cluster/afr: Make read child match check in afr optionalv3.5.4beta1

        Backport of: http://review.gluster.org/#/c/9917

Having this particular check which was introduced by
commit bb2df4e63fa8a5d65f18b4a5efc757e8d475fbff causes a drop in
performance in readdirp. So the behavior is made configurable with this
patch.

Change-Id: I9012a6bb955229a0cbb48f06e4e2edc0782dfead
BUG: 1202675
Signed-off-by: Krutika Dhananjay <kdhananj@redhat.com>
Reviewed-on: http://review.gluster.org/9924
Reviewed-by: Pranith Kumar Karampuri <pkarampu@redhat.com>
Reviewed-by: Atin Mukherjee <amukherj@redhat.com>
Tested-by: Gluster Build System <jenkins@build.gluster.com>
Reviewed-by: Niels de Vos <ndevos@redhat.com>

5 files changed, 30 insertions, 1 deletions

diff --git a/libglusterfs/src/globals.h b/libglusterfs/src/globals.h
index 3d2cdc306c1..3eceb4bb6df 100644
--- a/libglusterfs/src/globals.h
+++ b/libglusterfs/src/globals.h
@@ -34,10 +34,12 @@
  */
 #define GD_OP_VERSION_MIN  1 /* MIN is the fresh start op-version, mostly
                                 should not change */
-#define GD_OP_VERSION_MAX  30501 /* MAX VERSION is the maximum count in VME
+#define GD_OP_VERSION_MAX  30504 /* MAX VERSION is the maximum count in VME
                                     table, should keep changing with
                                     introduction of newer versions */
 
+#define GD_OP_VERSION_3_5_4 30504
+
 #include "xlator.h"
 
 /* THIS */
diff --git a/xlators/cluster/afr/src/afr-dir-read.c b/xlators/cluster/afr/src/afr-dir-read.c
index d94cb0ca699..2d368a49544 100644
--- a/xlators/cluster/afr/src/afr-dir-read.c
+++ b/xlators/cluster/afr/src/afr-dir-read.c
@@ -422,11 +422,14 @@ afr_readdirp_cbk (call_frame_t *frame, void *cookie, xlator_t *this,
                   int32_t op_ret, int32_t op_errno, gf_dirent_t *entries,
                   dict_t *xdata)
 {
+        afr_private_t   *priv           = NULL;
         afr_local_t     *local          = NULL;
         gf_dirent_t     *entry          = NULL;
         int              par_read_child = (long) cookie;
         int32_t          read_child     = -1;
 
+        priv = this->private;
+
         if (op_ret == -1)
                 goto out;
 
@@ -434,6 +437,9 @@ afr_readdirp_cbk (call_frame_t *frame, void *cookie, xlator_t *this,
 
         afr_readdir_filter_trash_dir (entries, local->fd);
 
+        if (!priv->consistent_metadata)
+                goto out;
+
         list_for_each_entry (entry, &entries->list, list) {
                 if (entry->inode) {
                         read_child = -1;
diff --git a/xlators/cluster/afr/src/afr.c b/xlators/cluster/afr/src/afr.c
index c020cbccac5..0739d0de93b 100644
--- a/xlators/cluster/afr/src/afr.c
+++ b/xlators/cluster/afr/src/afr.c
@@ -193,6 +193,8 @@ reconfigure (xlator_t *this, dict_t *options)
                           bool, out);
         GF_OPTION_RECONF ("ensure-durability", priv->ensure_durability, options,
                           bool, out);
+        GF_OPTION_RECONF ("consistent-metadata", priv->consistent_metadata,
+                          options, bool, out);
         priv->did_discovery = _gf_false;
 
         ret = 0;
@@ -339,6 +341,8 @@ init (xlator_t *this)
         GF_OPTION_INIT ("readdir-failover", priv->readdir_failover, bool, out);
         GF_OPTION_INIT ("ensure-durability", priv->ensure_durability, bool,
                         out);
+        GF_OPTION_INIT ("consistent-metadata", priv->consistent_metadata, bool,
+                        out);
 
         priv->wait_count = 1;
 
@@ -789,5 +793,15 @@ struct volume_options options[] = {
                          "written to the disk",
           .default_value = "on",
         },
+        { .key = {"consistent-metadata"},
+          .type = GF_OPTION_TYPE_BOOL,
+          .default_value = "no",
+          .description = "If this option is enabled, readdirp will force "
+                         "lookups on those entries read whose read child is "
+                         "not the same as that of the parent. This will "
+                         "guarantee that all read operations on a file serve "
+                         "attributes from the same subvol as long as it holds "
+                         " a good copy of the file/dir.",
+        },
         { .key  = {NULL} },
 };
diff --git a/xlators/cluster/afr/src/afr.h b/xlators/cluster/afr/src/afr.h
index e704dc2e929..990279d160b 100644
--- a/xlators/cluster/afr/src/afr.h
+++ b/xlators/cluster/afr/src/afr.h
@@ -180,6 +180,7 @@ typedef struct _afr_private {
         uint64_t               sh_readdir_size;
         gf_boolean_t           ensure_durability;
         char                   *sh_domain;
+        gf_boolean_t           consistent_metadata;
 } afr_private_t;
 
 typedef enum {
diff --git a/xlators/mgmt/glusterd/src/glusterd-volume-set.c b/xlators/mgmt/glusterd/src/glusterd-volume-set.c
index b8bad7622dc..869d11357dd 100644
--- a/xlators/mgmt/glusterd/src/glusterd-volume-set.c
+++ b/xlators/mgmt/glusterd/src/glusterd-volume-set.c
@@ -554,6 +554,12 @@ struct volopt_map_entry glusterd_volopt_map[] = {
           .op_version = 3,
           .flags      = OPT_FLAG_CLIENT_OPT
         },
+        { .key        = "cluster.consistent-metadata",
+          .voltype    = "cluster/replicate",
+          .type       = DOC,
+          .op_version = GD_OP_VERSION_3_5_4,
+          .flags      = OPT_FLAG_CLIENT_OPT
+        },
 
         /* Stripe xlator options */
         { .key         = "cluster.stripe-block-size",