BD Backend: Volume creation support

A new parameter type is added to volume create command. To use BD xlator
one has to specify following argument in addition to normal volume
create

device vg brick:<VG-NAME>

for example,
$ gluster volume create lv_volume device vg host:/vg1

Changes from previous version

* New type 'backend' added to volinfo structure to differentiate between
  posix and bd xlator
* Most of the volume related commands are updated to handle BD xlator,
  like add-brick, heal-brick etc refuse to work when volume is BD xlator
  type
* Only one VG (ie brick) can be specified for BD xlator during volume
  creation
* volume info shows VG info if its of type BD xlator

BUG: 805138
Change-Id: I0ff90aca04840c71f364fabb0ab43ce33f9278ce
Signed-off-by: M. Mohan Kumar <mohan@in.ibm.com>
Reviewed-on: http://review.gluster.org/3717
Reviewed-by: Vijay Bellur <vbellur@redhat.com>
Tested-by: Vijay Bellur <vbellur@redhat.com>

12 files changed, 217 insertions, 19 deletions

diff --git a/cli/src/cli-cmd-parser.c b/cli/src/cli-cmd-parser.c
index 4181e6c8162..f7ee29a10c7 100644
--- a/cli/src/cli-cmd-parser.c
+++ b/cli/src/cli-cmd-parser.c
@@ -159,13 +159,21 @@ cli_cmd_volume_create_parse (const char **words, int wordcount, dict_t **options
         int32_t index = 0;
         char    *bricks = NULL;
         int32_t brick_count = 0;
-        char    *opwords[] = { "replica", "stripe", "transport", NULL };
+        char    *opwords[] = { "replica", "stripe", "transport",
+#ifdef HAVE_BD_XLATOR
+                                "device",
+#endif
+                                NULL };
+
         char    *invalid_volnames[] = {"volume", "type", "subvolumes", "option",
                                       "end-volume", "all", NULL};
         char    *w = NULL;
         int      op_count = 0;
         int32_t  replica_count = 1;
         int32_t  stripe_count = 1;
+#ifdef HAVE_BD_XLATOR
+        char     *dev_type = NULL;
+#endif
 
         GF_ASSERT (words);
         GF_ASSERT (options);
@@ -300,7 +308,26 @@ cli_cmd_volume_create_parse (const char **words, int wordcount, dict_t **options
                                 goto out;
                         }
                         index += 2;
-                } else {
+                }
+#ifdef HAVE_BD_XLATOR
+                else if ((strcmp (w, "device")) == 0) {
+                        if (dev_type) {
+                                cli_err ("'device' option given more"
+                                         " than one time");
+                                goto out;
+                        }
+                        if ((strcasecmp (words[index+1], "vg") == 0)) {
+                                dev_type = gf_strdup ("vg");
+                        } else {
+                                gf_log ("", GF_LOG_ERROR, "incorrect"
+                                        " device type specified");
+                                ret = -1;
+                                goto out;
+                        }
+                        index += 2;
+                }
+#endif
+                else {
                         GF_ASSERT (!"opword mismatch");
                         ret = -1;
                         goto out;
@@ -336,6 +363,19 @@ cli_cmd_volume_create_parse (const char **words, int wordcount, dict_t **options
                 goto out;
         }
 
+        /* BD xlator does not support multiple bricks */
+#ifdef HAVE_BD_XLATOR
+        if (brick_count > 1 && dev_type) {
+                cli_err ("Block Device backend volume does not support multiple"
+                         " bricks");
+                gf_log ("", GF_LOG_ERROR,
+                          "Block Device backend volumer does not support multiple"
+                        " bricks");
+                ret = -1;
+                goto out;
+        }
+#endif
+
         if (brick_count % sub_count) {
                 if (type == GF_CLUSTER_TYPE_STRIPE)
                         cli_err ("number of bricks is not a multiple of "
@@ -369,6 +409,14 @@ cli_cmd_volume_create_parse (const char **words, int wordcount, dict_t **options
         if (ret)
                 goto out;
 
+#ifdef HAVE_BD_XLATOR
+        if (dev_type) {
+                ret = dict_set_dynstr (dict, "device", dev_type);
+                if (ret)
+                        goto out;
+        }
+#endif
+
         ret = dict_set_int32 (dict, "count", brick_count);
         if (ret)
                 goto out;
diff --git a/cli/src/cli-cmd-volume.c b/cli/src/cli-cmd-volume.c
index 5b02ef5d9fe..4dcb99ce26e 100644
--- a/cli/src/cli-cmd-volume.c
+++ b/cli/src/cli-cmd-volume.c
@@ -1795,7 +1795,7 @@ struct cli_cmd volume_cmds[] = {
           cli_cmd_volume_info_cbk,
           "list information of all volumes"},
 
-        { "volume create <NEW-VOLNAME> [stripe <COUNT>] [replica <COUNT>] [transport <tcp|rdma|tcp,rdma>] <NEW-BRICK> ...",
+        { "volume create <NEW-VOLNAME> [stripe <COUNT>] [replica <COUNT>] [device vg] [transport <tcp|rdma|tcp,rdma>] <NEW-BRICK> ...",
           cli_cmd_volume_create_cbk,
           "create a new volume of specified type with mentioned bricks"},
 
diff --git a/cli/src/cli-rpc-ops.c b/cli/src/cli-rpc-ops.c
index 703289b5b54..77e15ee296c 100644
--- a/cli/src/cli-rpc-ops.c
+++ b/cli/src/cli-rpc-ops.c
@@ -62,6 +62,10 @@ char *cli_vol_status_str[] = {"Created",
                               "Stopped",
                              };
 
+char *cli_volume_backend[] = {"",
+                           "Volume Group",
+};
+
 int32_t
 gf_cli_get_volume (call_frame_t *frame, xlator_t *this,
                       void *data);
@@ -493,6 +497,7 @@ gf_cli_get_volume_cbk (struct rpc_req *req, struct iovec *iov,
         char                       key[1024]            = {0};
         char                       err_str[2048]        = {0};
         gf_cli_rsp                 rsp                  = {0};
+        int32_t                    backend              = 0;
 
         if (-1 == req->rpc_status)
                 goto out;
@@ -643,6 +648,9 @@ xml_output:
                 if (ret)
                         goto out;
 
+                snprintf (key, 256, "volume%d.backend", i);
+                ret = dict_get_int32 (dict, key, &backend);
+
                 vol_type = type;
 
                 // Distributed (stripe/replicate/stripe-replica) setups
@@ -654,16 +662,17 @@ xml_output:
                 cli_out ("Volume ID: %s", volume_id_str);
                 cli_out ("Status: %s", cli_vol_status_str[status]);
 
+                if (backend)
+                        goto next;
+
                 if (type == GF_CLUSTER_TYPE_STRIPE_REPLICATE) {
                         cli_out ("Number of Bricks: %d x %d x %d = %d",
                                  (brick_count / dist_count),
                                  stripe_count,
                                  replica_count,
                                  brick_count);
-
                 } else if (type == GF_CLUSTER_TYPE_NONE) {
                         cli_out ("Number of Bricks: %d", brick_count);
-
                 } else {
                         /* For both replicate and stripe, dist_count is
                            good enough */
@@ -676,6 +685,12 @@ xml_output:
                          ((transport == 0)?"tcp":
                           (transport == 1)?"rdma":
                           "tcp,rdma"));
+
+next:
+                if (backend) {
+                        cli_out ("Backend Type: Block, %s",
+                                 cli_volume_backend[backend]);
+                }
                 j = 1;
 
                 GF_FREE (local->get_vol.volname);
diff --git a/xlators/mgmt/glusterd/src/Makefile.am b/xlators/mgmt/glusterd/src/Makefile.am
index 95d2aba9dcb..5ec9d2bef15 100644
--- a/xlators/mgmt/glusterd/src/Makefile.am
+++ b/xlators/mgmt/glusterd/src/Makefile.am
@@ -2,6 +2,9 @@ xlator_LTLIBRARIES = glusterd.la
 xlatordir = $(libdir)/glusterfs/$(PACKAGE_VERSION)/xlator/mgmt
 glusterd_la_CPPFLAGS = $(AM_CPPFLAGS) "-DFILTERDIR=\"$(libdir)/glusterfs/$(PACKAGE_VERSION)/filter\""
 glusterd_la_LDFLAGS = -module -avoidversion
+if ENABLE_BD_XLATOR
+glusterd_la_LDFLAGS += -llvm2app
+endif
 glusterd_la_SOURCES = glusterd.c glusterd-handler.c glusterd-sm.c \
 	glusterd-op-sm.c glusterd-utils.c glusterd-rpc-ops.c \
 	glusterd-store.c glusterd-handshake.c glusterd-pmap.c \
diff --git a/xlators/mgmt/glusterd/src/glusterd-brick-ops.c b/xlators/mgmt/glusterd/src/glusterd-brick-ops.c
index ade30fe951a..a14828e980b 100644
--- a/xlators/mgmt/glusterd/src/glusterd-brick-ops.c
+++ b/xlators/mgmt/glusterd/src/glusterd-brick-ops.c
@@ -1035,6 +1035,15 @@ glusterd_op_stage_add_brick (dict_t *dict, char **op_errstr)
                 goto out;
         }
 
+        if (volinfo->backend == GD_VOL_BK_BD) {
+                snprintf (msg, sizeof (msg), "Add brick is not supported for "
+                           "Block backend volume %s.", volname);
+                gf_log (THIS->name, GF_LOG_ERROR, "%s", msg);
+                *op_errstr = gf_strdup (msg);
+                ret = -1;
+                goto out;
+        }
+
         ret = glusterd_validate_volume_id (dict, volinfo);
         if (ret)
                 goto out;
diff --git a/xlators/mgmt/glusterd/src/glusterd-handler.c b/xlators/mgmt/glusterd/src/glusterd-handler.c
index cec94f89f6c..9433436d0ae 100644
--- a/xlators/mgmt/glusterd/src/glusterd-handler.c
+++ b/xlators/mgmt/glusterd/src/glusterd-handler.c
@@ -370,6 +370,13 @@ glusterd_add_volume_detail_to_dict (glusterd_volinfo_t *volinfo,
         if (ret)
                 goto out;
 
+#ifdef HAVE_BD_XLATOR
+        snprintf (key, 256, "volume%d.backend", count);
+        ret = dict_set_int32 (volumes, key, volinfo->backend);
+        if (ret)
+                goto out;
+#endif
+
         list_for_each_entry (brickinfo, &volinfo->bricks, brick_list) {
                 char    brick[1024] = {0,};
                 snprintf (key, 256, "volume%d.brick%d", count, i);
diff --git a/xlators/mgmt/glusterd/src/glusterd-replace-brick.c b/xlators/mgmt/glusterd/src/glusterd-replace-brick.c
index edc2627c1b3..bca306bd897 100644
--- a/xlators/mgmt/glusterd/src/glusterd-replace-brick.c
+++ b/xlators/mgmt/glusterd/src/glusterd-replace-brick.c
@@ -243,6 +243,13 @@ glusterd_op_stage_replace_brick (dict_t *dict, char **op_errstr,
                 goto out;
         }
 
+        if (volinfo->backend == GD_VOL_BK_BD) {
+                snprintf (msg, sizeof (msg), "replace brick not supported "
+                       "for Block backend volume");
+                *op_errstr = gf_strdup (msg);
+                goto out;
+        }
+
         if (GLUSTERD_STATUS_STARTED != volinfo->status) {
                 ret = -1;
                 snprintf (msg, sizeof (msg), "volume: %s is not started",
diff --git a/xlators/mgmt/glusterd/src/glusterd-store.c b/xlators/mgmt/glusterd/src/glusterd-store.c
index 471a24e6622..413c8a39abb 100644
--- a/xlators/mgmt/glusterd/src/glusterd-store.c
+++ b/xlators/mgmt/glusterd/src/glusterd-store.c
@@ -703,6 +703,15 @@ glusterd_volume_exclude_options_write (int fd, glusterd_volinfo_t *volinfo)
                 if (ret)
                         goto out;
         }
+
+        if (volinfo->backend == GD_VOL_BK_BD)  {
+                snprintf (buf, sizeof (buf), "%d", volinfo->backend);
+                ret = glusterd_store_save_value (fd,
+                        GLUSTERD_STORE_KEY_VOL_BACKEND, buf);
+                if (ret)
+                        goto out;
+        }
+
 out:
         if (ret)
                 gf_log ("", GF_LOG_ERROR, "Unable to write volume values"
@@ -2341,6 +2350,9 @@ glusterd_store_retrieve_volume (char    *volname)
                         gf_log ("", GF_LOG_DEBUG, "Parsed as "GEOREP" "
                                 " slave:key=%s,value:%s", key, value);
 
+                } else if (!strncmp (key, GLUSTERD_STORE_KEY_VOL_BACKEND,
+                                     strlen (GLUSTERD_STORE_KEY_VOL_BACKEND))) {
+                        volinfo->backend = atoi (value);
                 } else {
 
                         if (is_key_glusterd_hooks_friendly (key)) {
diff --git a/xlators/mgmt/glusterd/src/glusterd-store.h b/xlators/mgmt/glusterd/src/glusterd-store.h
index d8c1567a661..1ab398c0b77 100644
--- a/xlators/mgmt/glusterd/src/glusterd-store.h
+++ b/xlators/mgmt/glusterd/src/glusterd-store.h
@@ -64,6 +64,7 @@ typedef enum glusterd_store_ver_ac_{
 #define GLUSTERD_STORE_KEY_PEER_UUID      "uuid"
 #define GLUSTERD_STORE_KEY_PEER_HOSTNAME  "hostname"
 #define GLUSTERD_STORE_KEY_PEER_STATE     "state"
+#define GLUSTERD_STORE_KEY_VOL_BACKEND    "backend"
 
 #define glusterd_for_each_entry(entry, dir) \
         do {\
diff --git a/xlators/mgmt/glusterd/src/glusterd-volgen.c b/xlators/mgmt/glusterd/src/glusterd-volgen.c
index 76172dd9bce..f3de2feed52 100644
--- a/xlators/mgmt/glusterd/src/glusterd-volgen.c
+++ b/xlators/mgmt/glusterd/src/glusterd-volgen.c
@@ -1574,6 +1574,8 @@ server_graph_builder (volgen_graph_t *graph, glusterd_volinfo_t *volinfo,
         char     *password                = NULL;
         char     index_basepath[PATH_MAX] = {0};
         char     key[1024]                = {0};
+        char     *vgname                  = NULL;
+        char     *vg                      = NULL;
 
         path = param;
         volname = volinfo->volname;
@@ -1591,23 +1593,47 @@ server_graph_builder (volgen_graph_t *graph, glusterd_volinfo_t *volinfo,
                 }
         }
 
-        xl = volgen_graph_add (graph, "storage/posix", volname);
-        if (!xl)
-                return -1;
+        if (volinfo->backend == GD_VOL_BK_BD) {
+                xl = volgen_graph_add (graph, "storage/bd_map", volname);
+                if (!xl)
+                        return -1;
 
-        ret = xlator_set_option (xl, "directory", path);
-        if (ret)
-                return -1;
+                ret = xlator_set_option (xl, "device", "vg");
+                if (ret)
+                        return -1;
 
-        ret = xlator_set_option (xl, "volume-id",
-                                 uuid_utoa (volinfo->volume_id));
-        if (ret)
-                return -1;
+                vg = gf_strdup (path);
+                vgname = strrchr (vg, '/');
+                if (strchr(vg, '/') != vgname) {
+                        gf_log ("glusterd", GF_LOG_ERROR,
+                                  "invalid vg specified %s", path);
+                        GF_FREE (vg);
+                        goto out;
+                }
+                vgname++;
+                ret = xlator_set_option (xl, "export", vgname);
+                GF_FREE (vg);
+                if (ret)
+                        return -1;
+        } else {
+                xl = volgen_graph_add (graph, "storage/posix", volname);
+                if (!xl)
+                        return -1;
 
-        ret = check_and_add_debug_xl (graph, set_dict, volname, "posix");
-        if (ret)
-                return -1;
+                ret = xlator_set_option (xl, "directory", path);
+                if (ret)
+                        return -1;
 
+                ret = xlator_set_option (xl, "volume-id",
+                                 uuid_utoa (volinfo->volume_id));
+                if (ret)
+                        return -1;
+
+                ret = check_and_add_debug_xl (graph, set_dict, volname,
+                                                "posix");
+                if (ret)
+                        return -1;
+        }
         xl = volgen_graph_add (graph, "features/access-control", volname);
         if (!xl)
                 return -1;
diff --git a/xlators/mgmt/glusterd/src/glusterd-volume-ops.c b/xlators/mgmt/glusterd/src/glusterd-volume-ops.c
index d42694353af..8c76c8f09e1 100644
--- a/xlators/mgmt/glusterd/src/glusterd-volume-ops.c
+++ b/xlators/mgmt/glusterd/src/glusterd-volume-ops.c
@@ -12,6 +12,10 @@
 #include "config.h"
 #endif
 
+#ifdef HAVE_BD_XLATOR
+#include <lvm2app.h>
+#endif
+
 #include "common-utils.h"
 #include "syscall.h"
 #include "cli1-xdr.h"
@@ -554,6 +558,36 @@ out:
         return ret;
 }
 
+#ifdef HAVE_BD_XLATOR
+int
+glusterd_is_valid_vg (const char *name)
+{
+        lvm_t           handle   = NULL;
+        vg_t            vg       = NULL;
+        char            *vg_name = NULL;
+        int             retval   = -1;
+
+        handle = lvm_init (NULL);
+        if (!handle) {
+                gf_log ("", GF_LOG_ERROR, "lvm_init failed");
+                return -1;
+        }
+        vg_name = gf_strdup (name);
+        vg = lvm_vg_open (handle, basename (vg_name), "r", 0);
+        if (!vg) {
+                gf_log ("", GF_LOG_ERROR, "no such vg: %s", vg_name);
+                goto out;
+        }
+        retval = 0;
+out:
+        if (vg)
+                lvm_vg_close (vg);
+        lvm_quit (handle);
+        GF_FREE (vg_name);
+        return retval;
+}
+#endif
+
 /* op-sm */
 int
 glusterd_op_stage_create_volume (dict_t *dict, char **op_errstr)
@@ -575,7 +609,9 @@ glusterd_op_stage_create_volume (dict_t *dict, char **op_errstr)
         char                                    msg[2048] = {0};
         uuid_t                                  volume_uuid;
         char                                    *volume_uuid_str;
-
+#ifdef HAVE_BD_XLATOR
+        char                                    *dev_type = NULL;
+#endif
         this = THIS;
         if (!this) {
                 gf_log ("glusterd", GF_LOG_ERROR,
@@ -625,6 +661,11 @@ glusterd_op_stage_create_volume (dict_t *dict, char **op_errstr)
                 goto out;
         }
 
+
+#ifdef HAVE_BD_XLATOR
+        ret = dict_get_str (dict, "device", &dev_type);
+#endif
+
         ret = dict_get_str (dict, "bricks", &bricks);
         if (ret) {
                 gf_log ("", GF_LOG_ERROR, "Unable to get bricks");
@@ -670,6 +711,19 @@ glusterd_op_stage_create_volume (dict_t *dict, char **op_errstr)
                         goto out;
                 }
 
+#ifdef HAVE_BD_XLATOR
+                if (dev_type) {
+                        ret = glusterd_is_valid_vg (brick_info->path);
+                        if (ret) {
+                                snprintf (msg, sizeof(msg), "invalid vg %s",
+                                  brick_info->path);
+                                *op_errstr = gf_strdup (msg);
+                                goto out;
+                        }
+
+                        break;
+                } else
+#endif
                 if (!uuid_compare (brick_info->uuid, MY_UUID)) {
                         ret = glusterd_brick_create_path (brick_info->hostname,
                                                           brick_info->path,
@@ -1209,6 +1263,9 @@ glusterd_op_create_volume (dict_t *dict, char **op_errstr)
         char                 *str        = NULL;
         char                 *username   = NULL;
         char                 *password   = NULL;
+#ifdef HAVE_BD_XLATOR
+        char                 *device     = NULL;
+#endif
 
         this = THIS;
         GF_ASSERT (this);
@@ -1261,6 +1318,12 @@ glusterd_op_create_volume (dict_t *dict, char **op_errstr)
                 goto out;
         }
 
+#ifdef HAVE_BD_XLATOR
+        ret = dict_get_str (dict, "device", &device);
+        if (!ret)
+                volinfo->backend = GD_VOL_BK_BD;
+#endif
+
         /* replica-count 1 means, no replication, file is in one brick only */
         volinfo->replica_count = 1;
         /* stripe-count 1 means, no striping, file is present as a whole */
diff --git a/xlators/mgmt/glusterd/src/glusterd.h b/xlators/mgmt/glusterd/src/glusterd.h
index 95353f06514..ea0ea6061d3 100644
--- a/xlators/mgmt/glusterd/src/glusterd.h
+++ b/xlators/mgmt/glusterd/src/glusterd.h
@@ -229,6 +229,12 @@ struct _auth {
 
 typedef struct _auth auth_t;
 
+typedef enum glusterd_vol_backend_ {
+        GD_VOL_BK_DEFAULT = 0, /* POSIX */
+        GD_VOL_BK_BD = 1,
+} glusterd_vol_backend_t;
+
+
 struct glusterd_volinfo_ {
         char                    volname[GLUSTERD_MAX_VOLUME_NAME];
         int                     type;
@@ -278,6 +284,7 @@ struct glusterd_volinfo_ {
         xlator_t                *xl;
 
         gf_boolean_t             memory_accounting;
+        glusterd_vol_backend_t   backend;
 };
 
 typedef enum gd_node_type_ {