1 /* 2 * CDDL HEADER START 3 * 4 * The contents of this file are subject to the terms of the 5 * Common Development and Distribution License (the "License"). 6 * You may not use this file except in compliance with the License. 7 * 8 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE 9 * or http://www.opensolaris.org/os/licensing. 10 * See the License for the specific language governing permissions 11 * and limitations under the License. 12 * 13 * When distributing Covered Code, include this CDDL HEADER in each 14 * file and include the License file at usr/src/OPENSOLARIS.LICENSE. 15 * If applicable, add the following below this CDDL HEADER, with the 16 * fields enclosed by brackets "[]" replaced with your own identifying 17 * information: Portions Copyright [yyyy] [name of copyright owner] 18 * 19 * CDDL HEADER END 20 */ 21 22 /* 23 * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. 24 * Copyright (c) 2012 by Delphix. All rights reserved. 25 * Copyright (c) 2013 Steven Hartland. All rights reserved. 26 * Copyright (c) 2021, Colm Buckley <colm@tuatha.org> 27 */ 28 29 /* 30 * This file contains the functions which analyze the status of a pool. This 31 * include both the status of an active pool, as well as the status exported 32 * pools. Returns one of the ZPOOL_STATUS_* defines describing the status of 33 * the pool. This status is independent (to a certain degree) from the state of 34 * the pool. A pool's state describes only whether or not it is capable of 35 * providing the necessary fault tolerance for data. The status describes the 36 * overall status of devices. A pool that is online can still have a device 37 * that is experiencing errors. 38 * 39 * Only a subset of the possible faults can be detected using 'zpool status', 40 * and not all possible errors correspond to a FMA message ID. The explanation 41 * is left up to the caller, depending on whether it is a live pool or an 42 * import. 43 */ 44 45 #include <libzfs.h> 46 #include <libzutil.h> 47 #include <stdlib.h> 48 #include <string.h> 49 #include <unistd.h> 50 #include <sys/systeminfo.h> 51 #include "libzfs_impl.h" 52 #include "zfeature_common.h" 53 54 /* 55 * Message ID table. This must be kept in sync with the ZPOOL_STATUS_* defines 56 * in include/libzfs.h. Note that there are some status results which go past 57 * the end of this table, and hence have no associated message ID. 58 */ 59 static char *zfs_msgid_table[] = { 60 "ZFS-8000-14", /* ZPOOL_STATUS_CORRUPT_CACHE */ 61 "ZFS-8000-2Q", /* ZPOOL_STATUS_MISSING_DEV_R */ 62 "ZFS-8000-3C", /* ZPOOL_STATUS_MISSING_DEV_NR */ 63 "ZFS-8000-4J", /* ZPOOL_STATUS_CORRUPT_LABEL_R */ 64 "ZFS-8000-5E", /* ZPOOL_STATUS_CORRUPT_LABEL_NR */ 65 "ZFS-8000-6X", /* ZPOOL_STATUS_BAD_GUID_SUM */ 66 "ZFS-8000-72", /* ZPOOL_STATUS_CORRUPT_POOL */ 67 "ZFS-8000-8A", /* ZPOOL_STATUS_CORRUPT_DATA */ 68 "ZFS-8000-9P", /* ZPOOL_STATUS_FAILING_DEV */ 69 "ZFS-8000-A5", /* ZPOOL_STATUS_VERSION_NEWER */ 70 "ZFS-8000-EY", /* ZPOOL_STATUS_HOSTID_MISMATCH */ 71 "ZFS-8000-EY", /* ZPOOL_STATUS_HOSTID_ACTIVE */ 72 "ZFS-8000-EY", /* ZPOOL_STATUS_HOSTID_REQUIRED */ 73 "ZFS-8000-HC", /* ZPOOL_STATUS_IO_FAILURE_WAIT */ 74 "ZFS-8000-JQ", /* ZPOOL_STATUS_IO_FAILURE_CONTINUE */ 75 "ZFS-8000-MM", /* ZPOOL_STATUS_IO_FAILURE_MMP */ 76 "ZFS-8000-K4", /* ZPOOL_STATUS_BAD_LOG */ 77 "ZFS-8000-ER", /* ZPOOL_STATUS_ERRATA */ 78 /* 79 * The following results have no message ID. 80 * ZPOOL_STATUS_UNSUP_FEAT_READ 81 * ZPOOL_STATUS_UNSUP_FEAT_WRITE 82 * ZPOOL_STATUS_FAULTED_DEV_R 83 * ZPOOL_STATUS_FAULTED_DEV_NR 84 * ZPOOL_STATUS_VERSION_OLDER 85 * ZPOOL_STATUS_FEAT_DISABLED 86 * ZPOOL_STATUS_RESILVERING 87 * ZPOOL_STATUS_OFFLINE_DEV 88 * ZPOOL_STATUS_REMOVED_DEV 89 * ZPOOL_STATUS_REBUILDING 90 * ZPOOL_STATUS_REBUILD_SCRUB 91 * ZPOOL_STATUS_COMPATIBILITY_ERR 92 * ZPOOL_STATUS_INCOMPATIBLE_FEAT 93 * ZPOOL_STATUS_OK 94 */ 95 }; 96 97 #define NMSGID (sizeof (zfs_msgid_table) / sizeof (zfs_msgid_table[0])) 98 99 static int 100 vdev_missing(vdev_stat_t *vs, uint_t vsc) 101 { 102 (void) vsc; 103 return (vs->vs_state == VDEV_STATE_CANT_OPEN && 104 vs->vs_aux == VDEV_AUX_OPEN_FAILED); 105 } 106 107 static int 108 vdev_faulted(vdev_stat_t *vs, uint_t vsc) 109 { 110 (void) vsc; 111 return (vs->vs_state == VDEV_STATE_FAULTED); 112 } 113 114 static int 115 vdev_errors(vdev_stat_t *vs, uint_t vsc) 116 { 117 (void) vsc; 118 return (vs->vs_state == VDEV_STATE_DEGRADED || 119 vs->vs_read_errors != 0 || vs->vs_write_errors != 0 || 120 vs->vs_checksum_errors != 0); 121 } 122 123 static int 124 vdev_broken(vdev_stat_t *vs, uint_t vsc) 125 { 126 (void) vsc; 127 return (vs->vs_state == VDEV_STATE_CANT_OPEN); 128 } 129 130 static int 131 vdev_offlined(vdev_stat_t *vs, uint_t vsc) 132 { 133 (void) vsc; 134 return (vs->vs_state == VDEV_STATE_OFFLINE); 135 } 136 137 static int 138 vdev_removed(vdev_stat_t *vs, uint_t vsc) 139 { 140 (void) vsc; 141 return (vs->vs_state == VDEV_STATE_REMOVED); 142 } 143 144 static int 145 vdev_non_native_ashift(vdev_stat_t *vs, uint_t vsc) 146 { 147 if (getenv("ZPOOL_STATUS_NON_NATIVE_ASHIFT_IGNORE") != NULL) 148 return (0); 149 150 return (VDEV_STAT_VALID(vs_physical_ashift, vsc) && 151 vs->vs_configured_ashift < vs->vs_physical_ashift); 152 } 153 154 /* 155 * Detect if any leaf devices that have seen errors or could not be opened. 156 */ 157 static boolean_t 158 find_vdev_problem(nvlist_t *vdev, int (*func)(vdev_stat_t *, uint_t), 159 boolean_t ignore_replacing) 160 { 161 nvlist_t **child; 162 vdev_stat_t *vs; 163 uint_t c, vsc, children; 164 165 /* 166 * Ignore problems within a 'replacing' vdev, since we're presumably in 167 * the process of repairing any such errors, and don't want to call them 168 * out again. We'll pick up the fact that a resilver is happening 169 * later. 170 */ 171 if (ignore_replacing == B_TRUE) { 172 char *type; 173 174 verify(nvlist_lookup_string(vdev, ZPOOL_CONFIG_TYPE, 175 &type) == 0); 176 if (strcmp(type, VDEV_TYPE_REPLACING) == 0) 177 return (B_FALSE); 178 } 179 180 if (nvlist_lookup_nvlist_array(vdev, ZPOOL_CONFIG_CHILDREN, &child, 181 &children) == 0) { 182 for (c = 0; c < children; c++) 183 if (find_vdev_problem(child[c], func, ignore_replacing)) 184 return (B_TRUE); 185 } else { 186 verify(nvlist_lookup_uint64_array(vdev, ZPOOL_CONFIG_VDEV_STATS, 187 (uint64_t **)&vs, &vsc) == 0); 188 189 if (func(vs, vsc) != 0) 190 return (B_TRUE); 191 } 192 193 /* 194 * Check any L2 cache devs 195 */ 196 if (nvlist_lookup_nvlist_array(vdev, ZPOOL_CONFIG_L2CACHE, &child, 197 &children) == 0) { 198 for (c = 0; c < children; c++) 199 if (find_vdev_problem(child[c], func, ignore_replacing)) 200 return (B_TRUE); 201 } 202 203 return (B_FALSE); 204 } 205 206 /* 207 * Active pool health status. 208 * 209 * To determine the status for a pool, we make several passes over the config, 210 * picking the most egregious error we find. In order of importance, we do the 211 * following: 212 * 213 * - Check for a complete and valid configuration 214 * - Look for any faulted or missing devices in a non-replicated config 215 * - Check for any data errors 216 * - Check for any faulted or missing devices in a replicated config 217 * - Look for any devices showing errors 218 * - Check for any resilvering or rebuilding devices 219 * 220 * There can obviously be multiple errors within a single pool, so this routine 221 * only picks the most damaging of all the current errors to report. 222 */ 223 static zpool_status_t 224 check_status(nvlist_t *config, boolean_t isimport, 225 zpool_errata_t *erratap, const char *compat) 226 { 227 nvlist_t *nvroot; 228 vdev_stat_t *vs; 229 pool_scan_stat_t *ps = NULL; 230 uint_t vsc, psc; 231 uint64_t nerr; 232 uint64_t version; 233 uint64_t stateval; 234 uint64_t suspended; 235 uint64_t hostid = 0; 236 uint64_t errata = 0; 237 unsigned long system_hostid = get_system_hostid(); 238 239 verify(nvlist_lookup_uint64(config, ZPOOL_CONFIG_VERSION, 240 &version) == 0); 241 verify(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, 242 &nvroot) == 0); 243 verify(nvlist_lookup_uint64_array(nvroot, ZPOOL_CONFIG_VDEV_STATS, 244 (uint64_t **)&vs, &vsc) == 0); 245 verify(nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_STATE, 246 &stateval) == 0); 247 248 /* 249 * Currently resilvering a vdev 250 */ 251 (void) nvlist_lookup_uint64_array(nvroot, ZPOOL_CONFIG_SCAN_STATS, 252 (uint64_t **)&ps, &psc); 253 if (ps != NULL && ps->pss_func == POOL_SCAN_RESILVER && 254 ps->pss_state == DSS_SCANNING) 255 return (ZPOOL_STATUS_RESILVERING); 256 257 /* 258 * Currently rebuilding a vdev, check top-level vdevs. 259 */ 260 vdev_rebuild_stat_t *vrs = NULL; 261 nvlist_t **child; 262 uint_t c, i, children; 263 uint64_t rebuild_end_time = 0; 264 if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_CHILDREN, 265 &child, &children) == 0) { 266 for (c = 0; c < children; c++) { 267 if ((nvlist_lookup_uint64_array(child[c], 268 ZPOOL_CONFIG_REBUILD_STATS, 269 (uint64_t **)&vrs, &i) == 0) && (vrs != NULL)) { 270 uint64_t state = vrs->vrs_state; 271 272 if (state == VDEV_REBUILD_ACTIVE) { 273 return (ZPOOL_STATUS_REBUILDING); 274 } else if (state == VDEV_REBUILD_COMPLETE && 275 vrs->vrs_end_time > rebuild_end_time) { 276 rebuild_end_time = vrs->vrs_end_time; 277 } 278 } 279 } 280 281 /* 282 * If we can determine when the last scrub was run, and it 283 * was before the last rebuild completed, then recommend 284 * that the pool be scrubbed to verify all checksums. When 285 * ps is NULL we can infer the pool has never been scrubbed. 286 */ 287 if (rebuild_end_time > 0) { 288 if (ps != NULL) { 289 if ((ps->pss_state == DSS_FINISHED && 290 ps->pss_func == POOL_SCAN_SCRUB && 291 rebuild_end_time > ps->pss_end_time) || 292 ps->pss_state == DSS_NONE) 293 return (ZPOOL_STATUS_REBUILD_SCRUB); 294 } else { 295 return (ZPOOL_STATUS_REBUILD_SCRUB); 296 } 297 } 298 } 299 300 /* 301 * The multihost property is set and the pool may be active. 302 */ 303 if (vs->vs_state == VDEV_STATE_CANT_OPEN && 304 vs->vs_aux == VDEV_AUX_ACTIVE) { 305 mmp_state_t mmp_state; 306 nvlist_t *nvinfo; 307 308 nvinfo = fnvlist_lookup_nvlist(config, ZPOOL_CONFIG_LOAD_INFO); 309 mmp_state = fnvlist_lookup_uint64(nvinfo, 310 ZPOOL_CONFIG_MMP_STATE); 311 312 if (mmp_state == MMP_STATE_ACTIVE) 313 return (ZPOOL_STATUS_HOSTID_ACTIVE); 314 else if (mmp_state == MMP_STATE_NO_HOSTID) 315 return (ZPOOL_STATUS_HOSTID_REQUIRED); 316 else 317 return (ZPOOL_STATUS_HOSTID_MISMATCH); 318 } 319 320 /* 321 * Pool last accessed by another system. 322 */ 323 (void) nvlist_lookup_uint64(config, ZPOOL_CONFIG_HOSTID, &hostid); 324 if (hostid != 0 && (unsigned long)hostid != system_hostid && 325 stateval == POOL_STATE_ACTIVE) 326 return (ZPOOL_STATUS_HOSTID_MISMATCH); 327 328 /* 329 * Newer on-disk version. 330 */ 331 if (vs->vs_state == VDEV_STATE_CANT_OPEN && 332 vs->vs_aux == VDEV_AUX_VERSION_NEWER) 333 return (ZPOOL_STATUS_VERSION_NEWER); 334 335 /* 336 * Unsupported feature(s). 337 */ 338 if (vs->vs_state == VDEV_STATE_CANT_OPEN && 339 vs->vs_aux == VDEV_AUX_UNSUP_FEAT) { 340 nvlist_t *nvinfo; 341 342 verify(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_LOAD_INFO, 343 &nvinfo) == 0); 344 if (nvlist_exists(nvinfo, ZPOOL_CONFIG_CAN_RDONLY)) 345 return (ZPOOL_STATUS_UNSUP_FEAT_WRITE); 346 return (ZPOOL_STATUS_UNSUP_FEAT_READ); 347 } 348 349 /* 350 * Check that the config is complete. 351 */ 352 if (vs->vs_state == VDEV_STATE_CANT_OPEN && 353 vs->vs_aux == VDEV_AUX_BAD_GUID_SUM) 354 return (ZPOOL_STATUS_BAD_GUID_SUM); 355 356 /* 357 * Check whether the pool has suspended. 358 */ 359 if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_SUSPENDED, 360 &suspended) == 0) { 361 uint64_t reason; 362 363 if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_SUSPENDED_REASON, 364 &reason) == 0 && reason == ZIO_SUSPEND_MMP) 365 return (ZPOOL_STATUS_IO_FAILURE_MMP); 366 367 if (suspended == ZIO_FAILURE_MODE_CONTINUE) 368 return (ZPOOL_STATUS_IO_FAILURE_CONTINUE); 369 return (ZPOOL_STATUS_IO_FAILURE_WAIT); 370 } 371 372 /* 373 * Could not read a log. 374 */ 375 if (vs->vs_state == VDEV_STATE_CANT_OPEN && 376 vs->vs_aux == VDEV_AUX_BAD_LOG) { 377 return (ZPOOL_STATUS_BAD_LOG); 378 } 379 380 /* 381 * Bad devices in non-replicated config. 382 */ 383 if (vs->vs_state == VDEV_STATE_CANT_OPEN && 384 find_vdev_problem(nvroot, vdev_faulted, B_TRUE)) 385 return (ZPOOL_STATUS_FAULTED_DEV_NR); 386 387 if (vs->vs_state == VDEV_STATE_CANT_OPEN && 388 find_vdev_problem(nvroot, vdev_missing, B_TRUE)) 389 return (ZPOOL_STATUS_MISSING_DEV_NR); 390 391 if (vs->vs_state == VDEV_STATE_CANT_OPEN && 392 find_vdev_problem(nvroot, vdev_broken, B_TRUE)) 393 return (ZPOOL_STATUS_CORRUPT_LABEL_NR); 394 395 /* 396 * Corrupted pool metadata 397 */ 398 if (vs->vs_state == VDEV_STATE_CANT_OPEN && 399 vs->vs_aux == VDEV_AUX_CORRUPT_DATA) 400 return (ZPOOL_STATUS_CORRUPT_POOL); 401 402 /* 403 * Persistent data errors. 404 */ 405 if (!isimport) { 406 if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_ERRCOUNT, 407 &nerr) == 0 && nerr != 0) 408 return (ZPOOL_STATUS_CORRUPT_DATA); 409 } 410 411 /* 412 * Missing devices in a replicated config. 413 */ 414 if (find_vdev_problem(nvroot, vdev_faulted, B_TRUE)) 415 return (ZPOOL_STATUS_FAULTED_DEV_R); 416 if (find_vdev_problem(nvroot, vdev_missing, B_TRUE)) 417 return (ZPOOL_STATUS_MISSING_DEV_R); 418 if (find_vdev_problem(nvroot, vdev_broken, B_TRUE)) 419 return (ZPOOL_STATUS_CORRUPT_LABEL_R); 420 421 /* 422 * Devices with errors 423 */ 424 if (!isimport && find_vdev_problem(nvroot, vdev_errors, B_TRUE)) 425 return (ZPOOL_STATUS_FAILING_DEV); 426 427 /* 428 * Offlined devices 429 */ 430 if (find_vdev_problem(nvroot, vdev_offlined, B_TRUE)) 431 return (ZPOOL_STATUS_OFFLINE_DEV); 432 433 /* 434 * Removed device 435 */ 436 if (find_vdev_problem(nvroot, vdev_removed, B_TRUE)) 437 return (ZPOOL_STATUS_REMOVED_DEV); 438 439 /* 440 * Suboptimal, but usable, ashift configuration. 441 */ 442 if (find_vdev_problem(nvroot, vdev_non_native_ashift, B_FALSE)) 443 return (ZPOOL_STATUS_NON_NATIVE_ASHIFT); 444 445 /* 446 * Informational errata available. 447 */ 448 (void) nvlist_lookup_uint64(config, ZPOOL_CONFIG_ERRATA, &errata); 449 if (errata) { 450 *erratap = errata; 451 return (ZPOOL_STATUS_ERRATA); 452 } 453 454 /* 455 * Outdated, but usable, version 456 */ 457 if (SPA_VERSION_IS_SUPPORTED(version) && version != SPA_VERSION) { 458 /* "legacy" compatibility disables old version reporting */ 459 if (compat != NULL && strcmp(compat, ZPOOL_COMPAT_LEGACY) == 0) 460 return (ZPOOL_STATUS_OK); 461 else 462 return (ZPOOL_STATUS_VERSION_OLDER); 463 } 464 465 /* 466 * Usable pool with disabled or superfluous features 467 * (superfluous = beyond what's requested by 'compatibility') 468 */ 469 if (version >= SPA_VERSION_FEATURES) { 470 int i; 471 nvlist_t *feat; 472 473 if (isimport) { 474 feat = fnvlist_lookup_nvlist(config, 475 ZPOOL_CONFIG_LOAD_INFO); 476 if (nvlist_exists(feat, ZPOOL_CONFIG_ENABLED_FEAT)) 477 feat = fnvlist_lookup_nvlist(feat, 478 ZPOOL_CONFIG_ENABLED_FEAT); 479 } else { 480 feat = fnvlist_lookup_nvlist(config, 481 ZPOOL_CONFIG_FEATURE_STATS); 482 } 483 484 /* check against all features, or limited set? */ 485 boolean_t c_features[SPA_FEATURES]; 486 487 switch (zpool_load_compat(compat, c_features, NULL, 0)) { 488 case ZPOOL_COMPATIBILITY_OK: 489 case ZPOOL_COMPATIBILITY_WARNTOKEN: 490 break; 491 default: 492 return (ZPOOL_STATUS_COMPATIBILITY_ERR); 493 } 494 for (i = 0; i < SPA_FEATURES; i++) { 495 zfeature_info_t *fi = &spa_feature_table[i]; 496 if (!fi->fi_zfs_mod_supported) 497 continue; 498 if (c_features[i] && !nvlist_exists(feat, fi->fi_guid)) 499 return (ZPOOL_STATUS_FEAT_DISABLED); 500 if (!c_features[i] && nvlist_exists(feat, fi->fi_guid)) 501 return (ZPOOL_STATUS_INCOMPATIBLE_FEAT); 502 } 503 } 504 505 return (ZPOOL_STATUS_OK); 506 } 507 508 zpool_status_t 509 zpool_get_status(zpool_handle_t *zhp, char **msgid, zpool_errata_t *errata) 510 { 511 /* 512 * pass in the desired feature set, as 513 * it affects check for disabled features 514 */ 515 char compatibility[ZFS_MAXPROPLEN]; 516 if (zpool_get_prop(zhp, ZPOOL_PROP_COMPATIBILITY, compatibility, 517 ZFS_MAXPROPLEN, NULL, B_FALSE) != 0) 518 compatibility[0] = '\0'; 519 520 zpool_status_t ret = check_status(zhp->zpool_config, B_FALSE, errata, 521 compatibility); 522 523 if (msgid != NULL) { 524 if (ret >= NMSGID) 525 *msgid = NULL; 526 else 527 *msgid = zfs_msgid_table[ret]; 528 } 529 return (ret); 530 } 531 532 zpool_status_t 533 zpool_import_status(nvlist_t *config, char **msgid, zpool_errata_t *errata) 534 { 535 zpool_status_t ret = check_status(config, B_TRUE, errata, NULL); 536 537 if (ret >= NMSGID) 538 *msgid = NULL; 539 else 540 *msgid = zfs_msgid_table[ret]; 541 542 return (ret); 543 } 544