#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 devnm[32]; /* to sync with mdstat info */
long utime;
int err;
char *spare_group;
int active, working, failed, spare, raid;
+ int from_config;
+ int from_auto;
int expected_spares;
- int devstate[MaxDisks];
- unsigned devid[MaxDisks];
+ int devstate[MAX_DISKS];
+ dev_t devid[MAX_DISKS];
int percent;
- int parent_dev; /* For subarray, devnum of parent.
- * For others, NoMdDev
- */
+ char parent_devnm[32]; /* For subarray, devnm of parent.
+ * For others, ""
+ */
struct supertype *metadata;
struct state *subarray;/* for a container it is a link to first subarray
* for a subarray it is a link to next subarray
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);
-static int add_new_arrays(struct mdstat_ent *mdstat, struct state *statelist,
+ 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);
static void link_containers_with_subarrays(struct state *list);
int Monitor(struct mddev_dev *devlist,
char *mailaddr, char *alert_cmd,
- int period, int daemonise, int scan, int oneshot,
- int dosyslog, int test, char *pidfile, int increments,
+ struct context *c,
+ int daemonise, int oneshot,
+ int dosyslog, char *pidfile, int increments,
int share)
{
/*
*/
struct state *statelist = NULL;
+ struct state *st2;
int finished = 0;
struct mdstat_ent *mdstat = NULL;
char *mailfrom = NULL;
if (!mailaddr) {
mailaddr = conf_get_mailaddr();
- if (mailaddr && ! scan)
- fprintf(stderr, Name ": Monitor using email address \"%s\" from config file\n",
+ if (mailaddr && ! c->scan)
+ pr_err("Monitor using email address \"%s\" from config file\n",
mailaddr);
}
mailfrom = conf_get_mailfrom();
if (!alert_cmd) {
alert_cmd = conf_get_program();
- if (alert_cmd && ! scan)
- fprintf(stderr, Name ": Monitor using program \"%s\" from config file\n",
+ if (alert_cmd && ! c->scan)
+ pr_err("Monitor using program \"%s\" from config file\n",
alert_cmd);
}
- if (scan && !mailaddr && !alert_cmd) {
- fprintf(stderr, Name ": No mail address or alert command - not monitoring.\n");
+ if (c->scan && !mailaddr && !alert_cmd && !dosyslog) {
+ pr_err("No mail address or alert command - not monitoring.\n");
return 1;
}
info.alert_cmd = alert_cmd;
info.mailfrom = mailfrom;
info.dosyslog = dosyslog;
- if (daemonise)
- if (make_daemon(pidfile))
- return 1;
+ if (daemonise) {
+ int rv = make_daemon(pidfile);
+ if (rv >= 0)
+ return rv;
+ }
- if (share)
- if (check_one_sharer(scan))
+ if (share)
+ if (check_one_sharer(c->scan))
return 1;
if (devlist == NULL) {
continue;
if (strcasecmp(mdlist->devname, "<ignore>") == 0)
continue;
- st = calloc(1, sizeof *st);
- if (st == NULL)
- continue;
+ st = xcalloc(1, sizeof *st);
if (mdlist->devname[0] == '/')
- st->devname = strdup(mdlist->devname);
+ st->devname = xstrdup(mdlist->devname);
else {
- st->devname = malloc(8+strlen(mdlist->devname)+1);
+ st->devname = xmalloc(8+strlen(mdlist->devname)+1);
strcpy(strcpy(st->devname, "/dev/md/"),
mdlist->devname);
}
st->next = statelist;
- st->devnum = INT_MAX;
- st->percent = -2;
+ st->devnm[0] = 0;
+ st->percent = RESYNC_UNKNOWN;
+ st->from_config = 1;
st->expected_spares = mdlist->spare_disks;
if (mdlist->spare_group)
- st->spare_group = strdup(mdlist->spare_group);
+ st->spare_group = xstrdup(mdlist->spare_group);
statelist = st;
}
} else {
struct mddev_dev *dv;
for (dv=devlist ; dv; dv=dv->next) {
struct mddev_ident *mdlist = conf_get_ident(dv->devname);
- struct state *st = calloc(1, sizeof *st);
- if (st == NULL)
- continue;
- st->devname = strdup(dv->devname);
+ struct state *st = xcalloc(1, sizeof *st);
+ st->devname = xstrdup(dv->devname);
st->next = statelist;
- st->devnum = INT_MAX;
- st->percent = -2;
+ st->devnm[0] = 0;
+ st->percent = RESYNC_UNKNOWN;
st->expected_spares = -1;
if (mdlist) {
st->expected_spares = mdlist->spare_disks;
if (mdlist->spare_group)
- st->spare_group = strdup(mdlist->spare_group);
+ st->spare_group = xstrdup(mdlist->spare_group);
}
statelist = st;
}
}
-
while (! finished) {
int new_found = 0;
- struct state *st;
+ struct state *st, **stp;
int anydegraded = 0;
if (mdstat)
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, c->test, &info,
+ increments, c->prefer))
anydegraded = 1;
-
+
/* now check if there are any new devices found in mdstat */
- if (scan)
- new_found = add_new_arrays(mdstat, statelist, test,
+ if (c->scan)
+ new_found = add_new_arrays(mdstat, &statelist, c->test,
&info);
/* If an array has active < raid && spare == 0 && spare_group != NULL
if (oneshot)
break;
else
- mdstat_wait(period);
+ mdstat_wait(c->delay);
+ }
+ c->test = 0;
+
+ for (stp = &statelist; (st = *stp) != NULL; ) {
+ if (st->from_auto && st->err > 5) {
+ *stp = st->next;
+ free(st->devname);
+ free(st->spare_group);
+ free(st);
+ } else
+ stp = &st->next;
}
- test = 0;
}
+ for (st2 = statelist; st2; st2 = statelist) {
+ statelist = st2->next;
+ free(st2);
+ }
+
if (pidfile)
unlink(pidfile);
return 0;
static int make_daemon(char *pidfile)
{
+ /* Return:
+ * -1 in the forked daemon
+ * 0 in the parent
+ * 1 on error
+ * so a none-negative becomes the exit code.
+ */
int pid = fork();
if (pid > 0) {
if (!pidfile)
dup2(0,1);
dup2(0,2);
setsid();
- return 0;
+ return -1;
}
static int check_one_sharer(int scan)
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) {
if (scan) {
- fprintf(stderr, Name ": Only one "
- "autorebuild process allowed"
- " in scan mode, aborting\n");
+ pr_err("Only one autorebuild process allowed in scan mode, aborting\n");
fclose(fp);
return 1;
} else {
- fprintf(stderr, Name ": Warning: One"
- " autorebuild process already"
- " running.");
+ pr_err("Warning: One autorebuild process already running.\n");
}
}
fclose(fp);
}
if (scan) {
- fp = fopen("/var/run/mdadm/autorebuild.pid", "w");
- if (!fp)
- fprintf(stderr, Name ": Cannot create"
- " autorebuild.pid "
- "file\n");
- else {
- pid = getpid();
- fprintf(fp, "%d\n", pid);
- fclose(fp);
+ if (mkdir(MDMON_DIR, S_IRWXU) < 0 &&
+ errno != EEXIST) {
+ pr_err("Can't create autorebuild.pid file\n");
+ } else {
+ fp = fopen(path, "w");
+ if (!fp)
+ pr_err("Cannot create autorebuild.pidfile\n");
+ else {
+ pid = getpid();
+ fprintf(fp, "%d\n", pid);
+ fclose(fp);
+ }
}
}
return 0;
{
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");
if (info->mailfrom)
fprintf(mp, "From: %s\n", info->mailfrom);
else
- fprintf(mp, "From: " Name " monitoring <root>\n");
+ fprintf(mp, "From: %s monitoring <root>\n", Name);
fprintf(mp, "To: %s\n", info->mailaddr);
fprintf(mp, "Subject: %s event on %s:%s\n\n",
event, dev, hname);
fprintf(mp,
- "This is an automatically generated"
- " mail message from " Name "\n");
+ "This is an automatically generated mail message from %s\n", Name);
fprintf(mp, "running on %s\n\n", hname);
fprintf(mp,
- "A %s event had been detected on"
- " md device %s.\n\n", event, dev);
+ "A %s event had been detected on md device %s.\n\n", event, dev);
if (disc && disc[0] != ' ')
fprintf(mp,
- "It could be related to"
- " component device %s.\n\n", disc);
+ "It could be related to component device %s.\n\n", disc);
if (disc && disc[0] == ' ')
fprintf(mp, "Extra information:%s.\n\n", disc);
char buf[8192];
int n;
fprintf(mp,
- "\nP.S. The /proc/mdstat file"
- " currently contains the following:\n\n");
+ "\nP.S. The /proc/mdstat file currently contains the following:\n\n");
while ( (n=fread(buf, 1, sizeof(buf), mdstat)) > 0)
n=fwrite(buf, 1, n, mp);
fclose(mdstat);
else
priority = LOG_INFO;
- if (disc)
+ if (disc && disc[0] != ' ')
+ syslog(priority,
+ "%s event detected on md device %s, component device %s", event, dev, disc);
+ else if (disc)
syslog(priority,
- "%s event detected on md device %s,"
- " component device %s", event, dev, disc);
+ "%s event detected on md device %s: %s",
+ event, dev, disc);
else
syslog(priority,
"%s event detected on md device %s",
static int check_array(struct state *st, struct mdstat_ent *mdstat,
int test, struct alert_info *ainfo,
- int increments)
+ int increments, char *prefer)
{
- struct { int state, major, minor; } info[MaxDisks];
+ /* 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;
+ 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=1;
+ 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=1;
+ st->err++;
close(fd);
return 0;
}
* have a device disappear than all of them that can
*/
if (array.level == 0 || array.level == -1) {
- if (!st->err)
- alert("DeviceDisappeared", dev, "Wrong-Level", ainfo);
- st->err = 1;
+ if (!st->err && !st->from_config)
+ alert("DeviceDisappeared", dev, " Wrong-Level", ainfo);
+ st->err++;
close(fd);
return 0;
}
- if (st->devnum == INT_MAX) {
- struct stat stb;
- if (fstat(fd, &stb) == 0 &&
- (S_IFMT&stb.st_mode)==S_IFBLK) {
- if (major(stb.st_rdev) == MD_MAJOR)
- st->devnum = minor(stb.st_rdev);
- else
- st->devnum = -1- (minor(stb.st_rdev)>>6);
- }
- }
+ if (st->devnm[0] == 0)
+ strcpy(st->devnm, fd2devnm(fd));
for (mse2 = mdstat ; mse2 ; mse2=mse2->next)
- if (mse2->devnum == st->devnum) {
- mse2->devnum = INT_MAX; /* flag it as "used" */
+ 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 = 1;
+ st->err++;
close(fd);
return 0;
}
* 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 &&
mse->percent == st->percent
))) {
close(fd);
- st->err = 0;
- return 0;
+ 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 */
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,
* we should report that.
*/
struct mdinfo *sra =
- sysfs_read(-1, st->devnum, GET_MISMATCH);
+ sysfs_read(-1, st->devnm, 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 &&
- is_subarray(mse->metadata_version+9))
- st->parent_dev =
- devname2devnum(mse->metadata_version+10);
- else
- st->parent_dev = NoMdDev;
+ 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_dev == NoMdDev)
+ st->parent_devnm[0] == 0)
st->metadata = super_by_fd(fd, NULL);
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) {
- newstate = 0;
- disc.major = disc.minor = 0;
- } else if (info[i].major || info[i].minor) {
+ if (i < last_disk &&
+ (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;
- }
- disc.major = disc.minor = 0;
- }
+ } else
+ newstate = (1 << MD_DISK_REMOVED);
+
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 &&
- (((newstate&change)&(1<<MD_DISK_FAULTY)) ||
- ((st->devstate[i]&change)&(1<<MD_DISK_ACTIVE)) ||
- ((st->devstate[i]&change)&(1<<MD_DISK_SYNC)))
- )
+ if (st->utime && change && !st->err && !new_array) {
+ if ((st->devstate[i]&change)&(1<<MD_DISK_SYNC))
alert("Fail", dev, dv, ainfo);
- else if (i >= array.raid_disks &&
+ else if ((newstate & (1<<MD_DISK_FAULTY)) &&
(disc.major || disc.minor) &&
- st->devid[i] == makedev(disc.major, disc.minor) &&
- ((newstate&change)&(1<<MD_DISK_FAULTY))
- )
+ st->devid[i] == makedev(disc.major, disc.minor))
alert("FailSpare", dev, dv, ainfo);
- else if (i < array.raid_disks &&
- ! (newstate & (1<<MD_DISK_REMOVED)) &&
- (((st->devstate[i]&change)&(1<<MD_DISK_FAULTY)) ||
- ((newstate&change)&(1<<MD_DISK_ACTIVE)) ||
- ((newstate&change)&(1<<MD_DISK_SYNC)))
- )
+ else if ((newstate&change)&(1<<MD_DISK_SYNC))
alert("SpareActive", dev, dv, ainfo);
}
st->devstate[i] = newstate;
return 0;
}
-static int add_new_arrays(struct mdstat_ent *mdstat, struct state *statelist,
+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->devnum != INT_MAX &&
+ if (mse->devnm[0] &&
(!mse->level || /* retrieve containers */
(strcmp(mse->level, "raid0") != 0 &&
strcmp(mse->level, "linear") != 0))
) {
- struct state *st = calloc(1, sizeof *st);
+ struct state *st = xcalloc(1, sizeof *st);
mdu_array_info_t array;
int fd;
- if (st == NULL)
+
+ name = get_md_name(mse->devnm);
+ if (!name) {
+ free(st);
continue;
- st->devname = strdup(get_md_name(mse->devnum));
+ }
+
+ st->devname = xstrdup(name);
if ((fd = open(st->devname, O_RDONLY)) < 0 ||
ioctl(fd, GET_ARRAY_INFO, &array)< 0) {
/* no such array */
continue;
}
close(fd);
- st->next = statelist;
+ st->next = *statelist;
st->err = 1;
- st->devnum = mse->devnum;
- st->percent = -2;
+ st->from_auto = 1;
+ strcpy(st->devnm, mse->devnm);
+ st->percent = RESYNC_UNKNOWN;
st->expected_spares = -1;
- if (strncmp(mse->metadata_version, "external:", 9) == 0 &&
- is_subarray(mse->metadata_version+9))
- st->parent_dev =
- devname2devnum(mse->metadata_version+10);
- else
- st->parent_dev = NoMdDev;
- statelist = st;
+ 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);
- alert("NewArray", st->devname, NULL, info);
new_found = 1;
}
return new_found;
}
+static int get_min_spare_size_required(struct state *st, unsigned long long *sizep)
+{
+ int fd;
+
+ if (!st->metadata ||
+ !st->metadata->ss->min_acceptable_spare_size) {
+ *sizep = 0;
+ return 0;
+ }
+
+ fd = open(st->devname, O_RDONLY);
+ if (fd < 0)
+ 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);
+ if (!st->metadata->sb)
+ return 1;
+ *sizep = st->metadata->ss->min_acceptable_spare_size(st->metadata);
+ st->metadata->ss->free_super(st->metadata);
+
+ return 0;
+}
+
+static int check_donor(struct state *from, struct state *to)
+{
+ struct state *sub;
+
+ if (from == to)
+ return 0;
+ if (from->parent)
+ /* Cannot move from a member */
+ return 0;
+ if (from->err)
+ return 0;
+ for (sub = from->subarray; sub; sub = sub->subarray)
+ /* If source array has degraded subarrays, don't
+ * remove anything
+ */
+ if (sub->active < sub->raid)
+ return 0;
+ if (from->metadata->ss->external == 0)
+ if (from->active < from->raid)
+ return 0;
+ if (from->spare <= 0)
+ return 0;
+ return 1;
+}
+
+static dev_t choose_spare(struct state *from, struct state *to,
+ struct domainlist *domlist, unsigned long long min_size)
+{
+ int d;
+ dev_t dev = 0;
+
+ 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)
+ continue;
+
+ pol = devid_policy(from->devid[d]);
+ if (from->spare_group)
+ pol_add(&pol, pol_domain,
+ from->spare_group, NULL);
+ if (domain_test(domlist, pol, to->metadata->ss->name) == 1)
+ dev = from->devid[d];
+ dev_policy_free(pol);
+ }
+ }
+ return dev;
+}
+
+static dev_t container_choose_spare(struct state *from, struct state *to,
+ struct domainlist *domlist,
+ 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
+ */
+ struct mdinfo *list;
+ struct supertype *st = from->metadata;
+ int fd = open(from->devname, O_RDONLY);
+ int err;
+ dev_t dev = 0;
+
+ if (fd < 0)
+ return 0;
+ if (!st->ss->getinfo_super_disks) {
+ close(fd);
+ return 0;
+ }
+
+ err = st->ss->load_container(st, fd, NULL);
+ close(fd);
+ 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);
+ if (list) {
+ struct mdinfo *disks = list->devs;
+ if (disks)
+ dev = makedev(disks->disk.major, disks->disk.minor);
+ sysfs_free(list);
+ }
+ st->ss->free_super(st);
+ return dev;
+}
+
static void try_spare_migration(struct state *statelist, struct alert_info *info)
{
+ struct state *from;
struct state *st;
link_containers_with_subarrays(statelist);
- for (st = statelist; st; st=st->next)
+ for (st = statelist; st; st = st->next)
if (st->active < st->raid &&
- st->spare == 0 &&
- st->spare_group != NULL) {
- struct state *st2;
- for (st2=statelist ; st2 ; st2=st2->next)
- if (st2 != st &&
- st2->spare > 0 &&
- st2->active == st2->raid &&
- st2->spare_group != NULL &&
- strcmp(st->spare_group, st2->spare_group) == 0) {
- /* try to remove and add */
- int fd1 = open(st->devname, O_RDONLY);
- int fd2 = open(st2->devname, O_RDONLY);
- int dev = -1;
- int d;
- if (fd1 < 0 || fd2 < 0) {
- if (fd1>=0) close(fd1);
- if (fd2>=0) close(fd2);
- continue;
- }
- for (d=st2->raid; d < MaxDisks; d++) {
- if (st2->devid[d] > 0 &&
- st2->devstate[d] == 0) {
- dev = st2->devid[d];
- break;
- }
- }
- if (dev > 0) {
- struct mddev_dev devlist;
- char devname[20];
- devlist.next = NULL;
- devlist.used = 0;
- devlist.re_add = 0;
- devlist.writemostly = 0;
- devlist.devname = devname;
- sprintf(devname, "%d:%d", major(dev), minor(dev));
-
- devlist.disposition = 'r';
- if (Manage_subdevs(st2->devname, fd2, &devlist, -1, 0) == 0) {
- devlist.disposition = 'a';
- if (Manage_subdevs(st->devname, fd1, &devlist, -1, 0) == 0) {
- alert("MoveSpare", st->devname, st2->devname, info);
- close(fd1);
- close(fd2);
- break;
- }
- else Manage_subdevs(st2->devname, fd2, &devlist, -1, 0);
- }
- }
- close(fd1);
- close(fd2);
+ st->spare == 0 && !st->err) {
+ struct domainlist *domlist = NULL;
+ int d;
+ struct state *to = st;
+ unsigned long long min_size;
+
+ if (to->parent_devnm[0] && !to->parent)
+ /* subarray monitored without parent container
+ * we can't move spares here */
+ continue;
+
+ if (to->parent)
+ /* member of a container */
+ to = to->parent;
+
+ 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, st->active);
+ if (devid > 0)
+ continue;
+ }
+ for (d = 0; d < MAX_DISKS; d++)
+ if (to->devid[d])
+ domainlist_add_dev(&domlist,
+ to->devid[d],
+ to->metadata->ss->name);
+ if (to->spare_group)
+ domain_add(&domlist, to->spare_group);
+ /*
+ * No spare migration if the destination
+ * has no domain. Skip this array.
+ */
+ if (!domlist)
+ continue;
+ for (from=statelist ; from ; from=from->next) {
+ dev_t devid;
+ if (!check_donor(from, to))
+ continue;
+ if (from->metadata->ss->external)
+ devid = container_choose_spare(
+ from, to, domlist, min_size, 0);
+ else
+ devid = choose_spare(from, to, domlist,
+ min_size);
+ if (devid > 0
+ && move_spare(from->devname, to->devname, devid)) {
+ alert("MoveSpare", to->devname, from->devname, info);
+ break;
}
+ }
+ domain_free(domlist);
}
}
st->subarray = NULL;
}
for (st = list; st; st = st->next)
- if (st->parent_dev != NoMdDev)
+ if (st->parent_devnm[0])
for (cont = list; cont; cont = cont->next)
if (!cont->err &&
- cont->parent_dev == NoMdDev &&
- cont->devnum == st->parent_dev) {
+ cont->parent_devnm[0] == 0 &&
+ strcmp(cont->devnm, st->parent_devnm) == 0) {
st->parent = cont;
st->subarray = cont->subarray;
cont->subarray = st;
int Wait(char *dev)
{
struct stat stb;
- int devnum;
+ char devnm[32];
int rv = 1;
+ int frozen_remaining = 3;
if (stat(dev, &stb) != 0) {
- fprintf(stderr, Name ": Cannot find %s: %s\n", dev,
+ pr_err("Cannot find %s: %s\n", dev,
strerror(errno));
return 2;
}
- devnum = stat2devnum(&stb);
+ 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 (e->devnum == devnum)
+ if (strcmp(e->devnm, devnm) == 0)
break;
- if (!e || e->percent < 0) {
+ 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(devnum2devname(devnum));
+ ping_monitor(devnm);
}
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;
+ 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;
+}
+#endif /* MDASSEMBLE */