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 https://opensource.org/licenses/CDDL-1.0. 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 const char *const 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 uint_t c, children; 163 164 /* 165 * Ignore problems within a 'replacing' vdev, since we're presumably in 166 * the process of repairing any such errors, and don't want to call them 167 * out again. We'll pick up the fact that a resilver is happening 168 * later. 169 */ 170 if (ignore_replacing == B_TRUE) { 171 char *type = fnvlist_lookup_string(vdev, ZPOOL_CONFIG_TYPE); 172 if (strcmp(type, VDEV_TYPE_REPLACING) == 0) 173 return (B_FALSE); 174 } 175 176 if (nvlist_lookup_nvlist_array(vdev, ZPOOL_CONFIG_CHILDREN, &child, 177 &children) == 0) { 178 for (c = 0; c < children; c++) 179 if (find_vdev_problem(child[c], func, ignore_replacing)) 180 return (B_TRUE); 181 } else { 182 uint_t vsc; 183 vdev_stat_t *vs = (vdev_stat_t *)fnvlist_lookup_uint64_array( 184 vdev, ZPOOL_CONFIG_VDEV_STATS, &vsc); 185 if (func(vs, vsc) != 0) 186 return (B_TRUE); 187 } 188 189 /* 190 * Check any L2 cache devs 191 */ 192 if (nvlist_lookup_nvlist_array(vdev, ZPOOL_CONFIG_L2CACHE, &child, 193 &children) == 0) { 194 for (c = 0; c < children; c++) 195 if (find_vdev_problem(child[c], func, ignore_replacing)) 196 return (B_TRUE); 197 } 198 199 return (B_FALSE); 200 } 201 202 /* 203 * Active pool health status. 204 * 205 * To determine the status for a pool, we make several passes over the config, 206 * picking the most egregious error we find. In order of importance, we do the 207 * following: 208 * 209 * - Check for a complete and valid configuration 210 * - Look for any faulted or missing devices in a non-replicated config 211 * - Check for any data errors 212 * - Check for any faulted or missing devices in a replicated config 213 * - Look for any devices showing errors 214 * - Check for any resilvering or rebuilding devices 215 * 216 * There can obviously be multiple errors within a single pool, so this routine 217 * only picks the most damaging of all the current errors to report. 218 */ 219 static zpool_status_t 220 check_status(nvlist_t *config, boolean_t isimport, 221 zpool_errata_t *erratap, const char *compat) 222 { 223 pool_scan_stat_t *ps = NULL; 224 uint_t vsc, psc; 225 uint64_t suspended; 226 uint64_t hostid = 0; 227 uint64_t errata = 0; 228 unsigned long system_hostid = get_system_hostid(); 229 230 uint64_t version = fnvlist_lookup_uint64(config, ZPOOL_CONFIG_VERSION); 231 nvlist_t *nvroot = fnvlist_lookup_nvlist(config, 232 ZPOOL_CONFIG_VDEV_TREE); 233 vdev_stat_t *vs = (vdev_stat_t *)fnvlist_lookup_uint64_array(nvroot, 234 ZPOOL_CONFIG_VDEV_STATS, &vsc); 235 uint64_t stateval = fnvlist_lookup_uint64(config, 236 ZPOOL_CONFIG_POOL_STATE); 237 238 /* 239 * Currently resilvering a vdev 240 */ 241 (void) nvlist_lookup_uint64_array(nvroot, ZPOOL_CONFIG_SCAN_STATS, 242 (uint64_t **)&ps, &psc); 243 if (ps != NULL && ps->pss_func == POOL_SCAN_RESILVER && 244 ps->pss_state == DSS_SCANNING) 245 return (ZPOOL_STATUS_RESILVERING); 246 247 /* 248 * Currently rebuilding a vdev, check top-level vdevs. 249 */ 250 vdev_rebuild_stat_t *vrs = NULL; 251 nvlist_t **child; 252 uint_t c, i, children; 253 uint64_t rebuild_end_time = 0; 254 if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_CHILDREN, 255 &child, &children) == 0) { 256 for (c = 0; c < children; c++) { 257 if ((nvlist_lookup_uint64_array(child[c], 258 ZPOOL_CONFIG_REBUILD_STATS, 259 (uint64_t **)&vrs, &i) == 0) && (vrs != NULL)) { 260 uint64_t state = vrs->vrs_state; 261 262 if (state == VDEV_REBUILD_ACTIVE) { 263 return (ZPOOL_STATUS_REBUILDING); 264 } else if (state == VDEV_REBUILD_COMPLETE && 265 vrs->vrs_end_time > rebuild_end_time) { 266 rebuild_end_time = vrs->vrs_end_time; 267 } 268 } 269 } 270 271 /* 272 * If we can determine when the last scrub was run, and it 273 * was before the last rebuild completed, then recommend 274 * that the pool be scrubbed to verify all checksums. When 275 * ps is NULL we can infer the pool has never been scrubbed. 276 */ 277 if (rebuild_end_time > 0) { 278 if (ps != NULL) { 279 if ((ps->pss_state == DSS_FINISHED && 280 ps->pss_func == POOL_SCAN_SCRUB && 281 rebuild_end_time > ps->pss_end_time) || 282 ps->pss_state == DSS_NONE) 283 return (ZPOOL_STATUS_REBUILD_SCRUB); 284 } else { 285 return (ZPOOL_STATUS_REBUILD_SCRUB); 286 } 287 } 288 } 289 290 /* 291 * The multihost property is set and the pool may be active. 292 */ 293 if (vs->vs_state == VDEV_STATE_CANT_OPEN && 294 vs->vs_aux == VDEV_AUX_ACTIVE) { 295 mmp_state_t mmp_state; 296 nvlist_t *nvinfo; 297 298 nvinfo = fnvlist_lookup_nvlist(config, ZPOOL_CONFIG_LOAD_INFO); 299 mmp_state = fnvlist_lookup_uint64(nvinfo, 300 ZPOOL_CONFIG_MMP_STATE); 301 302 if (mmp_state == MMP_STATE_ACTIVE) 303 return (ZPOOL_STATUS_HOSTID_ACTIVE); 304 else if (mmp_state == MMP_STATE_NO_HOSTID) 305 return (ZPOOL_STATUS_HOSTID_REQUIRED); 306 else 307 return (ZPOOL_STATUS_HOSTID_MISMATCH); 308 } 309 310 /* 311 * Pool last accessed by another system. 312 */ 313 (void) nvlist_lookup_uint64(config, ZPOOL_CONFIG_HOSTID, &hostid); 314 if (hostid != 0 && (unsigned long)hostid != system_hostid && 315 stateval == POOL_STATE_ACTIVE) 316 return (ZPOOL_STATUS_HOSTID_MISMATCH); 317 318 /* 319 * Newer on-disk version. 320 */ 321 if (vs->vs_state == VDEV_STATE_CANT_OPEN && 322 vs->vs_aux == VDEV_AUX_VERSION_NEWER) 323 return (ZPOOL_STATUS_VERSION_NEWER); 324 325 /* 326 * Unsupported feature(s). 327 */ 328 if (vs->vs_state == VDEV_STATE_CANT_OPEN && 329 vs->vs_aux == VDEV_AUX_UNSUP_FEAT) { 330 nvlist_t *nvinfo = fnvlist_lookup_nvlist(config, 331 ZPOOL_CONFIG_LOAD_INFO); 332 if (nvlist_exists(nvinfo, ZPOOL_CONFIG_CAN_RDONLY)) 333 return (ZPOOL_STATUS_UNSUP_FEAT_WRITE); 334 return (ZPOOL_STATUS_UNSUP_FEAT_READ); 335 } 336 337 /* 338 * Check that the config is complete. 339 */ 340 if (vs->vs_state == VDEV_STATE_CANT_OPEN && 341 vs->vs_aux == VDEV_AUX_BAD_GUID_SUM) 342 return (ZPOOL_STATUS_BAD_GUID_SUM); 343 344 /* 345 * Check whether the pool has suspended. 346 */ 347 if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_SUSPENDED, 348 &suspended) == 0) { 349 uint64_t reason; 350 351 if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_SUSPENDED_REASON, 352 &reason) == 0 && reason == ZIO_SUSPEND_MMP) 353 return (ZPOOL_STATUS_IO_FAILURE_MMP); 354 355 if (suspended == ZIO_FAILURE_MODE_CONTINUE) 356 return (ZPOOL_STATUS_IO_FAILURE_CONTINUE); 357 return (ZPOOL_STATUS_IO_FAILURE_WAIT); 358 } 359 360 /* 361 * Could not read a log. 362 */ 363 if (vs->vs_state == VDEV_STATE_CANT_OPEN && 364 vs->vs_aux == VDEV_AUX_BAD_LOG) { 365 return (ZPOOL_STATUS_BAD_LOG); 366 } 367 368 /* 369 * Bad devices in non-replicated config. 370 */ 371 if (vs->vs_state == VDEV_STATE_CANT_OPEN && 372 find_vdev_problem(nvroot, vdev_faulted, B_TRUE)) 373 return (ZPOOL_STATUS_FAULTED_DEV_NR); 374 375 if (vs->vs_state == VDEV_STATE_CANT_OPEN && 376 find_vdev_problem(nvroot, vdev_missing, B_TRUE)) 377 return (ZPOOL_STATUS_MISSING_DEV_NR); 378 379 if (vs->vs_state == VDEV_STATE_CANT_OPEN && 380 find_vdev_problem(nvroot, vdev_broken, B_TRUE)) 381 return (ZPOOL_STATUS_CORRUPT_LABEL_NR); 382 383 /* 384 * Corrupted pool metadata 385 */ 386 if (vs->vs_state == VDEV_STATE_CANT_OPEN && 387 vs->vs_aux == VDEV_AUX_CORRUPT_DATA) 388 return (ZPOOL_STATUS_CORRUPT_POOL); 389 390 /* 391 * Persistent data errors. 392 */ 393 if (!isimport) { 394 uint64_t nerr; 395 if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_ERRCOUNT, 396 &nerr) == 0 && nerr != 0) 397 return (ZPOOL_STATUS_CORRUPT_DATA); 398 } 399 400 /* 401 * Missing devices in a replicated config. 402 */ 403 if (find_vdev_problem(nvroot, vdev_faulted, B_TRUE)) 404 return (ZPOOL_STATUS_FAULTED_DEV_R); 405 if (find_vdev_problem(nvroot, vdev_missing, B_TRUE)) 406 return (ZPOOL_STATUS_MISSING_DEV_R); 407 if (find_vdev_problem(nvroot, vdev_broken, B_TRUE)) 408 return (ZPOOL_STATUS_CORRUPT_LABEL_R); 409 410 /* 411 * Devices with errors 412 */ 413 if (!isimport && find_vdev_problem(nvroot, vdev_errors, B_TRUE)) 414 return (ZPOOL_STATUS_FAILING_DEV); 415 416 /* 417 * Offlined devices 418 */ 419 if (find_vdev_problem(nvroot, vdev_offlined, B_TRUE)) 420 return (ZPOOL_STATUS_OFFLINE_DEV); 421 422 /* 423 * Removed device 424 */ 425 if (find_vdev_problem(nvroot, vdev_removed, B_TRUE)) 426 return (ZPOOL_STATUS_REMOVED_DEV); 427 428 /* 429 * Suboptimal, but usable, ashift configuration. 430 */ 431 if (find_vdev_problem(nvroot, vdev_non_native_ashift, B_FALSE)) 432 return (ZPOOL_STATUS_NON_NATIVE_ASHIFT); 433 434 /* 435 * Informational errata available. 436 */ 437 (void) nvlist_lookup_uint64(config, ZPOOL_CONFIG_ERRATA, &errata); 438 if (errata) { 439 *erratap = errata; 440 return (ZPOOL_STATUS_ERRATA); 441 } 442 443 /* 444 * Outdated, but usable, version 445 */ 446 if (SPA_VERSION_IS_SUPPORTED(version) && version != SPA_VERSION) { 447 /* "legacy" compatibility disables old version reporting */ 448 if (compat != NULL && strcmp(compat, ZPOOL_COMPAT_LEGACY) == 0) 449 return (ZPOOL_STATUS_OK); 450 else 451 return (ZPOOL_STATUS_VERSION_OLDER); 452 } 453 454 /* 455 * Usable pool with disabled or superfluous features 456 * (superfluous = beyond what's requested by 'compatibility') 457 */ 458 if (version >= SPA_VERSION_FEATURES) { 459 int i; 460 nvlist_t *feat; 461 462 if (isimport) { 463 feat = fnvlist_lookup_nvlist(config, 464 ZPOOL_CONFIG_LOAD_INFO); 465 if (nvlist_exists(feat, ZPOOL_CONFIG_ENABLED_FEAT)) 466 feat = fnvlist_lookup_nvlist(feat, 467 ZPOOL_CONFIG_ENABLED_FEAT); 468 } else { 469 feat = fnvlist_lookup_nvlist(config, 470 ZPOOL_CONFIG_FEATURE_STATS); 471 } 472 473 /* check against all features, or limited set? */ 474 boolean_t c_features[SPA_FEATURES]; 475 476 switch (zpool_load_compat(compat, c_features, NULL, 0)) { 477 case ZPOOL_COMPATIBILITY_OK: 478 case ZPOOL_COMPATIBILITY_WARNTOKEN: 479 break; 480 default: 481 return (ZPOOL_STATUS_COMPATIBILITY_ERR); 482 } 483 for (i = 0; i < SPA_FEATURES; i++) { 484 zfeature_info_t *fi = &spa_feature_table[i]; 485 if (!fi->fi_zfs_mod_supported) 486 continue; 487 if (c_features[i] && !nvlist_exists(feat, fi->fi_guid)) 488 return (ZPOOL_STATUS_FEAT_DISABLED); 489 if (!c_features[i] && nvlist_exists(feat, fi->fi_guid)) 490 return (ZPOOL_STATUS_INCOMPATIBLE_FEAT); 491 } 492 } 493 494 return (ZPOOL_STATUS_OK); 495 } 496 497 zpool_status_t 498 zpool_get_status(zpool_handle_t *zhp, const char **msgid, 499 zpool_errata_t *errata) 500 { 501 /* 502 * pass in the desired feature set, as 503 * it affects check for disabled features 504 */ 505 char compatibility[ZFS_MAXPROPLEN]; 506 if (zpool_get_prop(zhp, ZPOOL_PROP_COMPATIBILITY, compatibility, 507 ZFS_MAXPROPLEN, NULL, B_FALSE) != 0) 508 compatibility[0] = '\0'; 509 510 zpool_status_t ret = check_status(zhp->zpool_config, B_FALSE, errata, 511 compatibility); 512 513 if (msgid != NULL) { 514 if (ret >= NMSGID) 515 *msgid = NULL; 516 else 517 *msgid = zfs_msgid_table[ret]; 518 } 519 return (ret); 520 } 521 522 zpool_status_t 523 zpool_import_status(nvlist_t *config, const char **msgid, 524 zpool_errata_t *errata) 525 { 526 zpool_status_t ret = check_status(config, B_TRUE, errata, NULL); 527 528 if (ret >= NMSGID) 529 *msgid = NULL; 530 else 531 *msgid = zfs_msgid_table[ret]; 532 533 return (ret); 534 } 535