/*
- * mdctl - manage Linux "md" devices aka RAID arrays.
+ * mdadm - manage Linux "md" devices aka RAID arrays.
*
* Copyright (C) 2001-2002 Neil Brown <neilb@cse.unsw.edu.au>
*
* Australia
*/
-#include "mdctl.h"
+#include "mdadm.h"
#include "md_p.h"
#include "md_u.h"
+#include <sys/wait.h>
#include <sys/signal.h>
static void alert(char *event, char *dev, char *disc, char *mailaddr, char *cmd);
+static char *percentalerts[] = {
+ "RebuildStarted",
+ "Rebuild20",
+ "Rebuild40",
+ "Rebuild60",
+ "Rebuild80",
+};
+
int Monitor(mddev_dev_t devlist,
char *mailaddr, char *alert_cmd,
- int period,
+ int period, int daemonise, int scan,
char *config)
{
/*
* Update time
* active/working/failed/spare drives
* State of each device.
+ * %rebuilt if rebuilding
*
* If the update time changes, check out all the data again
* It is possible that we cannot get the state of each device
* due to bugs in the md kernel module.
+ * We also read /proc/mdstat to get rebuild percent,
+ * and to get state on all active devices incase of kernel bug.
*
- * if active_drives decreases, generate a "Fail" event
- * if active_drives increases, generate a "SpareActive" event
+ * Events are:
+ * Fail
+ * An active device had Faulty set or Active/Sync removed
+ * FailSpare
+ * A spare device had Faulty set
+ * SpareActive
+ * An active device had a reverse transition
+ * RebuildStarted
+ * percent went from -1 to +ve
+ * Rebuild20 Rebuild40 Rebuild60 Rebuild80
+ * percent went from below to not-below that number
+ * DeviceDisappeared
+ * Couldn't access a device which was previously visible
*
* if we detect an array with active<raid and spare==0
* we look at other arrays that have same spare-group
* and if we can get_disk_info and find a name
* Then we hot-remove and hot-add to the other array
*
+ * If devlist is NULL, then we can monitor everything because --scan
+ * was given. We get an initial list from config file and add anything
+ * that appears in /proc/mdstat
*/
struct state {
char *devname;
+ int devnum; /* to sync with mdstat info */
long utime;
int err;
- int active, working, failed, spare;
+ char *spare_group;
+ int active, working, failed, spare, raid;
int devstate[MD_SB_DISKS];
+ int devid[MD_SB_DISKS];
+ int percent;
struct state *next;
} *statelist = NULL;
int finished = 0;
- while (! finished) {
- mddev_ident_t mdlist = NULL;
+ struct mdstat_ent *mdstat = NULL;
+
+ if (!mailaddr) {
+ mailaddr = conf_get_mailaddr(config);
+ if (mailaddr && ! scan)
+ fprintf(stderr, Name ": Monitor using email address \"%s\" from config file\n",
+ mailaddr);
+ }
+ if (!alert_cmd) {
+ alert_cmd = conf_get_program(config);
+ if (alert_cmd && ! scan)
+ fprintf(stderr, Name ": 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");
+ return 1;
+ }
+
+ if (daemonise) {
+ int pid = fork();
+ if (pid > 0) {
+ printf("%d\n", pid);
+ return 0;
+ }
+ if (pid < 0) {
+ perror("daemonise");
+ return 1;
+ }
+ close(0);
+ open("/dev/null", 3);
+ dup2(0,1);
+ dup2(0,2);
+ setsid();
+ }
+
+ if (devlist == NULL) {
+ mddev_ident_t mdlist = conf_get_ident(config, NULL);
+ for (; mdlist; mdlist=mdlist->next) {
+ struct state *st = malloc(sizeof *st);
+ if (st == NULL)
+ continue;
+ st->devname = strdup(mdlist->devname);
+ st->utime = 0;
+ st->next = statelist;
+ st->err = 0;
+ st->devnum = -1;
+ st->percent = -2;
+ if (mdlist->spare_group)
+ st->spare_group = strdup(mdlist->spare_group);
+ else
+ st->spare_group = NULL;
+ statelist = st;
+ }
+ } else {
mddev_dev_t dv;
- int dnum=0;
- if (devlist== NULL)
- mdlist = conf_get_ident(config, NULL);
- dv = devlist;
- while (dv || mdlist) {
- mddev_ident_t mdident;
- struct state *st;
+ for (dv=devlist ; dv; dv=dv->next) {
+ struct state *st = malloc(sizeof *st);
+ if (st == NULL)
+ continue;
+ st->devname = strdup(dv->devname);
+ st->utime = 0;
+ st->next = statelist;
+ st->err = 0;
+ st->devnum = -1;
+ st->percent = -2;
+ st->spare_group = NULL;
+ statelist = st;
+ }
+ }
+
+
+ while (! finished) {
+ struct state *st;
+
+ if (mdstat)
+ free_mdstat(mdstat);
+ mdstat = mdstat_read();
+
+ for (st=statelist; st; st=st->next) {
mdu_array_info_t array;
- char *dev;
+ struct mdstat_ent *mse;
+ char *dev = st->devname;
int fd;
- char *event = NULL;
int i;
- char *event_disc = NULL;
- if (dv) {
- dev = dv->devname;
- mdident = conf_get_ident(config, dev);
- dv = dv->next;
- } else {
- mdident = mdlist;
- dev = mdident->devname;
- mdlist = mdlist->next;
- }
- for (st=statelist; st ; st=st->next)
- if (strcmp(st->devname, dev)==0)
- break;
- if (!st) {
- st =malloc(sizeof *st);
- if (st == NULL)
- continue;
- st->devname = strdup(dev);
- st->utime = 0;
- st->next = statelist;
- st->err = 0;
- statelist = st;
- }
+
fd = open(dev, O_RDONLY);
if (fd < 0) {
if (!st->err)
- fprintf(stderr, Name ": cannot open %s: %s\n",
+ alert("DeviceDisappeared", dev, NULL,
+ mailaddr, alert_cmd);
+/* fprintf(stderr, Name ": cannot open %s: %s\n",
dev, strerror(errno));
- st->err=1;
+*/ st->err=1;
continue;
}
if (ioctl(fd, GET_ARRAY_INFO, &array)<0) {
if (!st->err)
- fprintf(stderr, Name ": cannot get array info for %s: %s\n",
+ alert("DeviceDisappeared", dev, NULL,
+ mailaddr, alert_cmd);
+/* fprintf(stderr, Name ": cannot get array info for %s: %s\n",
dev, strerror(errno));
- st->err=1;
+*/ st->err=1;
close(fd);
continue;
}
- st->err = 0;
-
- if (st->utime == array.utime &&
- st->failed == array.failed_disks) {
+ if (array.level != 1 && array.level != 5 && array.level != -4) {
+ if (!st->err)
+ alert("DeviceDisappeared", dev, "Wrong-Level",
+ mailaddr, alert_cmd);
+ st->err = 1;
close(fd);
continue;
}
- event = NULL;
- if (st->utime) {
- int i;
- if (st->active > array.active_disks)
- event = "Fail";
- else if (st->working > array.working_disks)
- event = "FailSpare";
- else if (st->active < array.active_disks)
- event = "ActiveSpare";
+ if (st->devnum < 0) {
+ struct stat stb;
+ if (fstat(fd, &stb) == 0 &&
+ (S_IFMT&stb.st_mode)==S_IFBLK)
+ st->devnum = MINOR(stb.st_rdev);
}
- for (i=0; i<array.raid_disks+array.spare_disks; i++) {
+
+ for (mse = mdstat ; mse ; mse=mse->next)
+ if (mse->devnum == st->devnum) {
+ mse->devnum = -1; /* flag it as "used" */
+ break;
+ }
+
+ 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);
+ st->err = 0;
+ continue;
+ }
+ if (mse &&
+ st->percent == -1 &&
+ mse->percent >= 0)
+ alert("RebuildStarted", dev, NULL, mailaddr, alert_cmd);
+ if (mse &&
+ st->percent >= 0 &&
+ mse->percent >= 0 &&
+ (mse->percent / 20) > (st->percent / 20))
+ alert(percentalerts[mse->percent/20],
+ dev, NULL, mailaddr, alert_cmd);
+
+ if (mse)
+ st->percent = mse->percent;
+
+ for (i=0; i<MD_SB_DISKS; i++) {
mdu_disk_info_t disc;
+ int newstate=0;
+ int change;
+ char *dv = NULL;
disc.number = i;
if (ioctl(fd, GET_DISK_INFO, &disc)>= 0) {
- if (event && event_disc == NULL &&
- st->devstate[i] != disc.state) {
- char * dv = map_dev(disc.major, disc.minor);
- if (dv)
- event_disc = strdup(dv);
+ newstate = disc.state;
+ dv = map_dev(disc.major, disc.minor);
+ } else if (mse && mse->pattern && i < strlen(mse->pattern))
+ switch(mse->pattern[i]) {
+ case 'U': newstate = 6 /* ACTIVE/SYNC */; break;
+ case '_': newstate = 0; break;
}
- st->devstate[i] = disc.state;
+ 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)))
+ )
+ alert("Fail", dev, dv, mailaddr, alert_cmd);
+ else if (i>=array.raid_disks &&
+ (disc.major || disc.minor) &&
+ st->devid[i] == MKDEV(disc.major, disc.minor) &&
+ ((newstate&change)&(1<<MD_DISK_FAULTY))
+ )
+ alert("FailSpare", dev, dv, mailaddr, alert_cmd);
+ else if (i < array.raid_disks &&
+ (((st->devstate[i]&change)&(1<<MD_DISK_FAULTY)) ||
+ ((newstate&change)&(1<<MD_DISK_ACTIVE)) ||
+ ((newstate&change)&(1<<MD_DISK_SYNC)))
+ )
+ alert("SpareActive", dev, dv, mailaddr, alert_cmd);
}
+ st->devstate[i] = disc.state;
+ st->devid[i] = MKDEV(disc.major, disc.minor);
}
close(fd);
st->active = array.active_disks;
st->spare = array.spare_disks;
st->failed = array.failed_disks;
st->utime = array.utime;
- if (event)
- alert(event, dev, event_disc, mailaddr, alert_cmd);
+ st->raid = array.raid_disks;
+ st->err = 0;
+ }
+ /* now check if there are any new devices found in mdstat */
+ if (scan) {
+ struct mdstat_ent *mse;
+ for (mse=mdstat; mse; mse=mse->next)
+ if (mse->devnum >= 0 &&
+ (strcmp(mse->level, "raid1")==0 ||
+ strcmp(mse->level, "raid5")==0 ||
+ strcmp(mse->level, "multipath")==0)
+ ) {
+ struct state *st = malloc(sizeof *st);
+ if (st == NULL)
+ continue;
+ st->devname = strdup(get_md_name(mse->devnum));
+ st->utime = 0;
+ st->next = statelist;
+ st->err = 1;
+ st->devnum = mse->devnum;
+ st->percent = -2;
+ st->spare_group = NULL;
+ statelist = st;
+ alert("NewArray", st->devname, NULL, mailaddr, alert_cmd);
+ }
}
+ /* If an array has active < raid && spare == 0 && spare_group != NULL
+ * Look for another array with spare > 0 and active == raid and same spare_group
+ * if found, choose a device and hotremove/hotadd
+ */
+ 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<MD_SB_DISKS; d++) {
+ if (st2->devid[d] > 0 &&
+ st2->devstate[d] == 0) {
+ dev = st2->devid[d];
+ break;
+ }
+ }
+ if (dev > 0) {
+ if (ioctl(fd2, HOT_REMOVE_DISK,
+ (unsigned long)dev) == 0) {
+ if (ioctl(fd1, HOT_ADD_DISK,
+ (unsigned long)dev) == 0) {
+ alert("MoveSpare", st->devname, st2->devname, mailaddr, alert_cmd);
+ close(fd1);
+ close(fd2);
+ break;
+ }
+ else ioctl(fd2, HOT_ADD_DISK, (unsigned long) dev);
+ }
+ }
+ close(fd1);
+ close(fd2);
+ }
+ }
+
sleep(period);
}
return 0;
if (!cmd && !mailaddr) {
time_t now = time(0);
- printf("%0.15s: %s on %s %s\n", ctime(&now)+4, event, dev, disc?disc:"unknown device");
+ printf("%1.15s: %s on %s %s\n", ctime(&now)+4, event, dev, disc?disc:"unknown device");
}
if (cmd) {
int pid = fork();
fprintf(mp, "A %s event had been detected on md device %s.\n\n", event, dev);
if (disc)
- fprintf(mp, "It could be related to sub-device %s.\n\n", disc);
+ fprintf(mp, "It could be related to component device %s.\n\n", disc);
fprintf(mp, "Faithfully yours, etc.\n");
fclose(mp);
projects / thirdparty / mdadm.git / blobdiff