cluster/afr: Make read child match check in afr optional

Having this particular check which was introduced by
commit c78998c39f0857ea7aacba360632c148afc54a55 causes a drop in
performance in readdirp. So the behavior is made configurable with this
patch.

Change-Id: I2858fc18b3539df7aa6d3f489e0d5cfaeb8a9b3c
BUG: 1202669
Signed-off-by: Krutika Dhananjay <kdhananj@redhat.com>
Reviewed-on: http://review.gluster.org/9917
Tested-by: Gluster Build System <jenkins@build.gluster.com>
Reviewed-by: Pranith Kumar Karampuri <pkarampu@redhat.com>
Tested-by: Pranith Kumar Karampuri <pkarampu@redhat.com>

4 files changed, 25 insertions, 0 deletions

diff --git a/xlators/cluster/afr/src/afr-dir-read.c b/xlators/cluster/afr/src/afr-dir-read.c
index 28bf89f2842..984ed9c6095 100644
--- a/xlators/cluster/afr/src/afr-dir-read.c
+++ b/xlators/cluster/afr/src/afr-dir-read.c
@@ -149,6 +149,10 @@ afr_validate_read_subvol (inode_t *inode, xlator_t *this, int par_read_subvol)
          * either afr_data_subvol_get() or afr_metadata_subvol_get() would
          * yield the same result. Hence, choosing afr_data_subvol_get() below.
          */
+
+        if (!priv->consistent_metadata)
+                return 0;
+
         entry_read_subvol = afr_data_subvol_get (inode, this, 0, 0);
         if (entry_read_subvol != par_read_subvol)
                 return -1;
diff --git a/xlators/cluster/afr/src/afr.c b/xlators/cluster/afr/src/afr.c
index f435767f5e4..f962fb6494e 100644
--- a/xlators/cluster/afr/src/afr.c
+++ b/xlators/cluster/afr/src/afr.c
@@ -213,6 +213,8 @@ reconfigure (xlator_t *this, dict_t *options)
 
         GF_OPTION_RECONF ("quorum-reads", priv->quorum_reads, options,
                           bool, out);
+        GF_OPTION_RECONF ("consistent-metadata", priv->consistent_metadata,
+                          options, bool, out);
 
         priv->did_discovery = _gf_false;
 
@@ -363,6 +365,8 @@ init (xlator_t *this)
         GF_OPTION_INIT ("heal-timeout", priv->shd.timeout, int32, out);
 
         GF_OPTION_INIT ("quorum-reads", priv->quorum_reads, bool, out);
+        GF_OPTION_INIT ("consistent-metadata", priv->consistent_metadata, bool,
+                        out);
 
         priv->wait_count = 1;
 
@@ -780,5 +784,15 @@ struct volume_options options[] = {
           .description = "time interval for checking the need to self-heal "
                          "in self-heal-daemon"
         },
+        { .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 dff70e89bb3..d7d15c69845 100644
--- a/xlators/cluster/afr/src/afr.h
+++ b/xlators/cluster/afr/src/afr.h
@@ -125,6 +125,7 @@ typedef struct _afr_private {
 	/* pump dependencies */
 	void                   *pump_private;
 	gf_boolean_t           use_afr_in_pump;
+        gf_boolean_t           consistent_metadata;
 } afr_private_t;
 
 
diff --git a/xlators/mgmt/glusterd/src/glusterd-volume-set.c b/xlators/mgmt/glusterd/src/glusterd-volume-set.c
index b0a3b0c8c39..891801fd755 100644
--- a/xlators/mgmt/glusterd/src/glusterd-volume-set.c
+++ b/xlators/mgmt/glusterd/src/glusterd-volume-set.c
@@ -517,6 +517,12 @@ 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_7_0,
+          .flags      = OPT_FLAG_CLIENT_OPT
+        },
 
         /* Stripe xlator options */
         { .key         = "cluster.stripe-block-size",