X-Git-Url: http://git.ipfire.org/?p=thirdparty%2Fmdadm.git;a=blobdiff_plain;f=monitor.c;h=81537ed37300159c6d260e68c0d9798fca01c446;hp=5c17910fe535249452cb908ffa39b43f5a950205;hb=1b7eb962db2cf9179d097e06cce74b84ac80e49d;hpb=6c40598f598874d1d4c2c4d0da0c2a9b873d768d diff --git a/monitor.c b/monitor.c index 5c17910f..81537ed3 100644 --- a/monitor.c +++ b/monitor.c @@ -1,8 +1,8 @@ /* * mdmon - monitor external metadata arrays * - * Copyright (C) 2007-2008 Neil Brown - * Copyright (C) 2007-2008 Intel Corporation + * Copyright (C) 2007-2009 Neil Brown + * Copyright (C) 2007-2009 Intel Corporation * * This program is free software; you can redistribute it and/or modify it * under the terms and conditions of the GNU General Public License, @@ -31,6 +31,11 @@ static char *sync_actions[] = { "idle", "reshape", "resync", "recover", "check", "repair", NULL }; +enum bb_action { + RECORD_BB = 1, + COMPARE_BB, +}; + static int write_attr(char *attr, int fd) { return write(fd, attr, strlen(attr)); @@ -38,8 +43,17 @@ static int write_attr(char *attr, int fd) static void add_fd(fd_set *fds, int *maxfd, int fd) { + struct stat st; if (fd < 0) return; + if (fstat(fd, &st) == -1) { + dprintf("Invalid fd %d\n", fd); + return; + } + if (st.st_nlink == 0) { + dprintf("fd %d was deleted\n", fd); + return; + } if (fd > *maxfd) *maxfd = fd; FD_SET(fd, fds); @@ -66,20 +80,39 @@ static int read_attr(char *buf, int len, int fd) return n; } -int get_resync_start(struct active_array *a) +static void read_resync_start(int fd, unsigned long long *v) { char buf[30]; int n; - n = read_attr(buf, 30, a->resync_start_fd); - if (n <= 0) - return n; + n = read_attr(buf, 30, fd); + if (n <= 0) { + dprintf("Failed to read resync_start (%d)\n", fd); + return; + } + if (strncmp(buf, "none", 4) == 0) + *v = MaxSector; + else + *v = strtoull(buf, NULL, 10); +} - a->resync_start = strtoull(buf, NULL, 10); +static unsigned long long read_sync_completed(int fd) +{ + unsigned long long val; + char buf[50]; + int n; + char *ep; - return 1; -} + n = read_attr(buf, 50, fd); + if (n <= 0) + return 0; + buf[n] = 0; + val = strtoull(buf, &ep, 0); + if (ep == buf || (*ep != 0 && *ep != '\n' && *ep != ' ')) + return 0; + return val; +} static enum array_state read_state(int fd) { @@ -103,8 +136,8 @@ static enum sync_action read_action( int fd) int read_dev_state(int fd) { - char buf[60]; - int n = read_attr(buf, 60, fd); + char buf[100]; + int n = read_attr(buf, sizeof(buf), fd); char *cp; int rv = 0; @@ -130,6 +163,177 @@ int read_dev_state(int fd) return rv; } +int process_ubb(struct active_array *a, struct mdinfo *mdi, const unsigned long + long sector, const int length, const char *buf, + const int buf_len) +{ + struct superswitch *ss = a->container->ss; + + /* + * record bad block in metadata first, then acknowledge it to the driver + * via sysfs file + */ + if ((ss->record_bad_block(a, mdi->disk.raid_disk, sector, length)) && + (write(mdi->bb_fd, buf, buf_len) == buf_len)) + return 1; + + /* + * failed to store or acknowledge bad block, switch of bad block support + * to get it out of blocked state + */ + sysfs_set_str(&a->info, mdi, "state", "-external_bbl"); + return -1; +} + +int compare_bb(struct active_array *a, struct mdinfo *mdi, const unsigned long + long sector, const unsigned int length, void *arg) +{ + struct superswitch *ss = a->container->ss; + struct md_bb *bb = (struct md_bb *) arg; + int record = 1; + int i; + + for (i = 0; i < bb->count; i++) { + unsigned long long start = bb->entries[i].sector; + unsigned long long len = bb->entries[i].length; + + /* + * bad block in metadata exactly matches bad block in kernel + * list, just remove it from a list + */ + if ((start == sector) && (len == length)) { + if (i < bb->count - 1) + bb->entries[i] = bb->entries[bb->count - 1]; + bb->count -= 1; + record = 0; + break; + } + /* + * bad block in metadata spans bad block in kernel list, + * clear it and record new bad block + */ + if ((sector >= start) && (sector + length <= start + len)) { + ss->clear_bad_block(a, mdi->disk.raid_disk, start, len); + break; + } + } + + /* record all bad blocks not in metadata list */ + if (record && (ss->record_bad_block(a, mdi->disk.raid_disk, sector, + length) <= 0)) { + sysfs_set_str(&a->info, mdi, "state", "-external_bbl"); + return -1; + } + + return 1; +} + +static int read_bb_file(int fd, struct active_array *a, struct mdinfo *mdi, + enum bb_action action, void *arg) +{ + char buf[30]; + int n = 0; + int ret = 0; + int read_again = 0; + int off = 0; + int pos = 0; + int preserve_pos = (action == RECORD_BB ? 0 : 1); + + if (lseek(fd, 0, SEEK_SET) == (off_t) -1) + return -1; + + do { + read_again = 0; + n = read(fd, buf + pos, sizeof(buf) - 1 - pos); + if (n < 0) + return -1; + n += pos; + + buf[n] = '\0'; + off = 0; + + while (off < n) { + unsigned long long sector; + int length; + char newline; + int consumed; + int matched; + int rc; + + /* kernel sysfs file format: "sector length\n" */ + matched = sscanf(buf + off, "%llu %d%c%n", §or, + &length, &newline, &consumed); + if ((matched != 3) && (off > 0)) { + /* truncated entry, read again */ + if (preserve_pos) { + pos = sizeof(buf) - off - 1; + memmove(buf, buf + off, pos); + } else { + if (lseek(fd, 0, SEEK_SET) == + (off_t) -1) + return -1; + } + read_again = 1; + break; + } + if (matched != 3) + return -1; + if (newline != '\n') + return -1; + if (length <= 0) + return -1; + + if (action == RECORD_BB) + rc = process_ubb(a, mdi, sector, length, + buf + off, consumed); + else if (action == COMPARE_BB) + rc = compare_bb(a, mdi, sector, length, arg); + else + rc = -1; + + if (rc < 0) + return rc; + ret += rc; + off += consumed; + } + } while (read_again); + + return ret; +} + +static int process_dev_ubb(struct active_array *a, struct mdinfo *mdi) +{ + return read_bb_file(mdi->ubb_fd, a, mdi, RECORD_BB, NULL); +} + +static int check_for_cleared_bb(struct active_array *a, struct mdinfo *mdi) +{ + struct superswitch *ss = a->container->ss; + struct md_bb *bb; + int i; + + /* + * Get a list of bad blocks for an array, then read list of + * acknowledged bad blocks from kernel and compare it against metadata + * list, clear all bad blocks remaining in metadata list + */ + bb = ss->get_bad_blocks(a, mdi->disk.raid_disk); + if (!bb) + return -1; + + if (read_bb_file(mdi->bb_fd, a, mdi, COMPARE_BB, bb) < 0) + return -1; + + for (i = 0; i < bb->count; i++) { + unsigned long long sector = bb->entries[i].sector; + int length = bb->entries[i].length; + + ss->clear_bad_block(a, mdi->disk.raid_disk, sector, length); + } + + return 0; +} + static void signal_manager(void) { /* tgkill(getpid(), mon_tid, SIGUSR1); */ @@ -194,47 +398,90 @@ static void signal_manager(void) * */ -static int read_and_act(struct active_array *a) +#define ARRAY_DIRTY 1 +#define ARRAY_BUSY 2 +static int read_and_act(struct active_array *a, fd_set *fds) { + unsigned long long sync_completed; int check_degraded = 0; + int check_reshape = 0; int deactivate = 0; struct mdinfo *mdi; + int ret = 0; + int count = 0; + struct timeval tv; a->next_state = bad_word; a->next_action = bad_action; a->curr_state = read_state(a->info.state_fd); a->curr_action = read_action(a->action_fd); + if (a->curr_state != clear) + /* + * In "clear" state, resync_start may wrongly be set to "0" + * when the kernel called md_clean but didn't remove the + * sysfs attributes yet + */ + read_resync_start(a->resync_start_fd, &a->info.resync_start); + sync_completed = read_sync_completed(a->sync_completed_fd); for (mdi = a->info.devs; mdi ; mdi = mdi->next) { mdi->next_state = 0; - if (mdi->state_fd >= 0) + mdi->curr_state = 0; + if (mdi->state_fd >= 0) { + read_resync_start(mdi->recovery_fd, + &mdi->recovery_start); mdi->curr_state = read_dev_state(mdi->state_fd); + } + /* + * If array is blocked and metadata handler is able to handle + * BB, check if you can acknowledge them to md driver. If + * successful, clear faulty state and unblock the array. + */ + if ((mdi->curr_state & DS_BLOCKED) && + a->container->ss->record_bad_block && + (process_dev_ubb(a, mdi) > 0)) { + mdi->next_state |= DS_UNBLOCK; + } + if (FD_ISSET(mdi->bb_fd, fds)) + check_for_cleared_bb(a, mdi); } - if (a->curr_state <= inactive && + gettimeofday(&tv, NULL); + dprintf("(%d): %ld.%06ld state:%s prev:%s action:%s prev: %s start:%llu\n", + a->info.container_member, + tv.tv_sec, tv.tv_usec, + array_states[a->curr_state], + array_states[a->prev_state], + sync_actions[a->curr_action], + sync_actions[a->prev_action], + a->info.resync_start + ); + + if ((a->curr_state == bad_word || a->curr_state <= inactive) && a->prev_state > inactive) { /* array has been stopped */ - get_resync_start(a); a->container->ss->set_array_state(a, 1); a->next_state = clear; deactivate = 1; } if (a->curr_state == write_pending) { - get_resync_start(a); a->container->ss->set_array_state(a, 0); a->next_state = active; + ret |= ARRAY_DIRTY; } if (a->curr_state == active_idle) { /* Set array to 'clean' FIRST, then mark clean * in the metadata */ a->next_state = clean; + ret |= ARRAY_DIRTY; } if (a->curr_state == clean) { - get_resync_start(a); a->container->ss->set_array_state(a, 1); } - + if (a->curr_state == active || + a->curr_state == suspended) + ret |= ARRAY_DIRTY; if (a->curr_state == readonly) { /* Well, I'm ready to handle things. If readonly * wasn't requested, transition to read-auto. @@ -245,11 +492,12 @@ static int read_and_act(struct active_array *a) /* explicit request for readonly array. Leave it alone */ ; } else { - get_resync_start(a); if (a->container->ss->set_array_state(a, 2)) a->next_state = read_auto; /* array is clean */ - else + else { a->next_state = active; /* Now active for recovery etc */ + ret |= ARRAY_DIRTY; + } } } @@ -261,7 +509,6 @@ static int read_and_act(struct active_array *a) * until the array goes inactive or readonly though. * Just check if we need to fiddle spares. */ - get_resync_start(a); a->container->ss->set_array_state(a, a->curr_state <= clean); check_degraded = 1; } @@ -277,9 +524,21 @@ static int read_and_act(struct active_array *a) mdi->curr_state); if (! (mdi->curr_state & DS_INSYNC)) check_degraded = 1; + count++; } + if (count != a->info.array.raid_disks) + check_degraded = 1; } + if (!deactivate && + a->curr_action == reshape && + a->prev_action != reshape) + /* reshape was requested by mdadm. Need to see if + * new devices have been added. Manager does that + * when it sees check_reshape + */ + check_reshape = 1; + /* Check for failures and if found: * 1/ Record the failure in the metadata and unblock the device. * FIXME update the kernel to stop notifying on failed drives when @@ -292,7 +551,8 @@ static int read_and_act(struct active_array *a) a->container->ss->set_disk(a, mdi->disk.raid_disk, mdi->curr_state); check_degraded = 1; - mdi->next_state |= DS_UNBLOCK; + if (mdi->curr_state & DS_BLOCKED) + mdi->next_state |= DS_UNBLOCK; if (a->curr_state == read_auto) { a->container->ss->set_array_state(a, 0); a->next_state = active; @@ -302,45 +562,98 @@ static int read_and_act(struct active_array *a) } } + /* Check for recovery checkpoint notifications. We need to be a + * minimum distance away from the last checkpoint to prevent + * over checkpointing. Note reshape checkpointing is handled + * in the second branch. + */ + if (sync_completed > a->last_checkpoint && + sync_completed - a->last_checkpoint > a->info.component_size >> 4 && + a->curr_action > reshape) { + /* A (non-reshape) sync_action has reached a checkpoint. + * Record the updated position in the metadata + */ + a->last_checkpoint = sync_completed; + a->container->ss->set_array_state(a, a->curr_state <= clean); + } else if ((a->curr_action == idle && a->prev_action == reshape) || + (a->curr_action == reshape && + sync_completed > a->last_checkpoint)) { + /* Reshape has progressed or completed so we need to + * update the array state - and possibly the array size + */ + if (sync_completed != 0) + a->last_checkpoint = sync_completed; + /* We might need to update last_checkpoint depending on + * the reason that reshape finished. + * if array reshape is really finished: + * set check point to the end, this allows + * set_array_state() to finalize reshape in metadata + * if reshape if broken: do not set checkpoint to the end + * this allows for reshape restart from checkpoint + */ + if ((a->curr_action != reshape) && + (a->prev_action == reshape)) { + char buf[40]; + if ((sysfs_get_str(&a->info, NULL, + "reshape_position", + buf, + sizeof(buf)) >= 0) && + strncmp(buf, "none", 4) == 0) + a->last_checkpoint = a->info.component_size; + } + a->container->ss->set_array_state(a, a->curr_state <= clean); + a->last_checkpoint = sync_completed; + } + + if (sync_completed > a->last_checkpoint) + a->last_checkpoint = sync_completed; + + if (sync_completed >= a->info.component_size) + a->last_checkpoint = 0; + a->container->ss->sync_metadata(a->container); - dprintf("%s(%d): state:%s action:%s next(", __func__, a->info.container_member, + dprintf("(%d): state:%s action:%s next(", a->info.container_member, array_states[a->curr_state], sync_actions[a->curr_action]); /* Effect state changes in the array */ if (a->next_state != bad_word) { - dprintf(" state:%s", array_states[a->next_state]); + dprintf_cont(" state:%s", array_states[a->next_state]); write_attr(array_states[a->next_state], a->info.state_fd); } if (a->next_action != bad_action) { write_attr(sync_actions[a->next_action], a->action_fd); - dprintf(" action:%s", sync_actions[a->next_action]); + dprintf_cont(" action:%s", sync_actions[a->next_action]); } for (mdi = a->info.devs; mdi ; mdi = mdi->next) { if (mdi->next_state & DS_UNBLOCK) { - dprintf(" %d:-blocked", mdi->disk.raid_disk); + dprintf_cont(" %d:-blocked", mdi->disk.raid_disk); write_attr("-blocked", mdi->state_fd); } if ((mdi->next_state & DS_REMOVE) && mdi->state_fd >= 0) { int remove_result; - /* the kernel may not be able to immediately remove the - * disk, we can simply wait until the next event to try - * again. + /* The kernel may not be able to immediately remove the + * disk. In that case we wait a little while and + * try again. */ remove_result = write_attr("remove", mdi->state_fd); if (remove_result > 0) { - dprintf(" %d:removed", mdi->disk.raid_disk); + dprintf_cont(" %d:removed", mdi->disk.raid_disk); close(mdi->state_fd); + close(mdi->recovery_fd); + close(mdi->bb_fd); + close(mdi->ubb_fd); mdi->state_fd = -1; - } + } else + ret |= ARRAY_BUSY; } if (mdi->next_state & DS_INSYNC) { write_attr("+in_sync", mdi->state_fd); - dprintf(" %d:+in_sync", mdi->disk.raid_disk); + dprintf_cont(" %d:+in_sync", mdi->disk.raid_disk); } } - dprintf(" )\n"); + dprintf_cont(" )\n"); /* move curr_ to prev_ */ a->prev_state = a->curr_state; @@ -352,16 +665,19 @@ static int read_and_act(struct active_array *a) mdi->next_state = 0; } - if (check_degraded) { + if (check_degraded || check_reshape) { /* manager will do the actual check */ - a->check_degraded = 1; + if (check_degraded) + a->check_degraded = 1; + if (check_reshape) + a->check_reshape = 1; signal_manager(); } if (deactivate) a->container = NULL; - return 1; + return ret; } static struct mdinfo * @@ -382,7 +698,7 @@ static void reconcile_failed(struct active_array *aa, struct mdinfo *failed) struct mdinfo *victim; for (a = aa; a; a = a->next) { - if (!a->container) + if (!a->container || a->to_remove) continue; victim = find_device(a, failed->disk.major, failed->disk.minor); if (!victim) @@ -442,7 +758,7 @@ static int wait_and_act(struct supertype *container, int nowait) /* once an array has been deactivated we want to * ask the manager to discard it. */ - if (!a->container) { + if (!a->container || a->to_remove) { if (discard_this) { ap = &(*ap)->next; continue; @@ -456,8 +772,12 @@ static int wait_and_act(struct supertype *container, int nowait) add_fd(&rfds, &maxfd, a->info.state_fd); add_fd(&rfds, &maxfd, a->action_fd); - for (mdi = a->info.devs ; mdi ; mdi = mdi->next) + add_fd(&rfds, &maxfd, a->sync_completed_fd); + for (mdi = a->info.devs ; mdi ; mdi = mdi->next) { add_fd(&rfds, &maxfd, mdi->state_fd); + add_fd(&rfds, &maxfd, mdi->bb_fd); + add_fd(&rfds, &maxfd, mdi->ubb_fd); + } ap = &(*ap)->next; } @@ -469,33 +789,58 @@ static int wait_and_act(struct supertype *container, int nowait) * problem as there are no active arrays, there is * nothing that we need to be ready to do. */ - int fd = open_dev_excl(container->devnum); + int fd; + if (sigterm) + fd = open_dev_excl(container->devnm); + else + fd = open_dev_flags(container->devnm, O_RDONLY|O_EXCL); if (fd >= 0 || errno != EBUSY) { /* OK, we are safe to leave */ if (sigterm && !dirty_arrays) dprintf("caught sigterm, all clean... exiting\n"); else dprintf("no arrays to monitor... exiting\n"); - remove_pidfile(container->devname); + if (!sigterm) + /* On SIGTERM, someone (the take-over mdmon) will + * clean up + */ + remove_pidfile(container->devnm); exit_now = 1; signal_manager(); + close(fd); exit(0); } } if (!nowait) { sigset_t set; + struct timespec ts; + ts.tv_sec = 24*3600; + ts.tv_nsec = 0; + if (*aap == NULL || container->retry_soon) { + /* just waiting to get O_EXCL access */ + ts.tv_sec = 0; + ts.tv_nsec = 20000000ULL; + } sigprocmask(SIG_UNBLOCK, NULL, &set); sigdelset(&set, SIGUSR1); monitor_loop_cnt |= 1; - rv = pselect(maxfd+1, &rfds, NULL, NULL, NULL, &set); + rv = pselect(maxfd+1, NULL, NULL, &rfds, &ts, &set); monitor_loop_cnt += 1; - if (rv == -1 && errno == EINTR) - rv = 0; + if (rv == -1) { + if (errno == EINTR) { + rv = 0; + FD_ZERO(&rfds); + dprintf("monitor: caught signal\n"); + } else + dprintf("monitor: error %d in pselect\n", + errno); + } #ifdef DEBUG - dprint_wake_reasons(&rfds); + else + dprint_wake_reasons(&rfds); #endif - + container->retry_soon = 0; } if (update_queue) { @@ -513,7 +858,6 @@ static int wait_and_act(struct supertype *container, int nowait) rv = 0; dirty_arrays = 0; for (a = *aap; a ; a = a->next) { - int is_dirty; if (a->replaces && !discard_this) { struct active_array **ap; @@ -527,37 +871,24 @@ static int wait_and_act(struct supertype *container, int nowait) /* FIXME check if device->state_fd need to be cleared?*/ signal_manager(); } - if (a->container) - rv += read_and_act(a); - else - continue; - - /* when terminating stop manipulating the array after it is - * clean, but make sure read_and_act() is given a chance to - * handle 'active_idle' - */ - switch (read_state(a->info.state_fd)) { - case active: - case active_idle: - case suspended: - case bad_word: - is_dirty = 1; - break; - default: - if (a->curr_state == active_idle) - is_dirty = 1; - else - is_dirty = 0; - break; + if (a->container && !a->to_remove) { + int ret = read_and_act(a, &rfds); + rv |= 1; + dirty_arrays += !!(ret & ARRAY_DIRTY); + /* when terminating stop manipulating the array after it + * is clean, but make sure read_and_act() is given a + * chance to handle 'active_idle' + */ + if (sigterm && !(ret & ARRAY_DIRTY)) + a->container = NULL; /* stop touching this array */ + if (ret & ARRAY_BUSY) + container->retry_soon = 1; } - dirty_arrays += is_dirty; - if (sigterm && !is_dirty) - a->container = NULL; /* stop touching this array */ } /* propagate failures across container members */ for (a = *aap; a ; a = a->next) { - if (!a->container) + if (!a->container || a->to_remove) continue; for (mdi = a->info.devs ; mdi ; mdi = mdi->next) if (mdi->curr_state & DS_FAULTY)