cluster/afr: Provide an option to disable afr durability

Change-Id: I40eec20ca6b3f857245a2438883822e251077ee9
BUG: 979365
Signed-off-by: Pranith Kumar K <pkarampu@redhat.com>
Reviewed-on: http://review.gluster.org/5269
Tested-by: Gluster Build System <jenkins@build.gluster.com>
Reviewed-by: Vijay Bellur <vbellur@redhat.com>

7 files changed, 78 insertions, 3 deletions

diff --git a/doc/admin-guide/en-US/admin_managing_volumes.xml b/doc/admin-guide/en-US/admin_managing_volumes.xml
index 0c4d2e922cf..70c1fe0b995 100644
--- a/doc/admin-guide/en-US/admin_managing_volumes.xml
+++ b/doc/admin-guide/en-US/admin_managing_volumes.xml
@@ -114,6 +114,12 @@ Set volume successful</programlisting></para>
                 <entry>On | Off</entry>
               </row>
               <row>
+                <entry>cluster.ensure-durability</entry>
+                <entry>This option makes sure the data/metadata is durable across abrupt shutdown of the brick. </entry>
+                <entry>on</entry>
+                <entry>On | Off</entry>
+              </row>
+              <row>
                 <entry>diagnostics.brick-log-level</entry>
                 <entry>Changes the log-level of the bricks. </entry>
                 <entry>INFO </entry>
diff --git a/doc/admin-guide/en-US/markdown/admin_managing_volumes.md b/doc/admin-guide/en-US/markdown/admin_managing_volumes.md
index 5c1b5347982..6375bf5257e 100644
--- a/doc/admin-guide/en-US/markdown/admin_managing_volumes.md
+++ b/doc/admin-guide/en-US/markdown/admin_managing_volumes.md
@@ -73,6 +73,8 @@ To tune volume options
 
       cluster.self-heal-daemon               Allows you to turn-off proactive self-heal on replicated volumes.                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               on                                 On | Off
 
+      cluster.ensure-durability              This option makes sure the data/metadata is durable across abrupt shutdown of the brick.                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                        on                                 On | Off
+
       diagnostics.brick-log-level            Changes the log-level of the bricks.                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                            INFO                               DEBUG|WARNING|ERROR|CRITICAL|NONE|TRACE
 
       diagnostics.client-log-level           Changes the log-level of the clients.                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                           INFO                               DEBUG|WARNING|ERROR|CRITICAL|NONE|TRACE
diff --git a/tests/bugs/bug-979365.t b/tests/bugs/bug-979365.t
new file mode 100755
index 00000000000..f5d9efd5487
--- /dev/null
+++ b/tests/bugs/bug-979365.t
@@ -0,0 +1,42 @@
+#!/bin/bash
+
+. $(dirname $0)/../include.rc
+. $(dirname $0)/../volume.rc
+
+#This script checks that ensure-durability option enables/disables afr
+#sending fsyncs
+cleanup;
+
+function num_fsyncs {
+        $CLI volume profile $V0 info | grep -w FSYNC | wc -l
+}
+
+TEST glusterd
+TEST pidof glusterd
+TEST $CLI volume create $V0 replica 2 $H0:$B0/${V0}{0,1}
+TEST $CLI volume set $V0 ensure-durability on
+TEST $CLI volume set $V0 eager-lock off
+TEST $CLI volume start $V0
+TEST $CLI volume profile $V0 start
+TEST glusterfs --volfile-id=/$V0 --volfile-server=$H0 $M0
+TEST kill_brick $V0 $H0 $B0/${V0}0
+TEST dd of=$M0/a if=/dev/zero bs=1M count=10
+#fsyncs take a while to complete.
+sleep 5
+TEST [[ $(num_fsyncs) -ne 0 ]]
+#Stop the volume to erase the profile info of old operations
+TEST $CLI volume profile $V0 stop
+TEST $CLI volume stop $V0
+umount $M0
+#Disable ensure-durability now to disable fsyncs in afr.
+TEST $CLI volume set $V0 ensure-durability off
+TEST $CLI volume start $V0
+TEST kill_brick $V0 $H0 $B0/${V0}0
+TEST glusterfs --volfile-id=/$V0 --volfile-server=$H0 $M0
+TEST $CLI volume profile $V0 start
+TEST dd of=$M0/a if=/dev/zero bs=1M count=10
+#fsyncs take a while to complete.
+sleep 5
+TEST [[ $(num_fsyncs) -eq 0 ]]
+
+cleanup
diff --git a/xlators/cluster/afr/src/afr-transaction.c b/xlators/cluster/afr/src/afr-transaction.c
index dbc74ceba53..d307d5b3eee 100644
--- a/xlators/cluster/afr/src/afr-transaction.c
+++ b/xlators/cluster/afr/src/afr-transaction.c
@@ -1495,8 +1495,10 @@ int
 afr_changelog_post_op_safe (call_frame_t *frame, xlator_t *this)
 {
 	afr_local_t    *local = NULL;
+        afr_private_t  *priv = NULL;
 
 	local = frame->local;
+        priv = this->private;
 
 	if (!local->fd || local->transaction.type != AFR_DATA_TRANSACTION) {
 		afr_changelog_post_op_now (frame, this);
@@ -1548,9 +1550,15 @@ afr_changelog_post_op_safe (call_frame_t *frame, xlator_t *this)
 		return 0;
 	}
 
-	/* Time to fsync() */
-
-	afr_changelog_fsync (frame, this);
+        /* Check whether users want durability and perform fsync/post-op
+         * accordingly.
+         */
+        if (priv->ensure_durability) {
+                /* Time to fsync() */
+                afr_changelog_fsync (frame, this);
+        } else {
+                afr_changelog_post_op_now (frame, this);
+        }
 
 	return 0;
 }
diff --git a/xlators/cluster/afr/src/afr.c b/xlators/cluster/afr/src/afr.c
index 6fabc00171f..9a13fbdacd9 100644
--- a/xlators/cluster/afr/src/afr.c
+++ b/xlators/cluster/afr/src/afr.c
@@ -191,6 +191,8 @@ reconfigure (xlator_t *this, dict_t *options)
         /* Reset this so we re-discover in case the topology changed.  */
         GF_OPTION_RECONF ("readdir-failover", priv->readdir_failover, options,
                           bool, out);
+        GF_OPTION_RECONF ("ensure-durability", priv->ensure_durability, options,
+                          bool, out);
         priv->did_discovery = _gf_false;
 
         ret = 0;
@@ -335,6 +337,8 @@ init (xlator_t *this)
 
 	GF_OPTION_INIT ("post-op-delay-secs", priv->post_op_delay_secs, uint32, out);
         GF_OPTION_INIT ("readdir-failover", priv->readdir_failover, bool, out);
+        GF_OPTION_INIT ("ensure-durability", priv->ensure_durability, bool,
+                        out);
 
         priv->wait_count = 1;
 
@@ -767,5 +771,12 @@ struct volume_options options[] = {
           .description = "readdir(p) will not failover if this option is off",
           .default_value = "on",
         },
+        { .key = {"ensure-durability"},
+          .type = GF_OPTION_TYPE_BOOL,
+          .description = "Afr performs fsyncs for transactions if this "
+                         "option is on to make sure the changelogs/data is "
+                         "written to the disk",
+          .default_value = "on",
+        },
         { .key  = {NULL} },
 };
diff --git a/xlators/cluster/afr/src/afr.h b/xlators/cluster/afr/src/afr.h
index 836a64d0a1f..e72398a62bc 100644
--- a/xlators/cluster/afr/src/afr.h
+++ b/xlators/cluster/afr/src/afr.h
@@ -170,6 +170,7 @@ typedef struct _afr_private {
         gf_boolean_t           did_discovery;
         gf_boolean_t           readdir_failover;
         uint64_t               sh_readdir_size;
+        gf_boolean_t           ensure_durability;
 } 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 6c4af6e3e61..f57f49f42e9 100644
--- a/xlators/mgmt/glusterd/src/glusterd-volume-set.c
+++ b/xlators/mgmt/glusterd/src/glusterd-volume-set.c
@@ -524,6 +524,11 @@ struct volopt_map_entry glusterd_volopt_map[] = {
           .op_version = 2,
           .flags      = OPT_FLAG_CLIENT_OPT
         },
+        { .key        = "cluster.ensure-durability",
+          .voltype    = "cluster/replicate",
+          .op_version = 2,
+          .flags      = OPT_FLAG_CLIENT_OPT
+        },
 
         /* Stripe xlator options */
         { .key         = "cluster.stripe-block-size",