struct mddev_dev *load_containers(void) { struct mdstat_ent *mdstat = mdstat_read(0, 0); struct mdstat_ent *ent; struct mddev_dev *d; struct mddev_dev *rv = NULL; struct map_ent *map = NULL, *me; if (!mdstat) return NULL; for (ent = mdstat; ent; ent = ent->next) if (ent->metadata_version && strncmp(ent->metadata_version, "external:", 9) == 0 && !is_subarray(&ent->metadata_version[9])) { d = xmalloc(sizeof(*d)); memset(d, 0, sizeof(*d)); me = map_by_devnm(&map, ent->dev); if (me) d->devname = xstrdup(me->path); else if (asprintf(&d->devname, "/dev/%s", ent->dev) < 0) { free(d); continue; } d->next = rv; rv = d; } free_mdstat(mdstat); map_free(map); return rv; }
int ack(int fd, int tmo) { struct metadata_update msg = { .len = 0 }; return send_message(fd, &msg, tmo); } int wait_reply(int fd, int tmo) { struct metadata_update msg; int err = receive_message(fd, &msg, tmo); /* mdmon sent extra data, but caller only cares that we got a * successful reply */ if (err == 0 && msg.len > 0) free(msg.buf); return err; } int connect_monitor(char *devname) { char path[100]; int sfd; long fl; struct sockaddr_un addr; int pos; char *c; pos = sprintf(path, "%s/", MDMON_DIR); if (is_subarray(devname)) { devname++; c = strchr(devname, '/'); if (!c) return -1; snprintf(&path[pos], c - devname + 1, "%s", devname); pos += c - devname; } else pos += sprintf(&path[pos], "%s", devname); sprintf(&path[pos], ".sock"); sfd = socket(PF_LOCAL, SOCK_STREAM, 0); if (sfd < 0) return -1; addr.sun_family = PF_LOCAL; strcpy(addr.sun_path, path); if (connect(sfd, &addr, sizeof(addr)) < 0) { close(sfd); return -1; } fl = fcntl(sfd, F_GETFL, 0); fl |= O_NONBLOCK; fcntl(sfd, F_SETFL, fl); return sfd; }
/* Not really Monitor but ... */ int Wait(char *dev) { struct stat stb; char devnm[32]; int rv = 1; int frozen_remaining = 3; if (stat(dev, &stb) != 0) { pr_err("Cannot find %s: %s\n", dev, strerror(errno)); return 2; } strcpy(devnm, stat2devnm(&stb)); while(1) { struct mdstat_ent *ms = mdstat_read(1, 0); struct mdstat_ent *e; for (e=ms ; e; e=e->next) if (strcmp(e->devnm, devnm) == 0) break; if (e && e->percent == RESYNC_NONE) { /* We could be in the brief pause before something * starts. /proc/mdstat doesn't show that, but * sync_action does. */ struct mdinfo mdi; char buf[21]; sysfs_init(&mdi, -1, devnm); if (sysfs_get_str(&mdi, NULL, "sync_action", buf, 20) > 0 && strcmp(buf,"idle\n") != 0) { e->percent = RESYNC_UNKNOWN; if (strcmp(buf, "frozen\n") == 0) { if (frozen_remaining == 0) e->percent = RESYNC_NONE; else frozen_remaining -= 1; } } } if (!e || e->percent == RESYNC_NONE) { if (e && e->metadata_version && strncmp(e->metadata_version, "external:", 9) == 0) { if (is_subarray(&e->metadata_version[9])) ping_monitor(&e->metadata_version[9]); else ping_monitor(devnm); } free_mdstat(ms); return rv; } free_mdstat(ms); rv = 0; mdstat_wait(5); } }
/* sets the proper subarray and container_dev according to the metadata * version super_by_fd does this automatically, this routine is meant as * a supplement for guess_super() */ static char *get_member_info(struct mdstat_ent *ent) { if (ent->metadata_version == NULL || strncmp(ent->metadata_version, "external:", 9) != 0) return NULL; if (is_subarray(&ent->metadata_version[9])) { char *subarray; subarray = strrchr(ent->metadata_version, '/'); return subarray + 1; } return NULL; }
/* Not really Monitor but ... */ int Wait(char *dev) { struct stat stb; int devnum; int rv = 1; if (stat(dev, &stb) != 0) { fprintf(stderr, Name ": Cannot find %s: %s\n", dev, strerror(errno)); return 2; } devnum = stat2devnum(&stb); while(1) { struct mdstat_ent *ms = mdstat_read(1, 0); struct mdstat_ent *e; for (e=ms ; e; e=e->next) if (e->devnum == devnum) break; if (!e || e->percent < 0) { if (e && e->metadata_version && strncmp(e->metadata_version, "external:", 9) == 0) { if (is_subarray(&e->metadata_version[9])) ping_monitor(&e->metadata_version[9]); else ping_monitor(devnum2devname(devnum)); } free_mdstat(ms); return rv; } free_mdstat(ms); rv = 0; mdstat_wait(5); } }
int main(int argc, char *argv[]) { char *container_name = NULL; char *devnm = NULL; int status = 0; int opt; int all = 0; int takeover = 0; int dofork = 1; static struct option options[] = { {"all", 0, NULL, 'a'}, {"takeover", 0, NULL, 't'}, {"help", 0, NULL, 'h'}, {"offroot", 0, NULL, OffRootOpt}, {"foreground", 0, NULL, 'F'}, {NULL, 0, NULL, 0} }; if (in_initrd()) { /* * set first char of argv[0] to @. This is used by * systemd to signal that the task was launched from * initrd/initramfs and should be preserved during shutdown */ argv[0][0] = '@'; } while ((opt = getopt_long(argc, argv, "thaF", options, NULL)) != -1) { switch (opt) { case 'a': container_name = argv[optind-1]; all = 1; break; case 't': takeover = 1; break; case 'F': dofork = 0; break; case OffRootOpt: argv[0][0] = '@'; break; case 'h': default: usage(); break; } } if (all == 0 && container_name == NULL) { if (argv[optind]) container_name = argv[optind]; } if (container_name == NULL) usage(); if (argc - optind > 1) usage(); if (strcmp(container_name, "/proc/mdstat") == 0) all = 1; if (all) { struct mdstat_ent *mdstat, *e; int container_len = strlen(container_name); /* launch an mdmon instance for each container found */ mdstat = mdstat_read(0, 0); for (e = mdstat; e; e = e->next) { if (e->metadata_version && strncmp(e->metadata_version, "external:", 9) == 0 && !is_subarray(&e->metadata_version[9])) { /* update cmdline so this mdmon instance can be * distinguished from others in a call to ps(1) */ if (strlen(e->devnm) <= (unsigned)container_len) { memset(container_name, 0, container_len); sprintf(container_name, "%s", e->devnm); } status |= mdmon(e->devnm, 1, takeover); } } free_mdstat(mdstat); return status; } else if (strncmp(container_name, "md", 2) == 0) { int id = devnm2devid(container_name); if (id) devnm = container_name; } else { struct stat st; if (stat(container_name, &st) == 0) devnm = xstrdup(stat2devnm(&st)); } if (!devnm) { pr_err("%s is not a valid md device name\n", container_name); exit(1); } return mdmon(devnm, dofork && do_fork(), takeover); }
static int add_new_arrays(struct mdstat_ent *mdstat, struct state **statelist, int test, struct alert_info *info) { struct mdstat_ent *mse; int new_found = 0; char *name; for (mse=mdstat; mse; mse=mse->next) if (mse->devnm[0] && (!mse->level || /* retrieve containers */ (strcmp(mse->level, "raid0") != 0 && strcmp(mse->level, "linear") != 0)) ) { struct state *st = xcalloc(1, sizeof *st); mdu_array_info_t array; int fd; name = get_md_name(mse->devnm); if (!name) { free(st); continue; } st->devname = xstrdup(name); if ((fd = open(st->devname, O_RDONLY)) < 0 || ioctl(fd, GET_ARRAY_INFO, &array)< 0) { /* no such array */ if (fd >=0) close(fd); put_md_name(st->devname); free(st->devname); if (st->metadata) { st->metadata->ss->free_super(st->metadata); free(st->metadata); } free(st); continue; } close(fd); st->next = *statelist; st->err = 1; st->from_auto = 1; strcpy(st->devnm, mse->devnm); st->percent = RESYNC_UNKNOWN; st->expected_spares = -1; if (mse->metadata_version && strncmp(mse->metadata_version, "external:", 9) == 0 && is_subarray(mse->metadata_version+9)) { char *sl; strcpy(st->parent_devnm, mse->metadata_version+10); sl = strchr(st->parent_devnm, '/'); *sl = 0; } else st->parent_devnm[0] = 0; *statelist = st; if (test) alert("TestMessage", st->devname, NULL, info); new_found = 1; } return new_found; }
static int check_array(struct state *st, struct mdstat_ent *mdstat, int test, struct alert_info *ainfo, int increments, char *prefer) { /* Update the state 'st' to reflect any changes shown in mdstat, * or found by directly examining the array, and return * '1' if the array is degraded, or '0' if it is optimal (or dead). */ struct { int state, major, minor; } info[MAX_DISKS]; mdu_array_info_t array; struct mdstat_ent *mse = NULL, *mse2; char *dev = st->devname; int fd = -1; int i; int remaining_disks; int last_disk; int new_array = 0; if (test) alert("TestMessage", dev, NULL, ainfo); if (st->devnm[0]) fd = open("/sys/block", O_RDONLY|O_DIRECTORY); if (fd >= 0) { /* Don't open the device unless it is present and * active in sysfs. */ char buf[10]; close(fd); fd = sysfs_open(st->devnm, NULL, "array_state"); if (fd < 0 || read(fd, buf, 10) < 5 || strncmp(buf,"clear",5) == 0 || strncmp(buf,"inact",5) == 0) { if (fd >= 0) close(fd); fd = sysfs_open(st->devnm, NULL, "level"); if (fd < 0 || read(fd, buf, 10) != 0) { if (fd >= 0) close(fd); if (!st->err) alert("DeviceDisappeared", dev, NULL, ainfo); st->err++; return 0; } } close(fd); } fd = open(dev, O_RDONLY); if (fd < 0) { if (!st->err) alert("DeviceDisappeared", dev, NULL, ainfo); st->err++; return 0; } fcntl(fd, F_SETFD, FD_CLOEXEC); if (ioctl(fd, GET_ARRAY_INFO, &array)<0) { if (!st->err) alert("DeviceDisappeared", dev, NULL, ainfo); st->err++; close(fd); return 0; } /* It's much easier to list what array levels can't * have a device disappear than all of them that can */ if (array.level == 0 || array.level == -1) { if (!st->err && !st->from_config) alert("DeviceDisappeared", dev, " Wrong-Level", ainfo); st->err++; close(fd); return 0; } if (st->devnm[0] == 0) strcpy(st->devnm, fd2devnm(fd)); for (mse2 = mdstat ; mse2 ; mse2=mse2->next) if (strcmp(mse2->devnm, st->devnm) == 0) { mse2->devnm[0] = 0; /* flag it as "used" */ mse = mse2; } if (!mse) { /* duplicated array in statelist * or re-created after reading mdstat*/ st->err++; close(fd); return 0; } /* this array is in /proc/mdstat */ if (array.utime == 0) /* external arrays don't update utime, so * just make sure it is always different. */ array.utime = st->utime + 1;; if (st->err) { /* New array appeared where previously had an error */ st->err = 0; st->percent = RESYNC_NONE; new_array = 1; alert("NewArray", st->devname, NULL, ainfo); } if (st->utime == array.utime && st->failed == array.failed_disks && st->working == array.working_disks && st->spare == array.spare_disks && (mse == NULL || ( mse->percent == st->percent ))) { close(fd); if ((st->active < st->raid) && st->spare == 0) return 1; else return 0; } if (st->utime == 0 && /* new array */ mse->pattern && strchr(mse->pattern, '_') /* degraded */ ) alert("DegradedArray", dev, NULL, ainfo); if (st->utime == 0 && /* new array */ st->expected_spares > 0 && array.spare_disks < st->expected_spares) alert("SparesMissing", dev, NULL, ainfo); if (st->percent < 0 && st->percent != RESYNC_UNKNOWN && mse->percent >= 0) alert("RebuildStarted", dev, NULL, ainfo); if (st->percent >= 0 && mse->percent >= 0 && (mse->percent / increments) > (st->percent / increments)) { char percentalert[15]; // "RebuildNN" (10 chars) or "RebuildStarted" (15 chars) if((mse->percent / increments) == 0) snprintf(percentalert, sizeof(percentalert), "RebuildStarted"); else snprintf(percentalert, sizeof(percentalert), "Rebuild%02d", mse->percent); alert(percentalert, dev, NULL, ainfo); } if (mse->percent == RESYNC_NONE && st->percent >= 0) { /* Rebuild/sync/whatever just finished. * If there is a number in /mismatch_cnt, * we should report that. */ struct mdinfo *sra = sysfs_read(-1, st->devnm, GET_MISMATCH); if (sra && sra->mismatch_cnt > 0) { char cnt[80]; snprintf(cnt, sizeof(cnt), " mismatches found: %d (on raid level %d)", sra->mismatch_cnt, array.level); alert("RebuildFinished", dev, cnt, ainfo); } else alert("RebuildFinished", dev, NULL, ainfo); if (sra) free(sra); } st->percent = mse->percent; remaining_disks = array.nr_disks; for (i=0; i<MAX_DISKS && remaining_disks > 0; i++) { mdu_disk_info_t disc; disc.number = i; if (ioctl(fd, GET_DISK_INFO, &disc) >= 0) { info[i].state = disc.state; info[i].major = disc.major; info[i].minor = disc.minor; if (disc.major || disc.minor) remaining_disks --; } else info[i].major = info[i].minor = 0; } last_disk = i; if (mse->metadata_version && strncmp(mse->metadata_version, "external:", 9) == 0 && is_subarray(mse->metadata_version+9)) { char *sl; strcpy(st->parent_devnm, mse->metadata_version+10); sl = strchr(st->parent_devnm, '/'); if (sl) *sl = 0; } else st->parent_devnm[0] = 0; if (st->metadata == NULL && st->parent_devnm[0] == 0) st->metadata = super_by_fd(fd, NULL); close(fd); for (i=0; i<MAX_DISKS; i++) { mdu_disk_info_t disc = {0,0,0,0,0}; int newstate=0; int change; char *dv = NULL; disc.number = i; if (i < last_disk && (info[i].major || info[i].minor)) { newstate = info[i].state; dv = map_dev_preferred( info[i].major, info[i].minor, 1, prefer); disc.state = newstate; disc.major = info[i].major; disc.minor = info[i].minor; } else newstate = (1 << MD_DISK_REMOVED); if (dv == NULL && st->devid[i]) dv = map_dev_preferred( major(st->devid[i]), minor(st->devid[i]), 1, prefer); change = newstate ^ st->devstate[i]; if (st->utime && change && !st->err && !new_array) { if ((st->devstate[i]&change)&(1<<MD_DISK_SYNC)) alert("Fail", dev, dv, ainfo); else if ((newstate & (1<<MD_DISK_FAULTY)) && (disc.major || disc.minor) && st->devid[i] == makedev(disc.major, disc.minor)) alert("FailSpare", dev, dv, ainfo); else if ((newstate&change)&(1<<MD_DISK_SYNC)) alert("SpareActive", dev, dv, ainfo); } st->devstate[i] = newstate; st->devid[i] = makedev(disc.major, disc.minor); } st->active = array.active_disks; st->working = array.working_disks; st->spare = array.spare_disks; st->failed = array.failed_disks; st->utime = array.utime; st->raid = array.raid_disks; st->err = 0; if ((st->active < st->raid) && st->spare == 0) return 1; return 0; }
int WaitClean(char *dev, int sock, int verbose) { int fd; struct mdinfo *mdi; int rv = 1; char devnm[32]; fd = open(dev, O_RDONLY); if (fd < 0) { if (verbose) pr_err("Couldn't open %s: %s\n", dev, strerror(errno)); return 1; } strcpy(devnm, fd2devnm(fd)); mdi = sysfs_read(fd, devnm, GET_VERSION|GET_LEVEL|GET_SAFEMODE); if (!mdi) { if (verbose) pr_err("Failed to read sysfs attributes for %s\n", dev); close(fd); return 0; } switch(mdi->array.level) { case LEVEL_LINEAR: case LEVEL_MULTIPATH: case 0: /* safemode delay is irrelevant for these levels */ rv = 0; } /* for internal metadata the kernel handles the final clean * transition, containers can never be dirty */ if (!is_subarray(mdi->text_version)) rv = 0; /* safemode disabled ? */ if (mdi->safe_mode_delay == 0) rv = 0; if (rv) { int state_fd = sysfs_open(fd2devnm(fd), NULL, "array_state"); char buf[20]; int delay = 5000; /* minimize the safe_mode_delay and prepare to wait up to 5s * for writes to quiesce */ sysfs_set_safemode(mdi, 1); /* wait for array_state to be clean */ while (1) { rv = read(state_fd, buf, sizeof(buf)); if (rv < 0) break; if (sysfs_match_word(buf, clean_states) <= 4) break; rv = sysfs_wait(state_fd, &delay); if (rv < 0 && errno != EINTR) break; lseek(state_fd, 0, SEEK_SET); } if (rv < 0) rv = 1; else if (fping_monitor(sock) == 0 || ping_monitor(mdi->text_version) == 0) { /* we need to ping to close the window between array * state transitioning to clean and the metadata being * marked clean */ rv = 0; } else rv = 1; if (rv && verbose) pr_err("Error waiting for %s to be clean\n", dev); /* restore the original safe_mode_delay */ sysfs_set_safemode(mdi, mdi->safe_mode_delay); close(state_fd); } sysfs_free(mdi); close(fd); return rv; }
int Manage_ro(char *devname, int fd, int readonly) { /* switch to readonly or rw * * requires >= 0.90.0 * first check that array is runing * use RESTART_ARRAY_RW or STOP_ARRAY_RO * */ mdu_array_info_t array; #ifndef MDASSEMBLE struct mdinfo *mdi; #endif if (md_get_version(fd) < 9000) { fprintf(stderr, Name ": need md driver version 0.90.0 or later\n"); return 1; } #ifndef MDASSEMBLE /* If this is an externally-manage array, we need to modify the * metadata_version so that mdmon doesn't undo our change. */ mdi = sysfs_read(fd, -1, GET_LEVEL|GET_VERSION); if (mdi && mdi->array.major_version == -1 && is_subarray(mdi->text_version)) { char vers[64]; strcpy(vers, "external:"); strcat(vers, mdi->text_version); if (readonly > 0) { int rv; /* We set readonly ourselves. */ vers[9] = '-'; sysfs_set_str(mdi, NULL, "metadata_version", vers); close(fd); rv = sysfs_set_str(mdi, NULL, "array_state", "readonly"); if (rv < 0) { fprintf(stderr, Name ": failed to set readonly for %s: %s\n", devname, strerror(errno)); vers[9] = mdi->text_version[0]; sysfs_set_str(mdi, NULL, "metadata_version", vers); return 1; } } else { char *cp; /* We cannot set read/write - must signal mdmon */ vers[9] = '/'; sysfs_set_str(mdi, NULL, "metadata_version", vers); cp = strchr(vers+10, '/'); if (*cp) *cp = 0; ping_monitor(vers+10); if (mdi->array.level <= 0) sysfs_set_str(mdi, NULL, "array_state", "active"); } return 0; } #endif if (ioctl(fd, GET_ARRAY_INFO, &array)) { fprintf(stderr, Name ": %s does not appear to be active.\n", devname); return 1; } if (readonly>0) { if (ioctl(fd, STOP_ARRAY_RO, NULL)) { fprintf(stderr, Name ": failed to set readonly for %s: %s\n", devname, strerror(errno)); return 1; } } else if (readonly < 0) { if (ioctl(fd, RESTART_ARRAY_RW, NULL)) { fprintf(stderr, Name ": failed to set writable for %s: %s\n", devname, strerror(errno)); return 1; } } return 0; }
int Manage_runstop(char *devname, int fd, int runstop, int quiet) { /* Run or stop the array. array must already be configured * required >= 0.90.0 * Only print failure messages if quiet == 0; * quiet > 0 means really be quiet * quiet < 0 means we will try again if it fails. */ mdu_param_t param; /* unused */ if (runstop == -1 && md_get_version(fd) < 9000) { if (ioctl(fd, STOP_MD, 0)) { if (quiet == 0) fprintf(stderr, Name ": stopping device %s " "failed: %s\n", devname, strerror(errno)); return 1; } } if (md_get_version(fd) < 9000) { fprintf(stderr, Name ": need md driver version 0.90.0 or later\n"); return 1; } /* if (ioctl(fd, GET_ARRAY_INFO, &array)) { fprintf(stderr, Name ": %s does not appear to be active.\n", devname); return 1; } */ if (runstop>0) { if (ioctl(fd, RUN_ARRAY, ¶m)) { fprintf(stderr, Name ": failed to run array %s: %s\n", devname, strerror(errno)); return 1; } if (quiet <= 0) fprintf(stderr, Name ": started %s\n", devname); } else if (runstop < 0){ struct map_ent *map = NULL; struct stat stb; struct mdinfo *mdi; int devnum; int err; int count; /* If this is an mdmon managed array, just write 'inactive' * to the array state and let mdmon clear up. */ devnum = fd2devnum(fd); /* Get EXCL access first. If this fails, then attempting * to stop is probably a bad idea. */ close(fd); fd = open(devname, O_RDONLY|O_EXCL); if (fd < 0 || fd2devnum(fd) != devnum) { if (fd >= 0) close(fd); fprintf(stderr, Name ": Cannot get exclusive access to %s:" "Perhaps a running " "process, mounted filesystem " "or active volume group?\n", devname); return 1; } mdi = sysfs_read(fd, -1, GET_LEVEL|GET_VERSION); if (mdi && mdi->array.level > 0 && is_subarray(mdi->text_version)) { int err; /* This is mdmon managed. */ close(fd); count = 25; while (count && (err = sysfs_set_str(mdi, NULL, "array_state", "inactive")) < 0 && errno == EBUSY) { usleep(200000); count--; } if (err && !quiet) { fprintf(stderr, Name ": failed to stop array %s: %s\n", devname, strerror(errno)); return 1; } /* Give monitor a chance to act */ ping_monitor(mdi->text_version); fd = open_dev_excl(devnum); if (fd < 0) { fprintf(stderr, Name ": failed to completely stop %s" ": Device is busy\n", devname); return 1; } } else if (mdi && mdi->array.major_version == -1 && mdi->array.minor_version == -2 && !is_subarray(mdi->text_version)) { struct mdstat_ent *mds, *m; /* container, possibly mdmon-managed. * Make sure mdmon isn't opening it, which * would interfere with the 'stop' */ ping_monitor(mdi->sys_name); /* now check that there are no existing arrays * which are members of this array */ mds = mdstat_read(0, 0); for (m=mds; m; m=m->next) if (m->metadata_version && strncmp(m->metadata_version, "external:", 9)==0 && is_subarray(m->metadata_version+9) && devname2devnum(m->metadata_version+10) == devnum) { if (!quiet) fprintf(stderr, Name ": Cannot stop container %s: " "member %s still active\n", devname, m->dev); free_mdstat(mds); if (mdi) sysfs_free(mdi); return 1; } } /* As we have an O_EXCL open, any use of the device * which blocks STOP_ARRAY is probably a transient use, * so it is reasonable to retry for a while - 5 seconds. */ count = 25; err = 0; while (count && fd >= 0 && (err = ioctl(fd, STOP_ARRAY, NULL)) < 0 && errno == EBUSY) { usleep(200000); count --; } if (fd >= 0 && err) { if (quiet == 0) { fprintf(stderr, Name ": failed to stop array %s: %s\n", devname, strerror(errno)); if (errno == EBUSY) fprintf(stderr, "Perhaps a running " "process, mounted filesystem " "or active volume group?\n"); } if (mdi) sysfs_free(mdi); return 1; } /* prior to 2.6.28, KOBJ_CHANGE was not sent when an md array * was stopped, so We'll do it here just to be sure. Drop any * partitions as well... */ if (fd >= 0) ioctl(fd, BLKRRPART, 0); if (mdi) sysfs_uevent(mdi, "change"); if (devnum != NoMdDev && (stat("/dev/.udev", &stb) != 0 || check_env("MDADM_NO_UDEV"))) { struct map_ent *mp = map_by_devnum(&map, devnum); remove_devices(devnum, mp ? mp->path : NULL); } if (quiet <= 0) fprintf(stderr, Name ": stopped %s\n", devname); map_lock(&map); map_remove(&map, devnum); map_unlock(&map); } return 0; }