297 | try: |
|---|
| 298 | rootcap, path = get_alias(options.aliases, where, DEFAULT_ALIAS) |
|---|
| 299 | except UnknownAliasError as e: |
|---|
| 300 | e.display(stderr) |
|---|
| 301 | return 1 |
|---|
| 302 | path = str(path, "utf-8") |
|---|
| 303 | if path == '/': |
|---|
| 304 | path = '' |
|---|
| 305 | url = nodeurl + "uri/%s" % url_quote(rootcap) |
|---|
| 306 | if path: |
|---|
| 307 | url += "/" + escape_path(path) |
|---|
| 308 | # todo: should it end with a slash? |
|---|
| 309 | url += "?t=stream-deep-check" |
|---|
| 310 | if options["verify"]: |
|---|
| 311 | url += "&verify=true" |
|---|
| 312 | if options["repair"]: |
|---|
| 313 | url += "&repair=true" |
|---|
| 314 | output = DeepCheckAndRepairOutput(self, options) |
|---|
| 315 | else: |
|---|
| 316 | output = DeepCheckOutput(self, options) |
|---|
| 317 | if options["add-lease"]: |
|---|
| 318 | url += "&add-lease=true" |
|---|
| 319 | resp = do_http("POST", url) |
|---|
| 320 | if resp.status not in (200, 302): |
|---|
| 321 | print(format_http_error("ERROR", resp), file=stderr) |
|---|
| 322 | return 1 |
|---|
| 323 | |
|---|
| 324 | # use Twisted to split this into lines |
|---|
| 325 | while True: |
|---|
| 326 | chunk = resp.read(100) |
|---|
| 327 | if not chunk: |
|---|
| 328 | break |
|---|
| 329 | if self.options["raw"]: |
|---|
| 330 | stdout.write(chunk.decode()) |
|---|
| 331 | else: |
|---|
| 332 | output.dataReceived(chunk) |
|---|
| 333 | if not self.options["raw"]: |
|---|
| 334 | output.done() |
|---|
| 335 | return 0 |
|---|
| 336 | |
|---|
| 337 | |
|---|
| 338 | def run(self, options): |
|---|
| 339 | if len(options.locations) == 0: |
|---|
| 340 | errno = self.deepcheck_location(options, str()) |
|---|
| 341 | if errno != 0: |
|---|
| 342 | return errno |
|---|
| 343 | return 0 |
|---|
| 344 | for location in options.locations: |
|---|
| 345 | errno = self.deepcheck_location(options, location) |
|---|
| 346 | if errno != 0: |
|---|
| 347 | return errno |
|---|
| 348 | return self.rc |
|---|
| 349 | |
|---|
| 350 | def deepcheck(options): |
|---|
| 351 | return DeepCheckStreamer().run(options) |
|---|
