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_OK 93 */ 94 }; 95 96 #define NMSGID (sizeof (zfs_msgid_table) / sizeof (zfs_msgid_table[0])) 97 98 /* ARGSUSED */ 99 static int 100 vdev_missing(vdev_stat_t *vs, uint_t vsc) 101 { 102 return (vs->vs_state == VDEV_STATE_CANT_OPEN && 103 vs->vs_aux == VDEV_AUX_OPEN_FAILED); 104 } 105 106 /* ARGSUSED */ 107 static int 108 vdev_faulted(vdev_stat_t *vs, uint_t vsc) 109 { 110 return (vs->vs_state == VDEV_STATE_FAULTED); 111 } 112 113 /* ARGSUSED */ 114 static int 115 vdev_errors(vdev_stat_t *vs, uint_t vsc) 116 { 117 return (vs->vs_state == VDEV_STATE_DEGRADED || 118 vs->vs_read_errors != 0 || vs->vs_write_errors != 0 || 119 vs->vs_checksum_errors != 0); 120 } 121 122 /* ARGSUSED */ 123 static int 124 vdev_broken(vdev_stat_t *vs, uint_t vsc) 125 { 126 return (vs->vs_state == VDEV_STATE_CANT_OPEN); 127 } 128 129 /* ARGSUSED */ 130 static int 131 vdev_offlined(vdev_stat_t *vs, uint_t vsc) 132 { 133 return (vs->vs_state == VDEV_STATE_OFFLINE); 134 } 135 136 /* ARGSUSED */ 137 static int 138 vdev_removed(vdev_stat_t *vs, uint_t vsc) 139 { 140 return (vs->vs_state == VDEV_STATE_REMOVED); 141 } 142 143 static int 144 vdev_non_native_ashift(vdev_stat_t *vs, uint_t vsc) 145 { 146 if (getenv("ZPOOL_STATUS_NON_NATIVE_ASHIFT_IGNORE") != NULL) 147 return (0); 148 149 return (VDEV_STAT_VALID(vs_physical_ashift, vsc) && 150 vs->vs_configured_ashift < vs->vs_physical_ashift); 151 } 152 153 /* 154 * Detect if any leaf devices that have seen errors or could not be opened. 155 */ 156 static boolean_t 157 find_vdev_problem(nvlist_t *vdev, int (*func)(vdev_stat_t *, uint_t), 158 boolean_t ignore_replacing) 159 { 160 nvlist_t **child; 161 vdev_stat_t *vs; 162 uint_t c, vsc, 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; 172 173 verify(nvlist_lookup_string(vdev, ZPOOL_CONFIG_TYPE, 174 &type) == 0); 175 if (strcmp(type, VDEV_TYPE_REPLACING) == 0) 176 return (B_FALSE); 177 } 178 179 if (nvlist_lookup_nvlist_array(vdev, ZPOOL_CONFIG_CHILDREN, &child, 180 &children) == 0) { 181 for (c = 0; c < children; c++) 182 if (find_vdev_problem(child[c], func, ignore_replacing)) 183 return (B_TRUE); 184 } else { 185 verify(nvlist_lookup_uint64_array(vdev, ZPOOL_CONFIG_VDEV_STATS, 186 (uint64_t **)&vs, &vsc) == 0); 187 188 if (func(vs, vsc) != 0) 189 return (B_TRUE); 190 } 191 192 /* 193 * Check any L2 cache devs 194 */ 195 if (nvlist_lookup_nvlist_array(vdev, ZPOOL_CONFIG_L2CACHE, &child, 196 &children) == 0) { 197 for (c = 0; c < children; c++) 198 if (find_vdev_problem(child[c], func, ignore_replacing)) 199 return (B_TRUE); 200 } 201 202 return (B_FALSE); 203 } 204 205 /* 206 * Active pool health status. 207 * 208 * To determine the status for a pool, we make several passes over the config, 209 * picking the most egregious error we find. In order of importance, we do the 210 * following: 211 * 212 * - Check for a complete and valid configuration 213 * - Look for any faulted or missing devices in a non-replicated config 214 * - Check for any data errors 215 * - Check for any faulted or missing devices in a replicated config 216 * - Look for any devices showing errors 217 * - Check for any resilvering or rebuilding devices 218 * 219 * There can obviously be multiple errors within a single pool, so this routine 220 * only picks the most damaging of all the current errors to report. 221 */ 222 static zpool_status_t 223 check_status(nvlist_t *config, boolean_t isimport, 224 zpool_errata_t *erratap, const char *compat) 225 { 226 nvlist_t *nvroot; 227 vdev_stat_t *vs; 228 pool_scan_stat_t *ps = NULL; 229 uint_t vsc, psc; 230 uint64_t nerr; 231 uint64_t version; 232 uint64_t stateval; 233 uint64_t suspended; 234 uint64_t hostid = 0; 235 uint64_t errata = 0; 236 unsigned long system_hostid = get_system_hostid(); 237 238 verify(nvlist_lookup_uint64(config, ZPOOL_CONFIG_VERSION, 239 &version) == 0); 240 verify(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, 241 &nvroot) == 0); 242 verify(nvlist_lookup_uint64_array(nvroot, ZPOOL_CONFIG_VDEV_STATS, 243 (uint64_t **)&vs, &vsc) == 0); 244 verify(nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_STATE, 245 &stateval) == 0); 246 247 /* 248 * Currently resilvering a vdev 249 */ 250 (void) nvlist_lookup_uint64_array(nvroot, ZPOOL_CONFIG_SCAN_STATS, 251 (uint64_t **)&ps, &psc); 252 if (ps != NULL && ps->pss_func == POOL_SCAN_RESILVER && 253 ps->pss_state == DSS_SCANNING) 254 return (ZPOOL_STATUS_RESILVERING); 255 256 /* 257 * Currently rebuilding a vdev, check top-level vdevs. 258 */ 259 vdev_rebuild_stat_t *vrs = NULL; 260 nvlist_t **child; 261 uint_t c, i, children; 262 uint64_t rebuild_end_time = 0; 263 if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_CHILDREN, 264 &child, &children) == 0) { 265 for (c = 0; c < children; c++) { 266 if ((nvlist_lookup_uint64_array(child[c], 267 ZPOOL_CONFIG_REBUILD_STATS, 268 (uint64_t **)&vrs, &i) == 0) && (vrs != NULL)) { 269 uint64_t state = vrs->vrs_state; 270 271 if (state == VDEV_REBUILD_ACTIVE) { 272 return (ZPOOL_STATUS_REBUILDING); 273 } else if (state == VDEV_REBUILD_COMPLETE && 274 vrs->vrs_end_time > rebuild_end_time) { 275 rebuild_end_time = vrs->vrs_end_time; 276 } 277 } 278 } 279 280 /* 281 * If we can determine when the last scrub was run, and it 282 * was before the last rebuild completed, then recommend 283 * that the pool be scrubbed to verify all checksums. When 284 * ps is NULL we can infer the pool has never been scrubbed. 285 */ 286 if (rebuild_end_time > 0) { 287 if (ps != NULL) { 288 if ((ps->pss_state == DSS_FINISHED && 289 ps->pss_func == POOL_SCAN_SCRUB && 290 rebuild_end_time > ps->pss_end_time) || 291 ps->pss_state == DSS_NONE) 292 return (ZPOOL_STATUS_REBUILD_SCRUB); 293 } else { 294 return (ZPOOL_STATUS_REBUILD_SCRUB); 295 } 296 } 297 } 298 299 /* 300 * The multihost property is set and the pool may be active. 301 */ 302 if (vs->vs_state == VDEV_STATE_CANT_OPEN && 303 vs->vs_aux == VDEV_AUX_ACTIVE) { 304 mmp_state_t mmp_state; 305 nvlist_t *nvinfo; 306 307 nvinfo = fnvlist_lookup_nvlist(config, ZPOOL_CONFIG_LOAD_INFO); 308 mmp_state = fnvlist_lookup_uint64(nvinfo, 309 ZPOOL_CONFIG_MMP_STATE); 310 311 if (mmp_state == MMP_STATE_ACTIVE) 312 return (ZPOOL_STATUS_HOSTID_ACTIVE); 313 else if (mmp_state == MMP_STATE_NO_HOSTID) 314 return (ZPOOL_STATUS_HOSTID_REQUIRED); 315 else 316 return (ZPOOL_STATUS_HOSTID_MISMATCH); 317 } 318 319 /* 320 * Pool last accessed by another system. 321 */ 322 (void) nvlist_lookup_uint64(config, ZPOOL_CONFIG_HOSTID, &hostid); 323 if (hostid != 0 && (unsigned long)hostid != system_hostid && 324 stateval == POOL_STATE_ACTIVE) 325 return (ZPOOL_STATUS_HOSTID_MISMATCH); 326 327 /* 328 * Newer on-disk version. 329 */ 330 if (vs->vs_state == VDEV_STATE_CANT_OPEN && 331 vs->vs_aux == VDEV_AUX_VERSION_NEWER) 332 return (ZPOOL_STATUS_VERSION_NEWER); 333 334 /* 335 * Unsupported feature(s). 336 */ 337 if (vs->vs_state == VDEV_STATE_CANT_OPEN && 338 vs->vs_aux == VDEV_AUX_UNSUP_FEAT) { 339 nvlist_t *nvinfo; 340 341 verify(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_LOAD_INFO, 342 &nvinfo) == 0); 343 if (nvlist_exists(nvinfo, ZPOOL_CONFIG_CAN_RDONLY)) 344 return (ZPOOL_STATUS_UNSUP_FEAT_WRITE); 345 return (ZPOOL_STATUS_UNSUP_FEAT_READ); 346 } 347 348 /* 349 * Check that the config is complete. 350 */ 351 if (vs->vs_state == VDEV_STATE_CANT_OPEN && 352 vs->vs_aux == VDEV_AUX_BAD_GUID_SUM) 353 return (ZPOOL_STATUS_BAD_GUID_SUM); 354 355 /* 356 * Check whether the pool has suspended. 357 */ 358 if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_SUSPENDED, 359 &suspended) == 0) { 360 uint64_t reason; 361 362 if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_SUSPENDED_REASON, 363 &reason) == 0 && reason == ZIO_SUSPEND_MMP) 364 return (ZPOOL_STATUS_IO_FAILURE_MMP); 365 366 if (suspended == ZIO_FAILURE_MODE_CONTINUE) 367 return (ZPOOL_STATUS_IO_FAILURE_CONTINUE); 368 return (ZPOOL_STATUS_IO_FAILURE_WAIT); 369 } 370 371 /* 372 * Could not read a log. 373 */ 374 if (vs->vs_state == VDEV_STATE_CANT_OPEN && 375 vs->vs_aux == VDEV_AUX_BAD_LOG) { 376 return (ZPOOL_STATUS_BAD_LOG); 377 } 378 379 /* 380 * Bad devices in non-replicated config. 381 */ 382 if (vs->vs_state == VDEV_STATE_CANT_OPEN && 383 find_vdev_problem(nvroot, vdev_faulted, B_TRUE)) 384 return (ZPOOL_STATUS_FAULTED_DEV_NR); 385 386 if (vs->vs_state == VDEV_STATE_CANT_OPEN && 387 find_vdev_problem(nvroot, vdev_missing, B_TRUE)) 388 return (ZPOOL_STATUS_MISSING_DEV_NR); 389 390 if (vs->vs_state == VDEV_STATE_CANT_OPEN && 391 find_vdev_problem(nvroot, vdev_broken, B_TRUE)) 392 return (ZPOOL_STATUS_CORRUPT_LABEL_NR); 393 394 /* 395 * Corrupted pool metadata 396 */ 397 if (vs->vs_state == VDEV_STATE_CANT_OPEN && 398 vs->vs_aux == VDEV_AUX_CORRUPT_DATA) 399 return (ZPOOL_STATUS_CORRUPT_POOL); 400 401 /* 402 * Persistent data errors. 403 */ 404 if (!isimport) { 405 if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_ERRCOUNT, 406 &nerr) == 0 && nerr != 0) 407 return (ZPOOL_STATUS_CORRUPT_DATA); 408 } 409 410 /* 411 * Missing devices in a replicated config. 412 */ 413 if (find_vdev_problem(nvroot, vdev_faulted, B_TRUE)) 414 return (ZPOOL_STATUS_FAULTED_DEV_R); 415 if (find_vdev_problem(nvroot, vdev_missing, B_TRUE)) 416 return (ZPOOL_STATUS_MISSING_DEV_R); 417 if (find_vdev_problem(nvroot, vdev_broken, B_TRUE)) 418 return (ZPOOL_STATUS_CORRUPT_LABEL_R); 419 420 /* 421 * Devices with errors 422 */ 423 if (!isimport && find_vdev_problem(nvroot, vdev_errors, B_TRUE)) 424 return (ZPOOL_STATUS_FAILING_DEV); 425 426 /* 427 * Offlined devices 428 */ 429 if (find_vdev_problem(nvroot, vdev_offlined, B_TRUE)) 430 return (ZPOOL_STATUS_OFFLINE_DEV); 431 432 /* 433 * Removed device 434 */ 435 if (find_vdev_problem(nvroot, vdev_removed, B_TRUE)) 436 return (ZPOOL_STATUS_REMOVED_DEV); 437 438 /* 439 * Suboptimal, but usable, ashift configuration. 440 */ 441 if (find_vdev_problem(nvroot, vdev_non_native_ashift, B_FALSE)) 442 return (ZPOOL_STATUS_NON_NATIVE_ASHIFT); 443 444 /* 445 * Informational errata available. 446 */ 447 (void) nvlist_lookup_uint64(config, ZPOOL_CONFIG_ERRATA, &errata); 448 if (errata) { 449 *erratap = errata; 450 return (ZPOOL_STATUS_ERRATA); 451 } 452 453 /* 454 * Outdated, but usable, version 455 */ 456 if (SPA_VERSION_IS_SUPPORTED(version) && version != SPA_VERSION) 457 return (ZPOOL_STATUS_VERSION_OLDER); 458 459 /* 460 * Usable pool with disabled features 461 */ 462 if (version >= SPA_VERSION_FEATURES) { 463 int i; 464 nvlist_t *feat; 465 466 if (isimport) { 467 feat = fnvlist_lookup_nvlist(config, 468 ZPOOL_CONFIG_LOAD_INFO); 469 if (nvlist_exists(feat, ZPOOL_CONFIG_ENABLED_FEAT)) 470 feat = fnvlist_lookup_nvlist(feat, 471 ZPOOL_CONFIG_ENABLED_FEAT); 472 } else { 473 feat = fnvlist_lookup_nvlist(config, 474 ZPOOL_CONFIG_FEATURE_STATS); 475 } 476 477 /* check against all features, or limited set? */ 478 boolean_t pool_features[SPA_FEATURES]; 479 480 if (zpool_load_compat(compat, pool_features, NULL, NULL) != 481 ZPOOL_COMPATIBILITY_OK) 482 return (ZPOOL_STATUS_COMPATIBILITY_ERR); 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 (pool_features[i] && 488 !nvlist_exists(feat, fi->fi_guid)) 489 return (ZPOOL_STATUS_FEAT_DISABLED); 490 } 491 } 492 493 return (ZPOOL_STATUS_OK); 494 } 495 496 zpool_status_t 497 zpool_get_status(zpool_handle_t *zhp, char **msgid, zpool_errata_t *errata) 498 { 499 /* 500 * pass in the desired feature set, as 501 * it affects check for disabled features 502 */ 503 char compatibility[ZFS_MAXPROPLEN]; 504 if (zpool_get_prop(zhp, ZPOOL_PROP_COMPATIBILITY, compatibility, 505 ZFS_MAXPROPLEN, NULL, B_FALSE) != 0) 506 compatibility[0] = '\0'; 507 508 zpool_status_t ret = check_status(zhp->zpool_config, B_FALSE, errata, 509 compatibility); 510 511 if (msgid != NULL) { 512 if (ret >= NMSGID) 513 *msgid = NULL; 514 else 515 *msgid = zfs_msgid_table[ret]; 516 } 517 return (ret); 518 } 519 520 zpool_status_t 521 zpool_import_status(nvlist_t *config, char **msgid, zpool_errata_t *errata) 522 { 523 zpool_status_t ret = check_status(config, B_TRUE, errata, NULL); 524 525 if (ret >= NMSGID) 526 *msgid = NULL; 527 else 528 *msgid = zfs_msgid_table[ret]; 529 530 return (ret); 531 } 532