#include <limits.h>
#include <syslog.h>
-/* The largest number of disks current arrays can manage is 384
- * This really should be dynamically, but that will have to wait
- * At least it isn't MD_SB_DISKS.
- */
-#define MaxDisks 384
struct state {
char *devname;
int devnum; /* to sync with mdstat info */
char *spare_group;
int active, working, failed, spare, raid;
int expected_spares;
- int devstate[MaxDisks];
- dev_t devid[MaxDisks];
+ int devstate[MAX_DISKS];
+ dev_t devid[MAX_DISKS];
int percent;
int parent_dev; /* For subarray, devnum of parent.
* For others, NoMdDev
static void alert(char *event, char *dev, char *disc, struct alert_info *info);
static int check_array(struct state *st, struct mdstat_ent *mdstat,
int test, struct alert_info *info,
- int increments);
+ int increments, char *prefer);
static int add_new_arrays(struct mdstat_ent *mdstat, struct state **statelist,
int test, struct alert_info *info);
static void try_spare_migration(struct state *statelist, struct alert_info *info);
char *mailaddr, char *alert_cmd,
int period, int daemonise, int scan, int oneshot,
int dosyslog, int test, char *pidfile, int increments,
- int share)
+ int share, char *prefer)
{
/*
* Every few seconds, scan every md device looking for changes
*/
struct state *statelist = NULL;
+ struct state *st2;
int finished = 0;
struct mdstat_ent *mdstat = NULL;
char *mailfrom = NULL;
fprintf(stderr, Name ": Monitor using program \"%s\" from config file\n",
alert_cmd);
}
- if (scan && !mailaddr && !alert_cmd) {
+ if (scan && !mailaddr && !alert_cmd && !dosyslog) {
fprintf(stderr, Name ": No mail address or alert command - not monitoring.\n");
return 1;
}
}
st->next = statelist;
st->devnum = INT_MAX;
- st->percent = -2;
+ st->percent = RESYNC_UNKNOWN;
st->expected_spares = mdlist->spare_disks;
if (mdlist->spare_group)
st->spare_group = strdup(mdlist->spare_group);
st->devname = strdup(dv->devname);
st->next = statelist;
st->devnum = INT_MAX;
- st->percent = -2;
+ st->percent = RESYNC_UNKNOWN;
st->expected_spares = -1;
if (mdlist) {
st->expected_spares = mdlist->spare_disks;
mdstat = mdstat_read(oneshot?0:1, 0);
for (st=statelist; st; st=st->next)
- if (check_array(st, mdstat, test, &info, increments))
+ if (check_array(st, mdstat, test, &info,
+ increments, prefer))
anydegraded = 1;
/* now check if there are any new devices found in mdstat */
}
test = 0;
}
+ for (st2 = statelist; st2; st2 = statelist) {
+ statelist = st2->next;
+ free(st2);
+ }
+
if (pidfile)
unlink(pidfile);
return 0;
int pid, rv;
FILE *fp;
char dir[20];
+ char path[100];
struct stat buf;
- fp = fopen("/var/run/mdadm/autorebuild.pid", "r");
+ sprintf(path, "%s/autorebuild.pid", MDMON_DIR);
+ fp = fopen(path, "r");
if (fp) {
- fscanf(fp, "%d", &pid);
+ if (fscanf(fp, "%d", &pid) != 1)
+ pid = -1;
sprintf(dir, "/proc/%d", pid);
rv = stat(dir, &buf);
if (rv != -1) {
fclose(fp);
}
if (scan) {
- if (mkdir("/var/run/mdadm", S_IRWXU) < 0 &&
+ if (mkdir(MDMON_DIR, S_IRWXU) < 0 &&
errno != EEXIST) {
fprintf(stderr, Name ": Can't create "
"autorebuild.pid file\n");
} else {
- fp = fopen("/var/run/mdadm/autorebuild.pid", "w");
+ fp = fopen(path, "w");
if (!fp)
fprintf(stderr, Name ": Cannot create"
" autorebuild.pid"
{
int priority;
- if (!info->alert_cmd && !info->mailaddr) {
+ if (!info->alert_cmd && !info->mailaddr && !info->dosyslog) {
time_t now = time(0);
printf("%1.15s: %s on %s %s\n", ctime(&now)+4, event, dev, disc?disc:"unknown device");
static int check_array(struct state *st, struct mdstat_ent *mdstat,
int test, struct alert_info *ainfo,
- int increments)
+ 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[MaxDisks];
+ 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;
int i;
+ int remaining_disks;
+ int last_disk;
if (test)
alert("TestMessage", dev, NULL, ainfo);
st->expected_spares > 0 &&
array.spare_disks < st->expected_spares)
alert("SparesMissing", dev, NULL, ainfo);
- if (st->percent == -1 &&
+ if (st->percent < 0 && st->percent != RESYNC_UNKNOWN &&
mse->percent >= 0)
alert("RebuildStarted", dev, NULL, ainfo);
if (st->percent >= 0 &&
alert(percentalert, dev, NULL, ainfo);
}
- if (mse->percent == -1 &&
+ if (mse->percent == RESYNC_NONE &&
st->percent >= 0) {
/* Rebuild/sync/whatever just finished.
* If there is a number in /mismatch_cnt,
struct mdinfo *sra =
sysfs_read(-1, st->devnum, GET_MISMATCH);
if (sra && sra->mismatch_cnt > 0) {
- char cnt[40];
- sprintf(cnt, " mismatches found: %d", sra->mismatch_cnt);
+ 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);
}
st->percent = mse->percent;
- for (i=0; i<MaxDisks && i <= array.raid_disks + array.nr_disks;
+ remaining_disks = array.nr_disks;
+ for (i=0; i<MAX_DISKS && remaining_disks > 0;
i++) {
mdu_disk_info_t disc;
disc.number = i;
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 (strncmp(mse->metadata_version, "external:", 9) == 0 &&
+ if (mse->metadata_version &&
+ strncmp(mse->metadata_version, "external:", 9) == 0 &&
is_subarray(mse->metadata_version+9))
st->parent_dev =
devname2devnum(mse->metadata_version+10);
close(fd);
- for (i=0; i<MaxDisks; i++) {
+ 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 > array.raid_disks + array.nr_disks) {
+ if (i >= last_disk) {
newstate = 0;
disc.major = disc.minor = 0;
} else if (info[i].major || info[i].minor) {
newstate = info[i].state;
- dv = map_dev(info[i].major, info[i].minor, 1);
+ 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 if (mse && mse->pattern && i < (int)strlen(mse->pattern)) {
switch(mse->pattern[i]) {
case 'U': newstate = 6 /* ACTIVE/SYNC */; break;
- case '_': newstate = 0; break;
+ case '_': newstate = 8 /* REMOVED */; break;
}
disc.major = disc.minor = 0;
}
if (dv == NULL && st->devid[i])
- dv = map_dev(major(st->devid[i]),
- minor(st->devid[i]), 1);
+ 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) {
if (i < array.raid_disks &&
st->next = *statelist;
st->err = 1;
st->devnum = mse->devnum;
- st->percent = -2;
+ st->percent = RESYNC_UNKNOWN;
st->expected_spares = -1;
- if (strncmp(mse->metadata_version, "external:", 9) == 0 &&
+ if (mse->metadata_version &&
+ strncmp(mse->metadata_version, "external:", 9) == 0 &&
is_subarray(mse->metadata_version+9))
st->parent_dev =
devname2devnum(mse->metadata_version+10);
return new_found;
}
-unsigned long long min_spare_size_required(struct state *st)
+static int get_min_spare_size_required(struct state *st, unsigned long long *sizep)
{
int fd;
- unsigned long long rv = 0;
if (!st->metadata ||
- !st->metadata->ss->min_acceptable_spare_size)
- return rv;
+ !st->metadata->ss->min_acceptable_spare_size) {
+ *sizep = 0;
+ return 0;
+ }
fd = open(st->devname, O_RDONLY);
if (fd < 0)
- return 0;
- st->metadata->ss->load_super(st->metadata, fd, st->devname);
+ return 1;
+ if (st->metadata->ss->external)
+ st->metadata->ss->load_container(st->metadata, fd, st->devname);
+ else
+ st->metadata->ss->load_super(st->metadata, fd, st->devname);
close(fd);
- rv = st->metadata->ss->min_acceptable_spare_size(st->metadata);
+ if (!st->metadata->sb)
+ return 1;
+ *sizep = st->metadata->ss->min_acceptable_spare_size(st->metadata);
st->metadata->ss->free_super(st->metadata);
- return rv;
-}
-
-static int move_spare(struct state *from, struct state *to,
- dev_t devid,
- struct alert_info *info)
-{
- struct mddev_dev devlist;
- char devname[20];
-
- /* try to remove and add */
- int fd1 = open(to->devname, O_RDONLY);
- int fd2 = open(from->devname, O_RDONLY);
-
- if (fd1 < 0 || fd2 < 0) {
- if (fd1>=0) close(fd1);
- if (fd2>=0) close(fd2);
- return 0;
- }
-
- devlist.next = NULL;
- devlist.used = 0;
- devlist.re_add = 0;
- devlist.writemostly = 0;
- devlist.devname = devname;
- sprintf(devname, "%d:%d", major(devid), minor(devid));
-
- devlist.disposition = 'r';
- if (Manage_subdevs(from->devname, fd2, &devlist, -1, 0, NULL) == 0) {
- devlist.disposition = 'a';
- if (Manage_subdevs(to->devname, fd1, &devlist, -1, 0, NULL) == 0) {
- alert("MoveSpare", to->devname, from->devname, info);
- /* make sure we will see newly added spare before next
- * time through loop
- */
- ping_manager(to->devname);
- ping_manager(from->devname);
- close(fd1);
- close(fd2);
- return 1;
- }
- else Manage_subdevs(from->devname, fd2, &devlist, -1, 0, NULL);
- }
- close(fd1);
- close(fd2);
return 0;
}
int d;
dev_t dev = 0;
- for (d = from->raid; !dev && d < MaxDisks; d++) {
+ for (d = from->raid; !dev && d < MAX_DISKS; d++) {
if (from->devid[d] > 0 &&
from->devstate[d] == 0) {
struct dev_policy *pol;
unsigned long long dev_size;
+ if (to->metadata->ss->external &&
+ test_partition_from_id(from->devid[d]))
+ continue;
+
if (min_size &&
dev_size_from_id(from->devid[d], &dev_size) &&
dev_size < min_size)
if (from->spare_group)
pol_add(&pol, pol_domain,
from->spare_group, NULL);
- if (domain_test(domlist, pol, to->metadata->ss->name))
+ if (domain_test(domlist, pol, to->metadata->ss->name) == 1)
dev = from->devid[d];
dev_policy_free(pol);
}
static dev_t container_choose_spare(struct state *from, struct state *to,
struct domainlist *domlist,
- unsigned long long min_size)
+ unsigned long long min_size, int active)
{
/* This is similar to choose_spare, but we cannot trust devstate,
* so we need to read the metadata instead
if (err)
return 0;
+ if (from == to) {
+ /* We must check if number of active disks has not increased
+ * since ioctl in main loop. mdmon may have added spare
+ * to subarray. If so we do not need to look for more spares
+ * so return non zero value */
+ int active_cnt = 0;
+ struct mdinfo *dp;
+ list = st->ss->getinfo_super_disks(st);
+ if (!list) {
+ st->ss->free_super(st);
+ return 1;
+ }
+ dp = list->devs;
+ while (dp) {
+ if (dp->disk.state & (1<<MD_DISK_SYNC) &&
+ !(dp->disk.state & (1<<MD_DISK_FAULTY)))
+ active_cnt++;
+ dp = dp->next;
+ }
+ sysfs_free(list);
+ if (active < active_cnt) {
+ /* Spare just activated.*/
+ st->ss->free_super(st);
+ return 1;
+ }
+ }
+
/* We only need one spare so full list not needed */
list = container_choose_spares(st, min_size, domlist, from->spare_group,
to->metadata->ss->name, 1);
struct state *to = st;
unsigned long long min_size;
+ if (to->parent_dev != NoMdDev && !to->parent)
+ /* subarray monitored without parent container
+ * we can't move spares here */
+ continue;
+
if (to->parent)
/* member of a container */
to = to->parent;
- min_size = min_spare_size_required(to);
+ if (get_min_spare_size_required(to, &min_size))
+ continue;
if (to->metadata->ss->external) {
/* We must make sure there is
* no suitable spare in container already.
* If there is we don't add more */
dev_t devid = container_choose_spare(
- to, to, NULL, min_size);
+ to, to, NULL, min_size, st->active);
if (devid > 0)
continue;
}
- for (d = 0; d < MaxDisks; d++)
+ for (d = 0; d < MAX_DISKS; d++)
if (to->devid[d])
domainlist_add_dev(&domlist,
to->devid[d],
continue;
if (from->metadata->ss->external)
devid = container_choose_spare(
- from, to, domlist, min_size);
+ from, to, domlist, min_size, 0);
else
devid = choose_spare(from, to, domlist,
min_size);
if (devid > 0
- && move_spare(from, to, devid, info))
- break;
+ && move_spare(from->devname, to->devname, devid)) {
+ alert("MoveSpare", to->devname, from->devname, info);
+ break;
+ }
}
domain_free(domlist);
}
if (is_subarray(&e->metadata_version[9]))
ping_monitor(&e->metadata_version[9]);
else
- ping_monitor(devnum2devname(devnum));
+ ping_monitor_by_id(devnum);
}
free_mdstat(ms);
return rv;
mdstat_wait(5);
}
}
+
+#ifndef MDASSEMBLE
+
+static char *clean_states[] = {
+ "clear", "inactive", "readonly", "read-auto", "clean", NULL };
+
+int WaitClean(char *dev, int sock, int verbose)
+{
+ int fd;
+ struct mdinfo *mdi;
+ int rv = 1;
+ int devnum;
+
+ fd = open(dev, O_RDONLY);
+ if (fd < 0) {
+ if (verbose)
+ fprintf(stderr, Name ": Couldn't open %s: %s\n", dev, strerror(errno));
+ return 1;
+ }
+
+ devnum = fd2devnum(fd);
+ mdi = sysfs_read(fd, devnum, GET_VERSION|GET_LEVEL|GET_SAFEMODE);
+ if (!mdi) {
+ if (verbose)
+ fprintf(stderr, Name ": 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(fd2devnum(fd), NULL, "array_state");
+ char buf[20];
+ fd_set fds;
+ struct timeval tm;
+
+ /* minimize the safe_mode_delay and prepare to wait up to 5s
+ * for writes to quiesce
+ */
+ sysfs_set_safemode(mdi, 1);
+ tm.tv_sec = 5;
+ tm.tv_usec = 0;
+
+ FD_ZERO(&fds);
+
+ /* 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;
+ FD_SET(state_fd, &fds);
+ rv = select(state_fd + 1, NULL, NULL, &fds, &tm);
+ 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)
+ fprintf(stderr, Name ": 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;
+}
+#endif /* MDASSEMBLE */