b=22078 Bind ost/mdt to specific nids

The goal of the patch attached to this bug is to make NUMIOA factor management
easier, by giving the ability to restrict the NIDs that a target (mdt or ost)
registers on the MGS.

i=nathan
i=sebastien

diff --git a/lustre/doc/mkfs.lustre.8 b/lustre/doc/mkfs.lustre.8
index 32a4098..ea32c54 100644 (file)
--- a/lustre/doc/mkfs.lustre.8
+++ b/lustre/doc/mkfs.lustre.8
@@ -68,6 +68,9 @@ OST: \fIerrors=remount-ro,mballoc,extents\fR;
 MGS/MDT: \fIerrors=remount-ro,iopen_nopriv,user_xattr\fR.
 \fBDO NOT\fR alter the default mount options unless you know what you are doing.
 .TP
 MGS/MDT: \fIerrors=remount-ro,iopen_nopriv,user_xattr\fR.
 \fBDO NOT\fR alter the default mount options unless you know what you are doing.
 .TP

+.BI \--network= net,...
+Network(s) to restrict this ost/mdt to. This option can be repeated as desired.
+.TP

 .BI \--mgsnode= nid,...  
 Set the NID(s) of the MGS node, required for all targets other than the MGS.
 .TP
 .BI \--mgsnode= nid,...  
 Set the NID(s) of the MGS node, required for all targets other than the MGS.
 .TP

diff --git a/lustre/doc/tunefs.lustre.8 b/lustre/doc/tunefs.lustre.8
index ec4db1a..16b40a8 100644 (file)
--- a/lustre/doc/tunefs.lustre.8
+++ b/lustre/doc/tunefs.lustre.8
@@ -52,6 +52,9 @@ OST: \fIerrors=remount-ro,mballoc,extents\fR;
 MGS/MDT: \fIerrors=remount-ro,iopen_nopriv,user_xattr\fR.
 \fBDO NOT\fR alter the default mount options unless you know what you are doing.
 .TP
 MGS/MDT: \fIerrors=remount-ro,iopen_nopriv,user_xattr\fR.
 \fBDO NOT\fR alter the default mount options unless you know what you are doing.
 .TP

+.BI \--network= net,...
+Network(s) to restrict this ost/mdt to. This option can be repeated as desired.
+.TP

 .BI \--mgs
 Add a configuration management service to this target
 .TP
 .BI \--mgs
 Add a configuration management service to this target
 .TP

diff --git a/lustre/include/lustre_param.h b/lustre/include/lustre_param.h
index 883eae3..cfc510c 100644 (file)
--- a/lustre/include/lustre_param.h
+++ b/lustre/include/lustre_param.h
@@ -53,6 +53,8 @@ int class_find_param(char *buf, char *key, char **valp);
 int class_get_next_param(char **params, char *copy);
 int class_match_param(char *buf, char *key, char **valp);
 int class_parse_nid(char *buf, lnet_nid_t *nid, char **endh);
 int class_get_next_param(char **params, char *copy);
 int class_match_param(char *buf, char *key, char **valp);
 int class_parse_nid(char *buf, lnet_nid_t *nid, char **endh);

+int class_parse_net(char *buf, __u32 *net, char **endh);
+int class_match_net(char *buf, lnet_nid_t nid);

 /* obd_mount.c */
 int do_lcfg(char *cfgname, lnet_nid_t nid, int cmd,
             char *s1, char *s2, char *s3, char *s4);
 /* obd_mount.c */
 int do_lcfg(char *cfgname, lnet_nid_t nid, int cmd,
             char *s1, char *s2, char *s3, char *s4);


@@ -83,6 +85,7 @@ int do_lcfg(char *cfgname, lnet_nid_t nid, int cmd,
 #define PARAM_FAILNODE             "failover.node="    /* add failover nid */
 #define PARAM_FAILMODE             "failover.mode="    /* initial mount only */
 #define PARAM_ACTIVE               "active="           /* activate/deactivate */
 #define PARAM_FAILNODE             "failover.node="    /* add failover nid */
 #define PARAM_FAILMODE             "failover.mode="    /* initial mount only */
 #define PARAM_ACTIVE               "active="           /* activate/deactivate */

+#define PARAM_NETWORK              "network="          /* bind on nid */

 
 /* Prefixes for parameters handled by obd's proc methods (XXX_process_config) */
 #define PARAM_OST                  "ost."
 
 /* Prefixes for parameters handled by obd's proc methods (XXX_process_config) */
 #define PARAM_OST                  "ost."

diff --git a/lustre/obdclass/obd_config.c b/lustre/obdclass/obd_config.c
index 1fdfb12..96c853a 100644 (file)
--- a/lustre/obdclass/obd_config.c
+++ b/lustre/obdclass/obd_config.c
@@ -157,13 +157,41 @@ int class_match_param(char *buf, char *key, char **valp)
         return 0;
 }
 
         return 0;
 }
 

+static int parse_nid(char *buf, void *value)
+{
+        lnet_nid_t *nid = (lnet_nid_t *)value;
+
+        *nid = libcfs_str2nid(buf);
+        if (*nid != LNET_NID_ANY)
+                return 0;
+
+        LCONSOLE_ERROR_MSG(0x159, "Can't parse NID '%s'\n", buf);
+        return -EINVAL;
+}
+
+static int parse_net(char *buf, void *value)
+{
+        __u32 *net = (__u32 *)net;
+
+        *net = libcfs_str2net(buf);
+        CDEBUG(D_INFO, "Net %s\n", libcfs_net2str(*net));
+        return 0;
+}
+
+enum {
+        CLASS_PARSE_NID = 1,
+        CLASS_PARSE_NET,
+};
+

 /* 0 is good nid,
    1 not found
    < 0 error
    endh is set to next separator */
 /* 0 is good nid,
    1 not found
    < 0 error
    endh is set to next separator */

-int class_parse_nid(char *buf, lnet_nid_t *nid, char **endh)
+static int class_parse_value(char *buf, int opc, void *value, char **endh)

 {
 {

-        char tmp, *endp;
+        char *endp;
+        char  tmp;
+        int   rc = 0;

 
         if (!buf)
                 return 1;
 
         if (!buf)
                 return 1;


@@ -179,17 +207,46 @@ int class_parse_nid(char *buf, lnet_nid_t *nid, char **endh)
 
         tmp = *endp;
         *endp = '\0';
 
         tmp = *endp;
         *endp = '\0';

-        *nid = libcfs_str2nid(buf);
-        if (*nid == LNET_NID_ANY) {
-                LCONSOLE_ERROR_MSG(0x159, "Can't parse NID '%s'\n", buf);
-                *endp = tmp;
-                return -EINVAL;
+        switch (opc) {
+        default:
+                LBUG();
+        case CLASS_PARSE_NID:
+                rc = parse_nid(buf, value);
+                break;
+        case CLASS_PARSE_NET:
+                rc = parse_net(buf, value);
+                break;

         }
         *endp = tmp;
         }
         *endp = tmp;

-
+        if (rc != 0)
+                return rc;

         if (endh)
                 *endh = endp;
         if (endh)
                 *endh = endp;

-        CDEBUG(D_INFO, "Nid %s\n", libcfs_nid2str(*nid));
+        return 0;
+}
+
+int class_parse_nid(char *buf, lnet_nid_t *nid, char **endh)
+{
+        return class_parse_value(buf, CLASS_PARSE_NID, (void *)nid, endh);
+}
+
+int class_parse_net(char *buf, __u32 *net, char **endh)
+{
+        return class_parse_value(buf, CLASS_PARSE_NET, (void *)net, endh);
+}
+
+int class_match_net(char *buf, lnet_nid_t nid)
+{
+        __u32 net;
+
+        while (class_find_param(buf, PARAM_NETWORK, &buf) == 0) {
+                /* please restrict to the nids pertaining to
+                 * the specified networks */
+                while (class_parse_net(buf, &net, &buf) == 0) {
+                        if (LNET_NIDNET(nid) == net)
+                                return 1;
+                }
+        }

         return 0;
 }
 
         return 0;
 }
 


@@ -197,6 +254,8 @@ EXPORT_SYMBOL(class_find_param);
 EXPORT_SYMBOL(class_get_next_param);
 EXPORT_SYMBOL(class_match_param);
 EXPORT_SYMBOL(class_parse_nid);
 EXPORT_SYMBOL(class_get_next_param);
 EXPORT_SYMBOL(class_match_param);
 EXPORT_SYMBOL(class_parse_nid);

+EXPORT_SYMBOL(class_parse_net);
+EXPORT_SYMBOL(class_match_net);

 
 /********************** class fns **********************/
 
 
 /********************** class fns **********************/
 

diff --git a/lustre/obdclass/obd_mount.c b/lustre/obdclass/obd_mount.c
index c48f8b4..8531464 100644 (file)
--- a/lustre/obdclass/obd_mount.c
+++ b/lustre/obdclass/obd_mount.c
@@ -964,7 +964,7 @@ static int server_sb2mti(struct super_block *sb, struct mgs_target_info *mti)
         struct lustre_sb_info    *lsi = s2lsi(sb);
         struct lustre_disk_data  *ldd = lsi->lsi_ldd;
         lnet_process_id_t         id;
         struct lustre_sb_info    *lsi = s2lsi(sb);
         struct lustre_disk_data  *ldd = lsi->lsi_ldd;
         lnet_process_id_t         id;

-        int i = 0;
+        int                       i = 0;

         ENTRY;
 
         if (!(lsi->lsi_flags & LSI_SERVER))
         ENTRY;
 
         if (!(lsi->lsi_flags & LSI_SERVER))


@@ -979,6 +979,14 @@ static int server_sb2mti(struct super_block *sb, struct mgs_target_info *mti)
         while (LNetGetId(i++, &id) != -ENOENT) {
                 if (LNET_NETTYP(LNET_NIDNET(id.nid)) == LOLND)
                         continue;
         while (LNetGetId(i++, &id) != -ENOENT) {
                 if (LNET_NETTYP(LNET_NIDNET(id.nid)) == LOLND)
                         continue;

+
+                if (class_find_param(ldd->ldd_params,
+                                     PARAM_NETWORK, NULL) == 0 &&
+                    !class_match_net(ldd->ldd_params, id.nid)) {
+                        /* can't match specified network */
+                        continue;
+                }
+

                 mti->mti_nids[mti->mti_nid_count] = id.nid;
                 mti->mti_nid_count++;
                 if (mti->mti_nid_count >= MTI_NIDS_MAX) {
                 mti->mti_nids[mti->mti_nid_count] = id.nid;
                 mti->mti_nid_count++;
                 if (mti->mti_nid_count >= MTI_NIDS_MAX) {

diff --git a/lustre/utils/mkfs_lustre.c b/lustre/utils/mkfs_lustre.c
index c4fc9ff..f3e0b55 100644 (file)
--- a/lustre/utils/mkfs_lustre.c
+++ b/lustre/utils/mkfs_lustre.c
@@ -130,6 +130,7 @@ void usage(FILE *out)
                 */
                 "\t\t--comment=<user comment>: arbitrary user string (%d bytes)\n"
                 "\t\t--mountfsoptions=<opts> : permanent mount options\n"
                 */
                 "\t\t--comment=<user comment>: arbitrary user string (%d bytes)\n"
                 "\t\t--mountfsoptions=<opts> : permanent mount options\n"

+                "\t\t--network=<net>[,<...>] : network(s) to restrict this ost/mdt to\n"

 #ifndef TUNEFS
                 "\t\t--backfstype=<fstype> : backing fs type (ext3, ldiskfs)\n"
                 "\t\t--device-size=#N(KB) : device size for loop devices\n"
 #ifndef TUNEFS
                 "\t\t--backfstype=<fstype> : backing fs type (ext3, ldiskfs)\n"
                 "\t\t--device-size=#N(KB) : device size for loop devices\n"


@@ -1224,6 +1225,7 @@ int parse_opts(int argc, char *const argv[], struct mkfs_opts *mop,
                 {"verbose", 0, 0, 'v'},
                 {"writeconf", 0, 0, 'w'},
                 {"upgrade_to_18", 0, 0, 'U'},
                 {"verbose", 0, 0, 'v'},
                 {"writeconf", 0, 0, 'w'},
                 {"upgrade_to_18", 0, 0, 'U'},

+                {"network", 1, 0, 't'},

                 {0, 0, 0, 0}
         };
         char *optstring = "b:c:C:d:ef:Ghi:k:L:m:MnNo:Op:Pqru:vw";
                 {0, 0, 0, 0}
         };
         char *optstring = "b:c:C:d:ef:Ghi:k:L:m:MnNo:Op:Pqru:vw";


@@ -1379,6 +1381,22 @@ int parse_opts(int argc, char *const argv[], struct mkfs_opts *mop,
                 case 'r':
                         mop->mo_flags |= MO_FORCEFORMAT;
                         break;
                 case 'r':
                         mop->mo_flags |= MO_FORCEFORMAT;
                         break;

+                case 't':
+                        if (!IS_MDT(&mop->mo_ldd) && !IS_OST(&mop->mo_ldd)) {
+                                badopt(long_opt[longidx].name, "MDT,OST");
+                                return 1;
+                        }
+
+                        if (!optarg)
+                                return 1;
+
+                        rc = add_param(mop->mo_ldd.ldd_params,
+                                       PARAM_NETWORK, optarg);
+                        if (rc != 0)
+                                return rc;
+                        /* Must update the mgs logs */
+                        mop->mo_ldd.ldd_flags |= LDD_F_UPDATE;
+                        break;

                 case 'u':
                         strscpy(mop->mo_ldd.ldd_userdata, optarg,
                                 sizeof(mop->mo_ldd.ldd_userdata));
                 case 'u':
                         strscpy(mop->mo_ldd.ldd_userdata, optarg,
                                 sizeof(mop->mo_ldd.ldd_userdata));