glusterd: Change the volfile to have readdir-ahead as a child

          of dht

As mentioned in feature page http://review.gluster.org/#/c/16090/
readdir-ahead will be optionally placed below dht.

There are two options:
1. performance.readdir-ahead
2. performance.parallel-readdir

If only option is enabled, then readdir ahead is placed at its
original place as an ancestor of dht. If both the options 1 and 2
are enabled then readdir ahead is placed as a child of dht.

Also changes have been made to retain the rebalance, quotad,
snapd vol files to remain unchanged.

Change-Id: I0adf0b476fcbf91251f5a2fee2241786a3d8255a
BUG: 1401812
Signed-off-by: Poornima G <pgurusid@redhat.com>
Reviewed-on: http://review.gluster.org/16072
Reviewed-by: Raghavendra G <rgowdapp@redhat.com>
Smoke: Gluster Build System <jenkins@build.gluster.org>
NetBSD-regression: NetBSD Build System <jenkins@build.gluster.org>
CentOS-regression: Gluster Build System <jenkins@build.gluster.org>
Reviewed-by: Atin Mukherjee <amukherj@redhat.com>

4 files changed, 63 insertions, 18 deletions

diff --git a/xlators/mgmt/glusterd/src/glusterd-utils.c b/xlators/mgmt/glusterd/src/glusterd-utils.c
index 7d1835236fe..c408f12c10d 100644
--- a/xlators/mgmt/glusterd/src/glusterd-utils.c
+++ b/xlators/mgmt/glusterd/src/glusterd-utils.c
@@ -11028,22 +11028,6 @@ glusterd_enable_default_options (glusterd_volinfo_t *volinfo, char *option)
                  * }
                  * */
 
-                /* readdir-ahead needs to be enabled for new volumes with
-                 * >= gluster version 3.7
-                 */
-                if (!option || !strcmp ("performance.readdir-ahead", option)) {
-                        ret = dict_set_dynstr_with_alloc (volinfo->dict,
-                                        "performance.readdir-ahead", "on");
-                        if (ret) {
-                                gf_msg (this->name, GF_LOG_ERROR, errno,
-                                        GD_MSG_DICT_SET_FAILED,
-                                        "Failed to set option "
-                                        "'performance.readdir-ahead' on volume "
-                                        "%s", volinfo->volname);
-                                goto out;
-                        }
-                }
-
                 /* Option 'features.quota-deem-statfs' should not be turned off
                  * with 'gluster volume reset <VOLNAME>', since quota features
                  * can be reset only with 'gluster volume quota <VOLNAME>
diff --git a/xlators/mgmt/glusterd/src/glusterd-volgen.c b/xlators/mgmt/glusterd/src/glusterd-volgen.c
index 9537156b6c3..144bd5a4d57 100644
--- a/xlators/mgmt/glusterd/src/glusterd-volgen.c
+++ b/xlators/mgmt/glusterd/src/glusterd-volgen.c
@@ -2647,6 +2647,13 @@ perfxl_option_handler (volgen_graph_t *graph, struct volopt_map_entry *vme,
             (vme->op_version > volinfo->client_op_version))
                 return 0;
 
+        /* if VKEY_READDIR_AHEAD is enabled and parallel readdir is
+         * not enabled then load readdir-ahead here else it will be
+         * loaded as a child of dht */
+        if (!strcmp (vme->key, VKEY_READDIR_AHEAD) &&
+            glusterd_volinfo_get_boolean (volinfo, VKEY_PARALLEL_READDIR))
+                return 0;
+
         if (volgen_graph_add (graph, vme->voltype, volinfo->volname))
                 return 0;
         else
@@ -3391,6 +3398,30 @@ out:
 }
 
 static int
+volgen_graph_build_readdir_ahead (volgen_graph_t *graph,
+                                  glusterd_volinfo_t *volinfo,
+                                  size_t child_count)
+{
+        int32_t                 clusters                 = 0;
+
+        if (graph->type == GF_REBALANCED ||
+            graph->type == GF_QUOTAD ||
+            graph->type == GF_SNAPD ||
+            !glusterd_volinfo_get_boolean (volinfo, VKEY_PARALLEL_READDIR) ||
+            !glusterd_volinfo_get_boolean (volinfo, VKEY_READDIR_AHEAD))
+                goto out;
+
+        clusters = volgen_link_bricks_from_list_tail (graph,
+                                                      volinfo,
+                                                      "performance/readdir-ahead",
+                                                      "%s-readdir-ahead-%d",
+                                                      child_count,
+                                                      1);
+out:
+        return clusters;
+}
+
+static int
 volgen_graph_build_dht_cluster (volgen_graph_t *graph,
                                 glusterd_volinfo_t *volinfo, size_t child_count,
                                 gf_boolean_t is_quotad)
@@ -3726,6 +3757,11 @@ build_distribute:
                 else
                         strcat (volinfo->volname, "-cold");
         }
+        clusters = volgen_graph_build_readdir_ahead (graph, volinfo,
+                                                     dist_count);
+        if (clusters < 0)
+                goto out;
+
         ret = volgen_graph_build_dht_cluster (graph, volinfo,
                                               dist_count, is_quotad);
         if (volinfo->tier_info.hot_brick_count)
@@ -4750,6 +4786,8 @@ build_rebalance_volfile (glusterd_volinfo_t *volinfo, char *filepath,
 
         this = THIS;
 
+        graph.type = GF_REBALANCED;
+
         if (volinfo->brick_count <= volinfo->dist_leaf_count) {
                 /*
                  * Volume is not a distribute volume or
@@ -5186,6 +5224,8 @@ build_quotad_graph (volgen_graph_t *graph, dict_t *mod_dict)
         priv = this->private;
         GF_ASSERT (priv);
 
+        graph->type = GF_QUOTAD;
+
         set_dict = dict_new ();
         if (!set_dict) {
                 ret = -ENOMEM;
@@ -6050,6 +6090,7 @@ glusterd_snapdsvc_create_volfile (glusterd_volinfo_t *volinfo)
         int             ret                     = -1;
         char            filename [PATH_MAX]     = {0,};
 
+        graph.type = GF_SNAPD;
         glusterd_svc_build_snapd_volfile (volinfo, filename, PATH_MAX);
 
         ret = glusterd_snapdsvc_generate_volfile (&graph, volinfo);
diff --git a/xlators/mgmt/glusterd/src/glusterd-volgen.h b/xlators/mgmt/glusterd/src/glusterd-volgen.h
index 2d62f720c8a..de1ba9428f5 100644
--- a/xlators/mgmt/glusterd/src/glusterd-volgen.h
+++ b/xlators/mgmt/glusterd/src/glusterd-volgen.h
@@ -32,6 +32,8 @@
 #define VKEY_FEATURES_TRASH       "features.trash"
 #define VKEY_FEATURES_BITROT      "features.bitrot"
 #define VKEY_FEATURES_SCRUB       "features.scrub"
+#define VKEY_PARALLEL_READDIR     "performance.parallel-readdir"
+#define VKEY_READDIR_AHEAD        "performance.readdir-ahead"
 
 #define AUTH_ALLOW_MAP_KEY "auth.allow"
 #define AUTH_REJECT_MAP_KEY "auth.reject"
@@ -55,9 +57,17 @@ typedef enum {
         GF_CLIENT_OTHER
 } glusterd_client_type_t;
 
+/* It indicates the type of volfile that the graph is built for */
+typedef enum {
+        GF_REBALANCED = 1,
+        GF_QUOTAD,
+        GF_SNAPD,
+} glusterd_graph_type_t;
+
 struct volgen_graph {
         char **errstr;
         glusterfs_graph_t graph;
+        glusterd_graph_type_t type;
 };
 typedef struct volgen_graph volgen_graph_t;
 
diff --git a/xlators/mgmt/glusterd/src/glusterd-volume-set.c b/xlators/mgmt/glusterd/src/glusterd-volume-set.c
index a75770220b3..ae569c66a23 100644
--- a/xlators/mgmt/glusterd/src/glusterd-volume-set.c
+++ b/xlators/mgmt/glusterd/src/glusterd-volume-set.c
@@ -1867,12 +1867,11 @@ struct volopt_map_entry glusterd_volopt_map[] = {
         { .key         = "performance.readdir-ahead",
           .voltype     = "performance/readdir-ahead",
           .option      = "!perf",
-          .value       = "off",
+          .value       = "on",
           .op_version  = 3,
           .description = "enable/disable readdir-ahead translator in the volume.",
           .flags       = OPT_FLAG_CLIENT_OPT | OPT_FLAG_XLATOR_OPT
         },
-
         { .key         = "performance.io-cache",
           .voltype     = "performance/io-cache",
           .option      = "!perf",
@@ -3092,6 +3091,17 @@ struct volopt_map_entry glusterd_volopt_map[] = {
           .op_version = GD_OP_VERSION_3_8_4,
           .flags      = OPT_FLAG_CLIENT_OPT
         },
+        { .key         = "performance.parallel-readdir",
+          .voltype     = "performance/readdir-ahead",
+          .option      = "parallel-readdir",
+          .value       = "off",
+          .type        = DOC,
+          .op_version  = GD_OP_VERSION_3_10_0,
+          .description = "If this option is enabled, the readdir operation is "
+                         "performed parallely on all the bricks, thus improving"
+                         " the performance of readdir. Note that the performance"
+                         "improvement is higher in large clusters"
+        },
 	{ .key         = "performance.rda-request-size",
 	  .voltype     = "performance/readdir-ahead",
           .option      = "rda-request-size",