cluster/afr: Adding option to take full file lock

Problem:
In replica 3 volumes there is a possibilities of ending up in split
brain scenario, when multiple clients writing data on the same file
at non overlapping regions in parallel.

Scenario:
- Initially all the copies are good and all the clients gets the value
  of data readables as all good.
- Client C0 performs write W1 which fails on brick B0 and succeeds on
  other two bricks.
- C1 performs write W2 which fails on B1 and succeeds on other two bricks.
- C2 performs write W3 which fails on B2 and succeeds on other two bricks.
- All the 3 writes above happen in parallel and fall on different ranges
  so afr takes granular locks and all the writes are performed in parallel.
  Since each client had data-readables as good, it does not see
  file going into split-brain in the in_flight_split_brain check, hence
  performs the post-op marking the pending xattrs. Now all the bricks
  are being blamed by each other, ending up in split-brain.

Fix:
Have an option to take either full lock or range lock on files while
doing data transactions, to prevent the possibility of ending up in
split brains. With this change, by default the files will take full
lock while doing IO. If you want to make use of the old range lock
change the value of "cluster.full-lock" to "no".

Change-Id: I7893fa33005328ed63daa2f7c35eeed7c5218962
BUG: 1535438
Signed-off-by: karthik-us <ksubrahm@redhat.com>

5 files changed, 25 insertions, 4 deletions

diff --git a/libglusterfs/src/globals.h b/libglusterfs/src/globals.h
index bbc87f502dd..9231fbc1380 100644
--- a/libglusterfs/src/globals.h
+++ b/libglusterfs/src/globals.h
@@ -38,7 +38,7 @@
  */
 #define GD_OP_VERSION_MIN  1 /* MIN is the fresh start op-version, mostly
                                 should not change */
-#define GD_OP_VERSION_MAX  GD_OP_VERSION_3_13_1 /* MAX VERSION is the maximum
+#define GD_OP_VERSION_MAX  GD_OP_VERSION_3_13_2 /* MAX VERSION is the maximum
                                                   count in VME table, should
                                                   keep changing with
                                                   introduction of newer
@@ -96,6 +96,9 @@
 
 #define GD_OP_VERSION_3_13_1   31301 /* Op-version for GlusterFS 3.13.1 */
 
+#define GD_OP_VERSION_3_13_2   31302 /* Op-version for GlusterFS 3.13.2 */
+
+
 #define GD_OP_VER_PERSISTENT_AFR_XATTRS GD_OP_VERSION_3_6_0
 
 #include "xlator.h"
diff --git a/xlators/cluster/afr/src/afr-transaction.c b/xlators/cluster/afr/src/afr-transaction.c
index 7e40bba5b53..ba621204fdc 100644
--- a/xlators/cluster/afr/src/afr-transaction.c
+++ b/xlators/cluster/afr/src/afr-transaction.c
@@ -1962,7 +1962,7 @@ afr_set_transaction_flock (xlator_t *this, afr_local_t *local)
         inodelk = afr_get_inodelk (int_lock, int_lock->domain);
         priv = this->private;
 
-        if (priv->arbiter_count &&
+        if ((priv->arbiter_count || priv->full_lock) &&
             local->transaction.type == AFR_DATA_TRANSACTION) {
                 /*Lock entire file to avoid network split brains.*/
                 inodelk->flock.l_len   = 0;
diff --git a/xlators/cluster/afr/src/afr.c b/xlators/cluster/afr/src/afr.c
index b18f60fcc7c..43389390730 100644
--- a/xlators/cluster/afr/src/afr.c
+++ b/xlators/cluster/afr/src/afr.c
@@ -244,6 +244,7 @@ reconfigure (xlator_t *this, dict_t *options)
                           out);
         GF_OPTION_RECONF ("locking-scheme", priv->locking_scheme, options, str,
                           out);
+        GF_OPTION_RECONF ("full-lock", priv->full_lock, options, bool, out);
         GF_OPTION_RECONF ("use-compound-fops", priv->use_compound_fops,
                           options, bool,
                           out);
@@ -532,6 +533,7 @@ init (xlator_t *this)
 
         GF_OPTION_INIT ("pre-op-compat", priv->pre_op_compat, bool, out);
         GF_OPTION_INIT ("locking-scheme", priv->locking_scheme, str, out);
+        GF_OPTION_INIT ("full-lock", priv->full_lock, bool, out);
         GF_OPTION_INIT ("use-compound-fops", priv->use_compound_fops,
                         bool, out);
         GF_OPTION_INIT ("granular-entry-heal", priv->esh_granular, bool, out);
@@ -1088,6 +1090,15 @@ struct volume_options options[] = {
                          "stop being compatible with afr-v1, which helps afr "
                          "be more granular while self-healing",
         },
+        { .key = {"full-lock"},
+          .type = GF_OPTION_TYPE_BOOL,
+          .default_value = "yes",
+          .op_version = {GD_OP_VERSION_3_13_2},
+          .flags = OPT_FLAG_CLIENT_OPT | OPT_FLAG_SETTABLE,
+          .tags = {"replicate"},
+          .description = "If this option is disabled, then the IOs will take "
+                         "range locks same as versions till 3.13.1."
+        },
         { .key = {"granular-entry-heal"},
           .type = GF_OPTION_TYPE_BOOL,
           .default_value = "no",
diff --git a/xlators/cluster/afr/src/afr.h b/xlators/cluster/afr/src/afr.h
index 18bf249fa34..dbc9e3cc554 100644
--- a/xlators/cluster/afr/src/afr.h
+++ b/xlators/cluster/afr/src/afr.h
@@ -177,9 +177,10 @@ typedef struct _afr_private {
 	void                   *pump_private;
 	gf_boolean_t           use_afr_in_pump;
 	char                   *locking_scheme;
-        gf_boolean_t            esh_granular;
+        gf_boolean_t           full_lock;
+        gf_boolean_t           esh_granular;
         gf_boolean_t           consistent_io;
-        gf_boolean_t            use_compound_fops;
+        gf_boolean_t           use_compound_fops;
 } afr_private_t;
 
 
diff --git a/xlators/mgmt/glusterd/src/glusterd-volume-set.c b/xlators/mgmt/glusterd/src/glusterd-volume-set.c
index edb31b685c4..9de9cc560fb 100644
--- a/xlators/mgmt/glusterd/src/glusterd-volume-set.c
+++ b/xlators/mgmt/glusterd/src/glusterd-volume-set.c
@@ -1506,6 +1506,12 @@ struct volopt_map_entry glusterd_volopt_map[] = {
           .op_version = GD_OP_VERSION_3_7_12,
           .flags      = VOLOPT_FLAG_CLIENT_OPT
         },
+        { .key        = "cluster.full-lock",
+          .voltype    = "cluster/replicate",
+          .type       = NO_DOC,
+          .op_version = GD_OP_VERSION_3_13_2,
+          .flags      = VOLOPT_FLAG_CLIENT_OPT
+        },
 
         /* stripe xlator options */
         { .key         = "cluster.stripe-block-size",