afr: don't treat all cases all bricks being blamed as split-brain

Problem:
We currently don't have a roll-back/undoing of post-ops if quorum is not
met. Though the FOP is still unwound with failure, the xattrs remain on
the disk.  Due to these partial post-ops and partial heals (healing only when
2 bricks are up), we can end up in split-brain purely from the afr
xattrs point of view i.e each brick is blamed by atleast one of the
others. These scenarios are hit when there is frequent
connect/disconnect of the client/shd to the bricks while I/O or heal
are in progress.

Fix:
Instead of undoing the post-op, pick a source based on the xattr values.
If 2 bricks blame one, the blamed one must be treated as sink.
If there is no majority, all are sources. Once we pick a source,
self-heal will then do the heal instead of erroring out due to
split-brain.

Change-Id: I3d0224b883eb0945785ade0e9697a1c828aec0ae
BUG: 1539358
Signed-off-by: Ravishankar N <ravishankar@redhat.com>

2 files changed, 48 insertions, 9 deletions

diff --git a/xlators/cluster/afr/src/afr-self-heal-common.c b/xlators/cluster/afr/src/afr-self-heal-common.c
index 311da68e8ee..7e2a6027126 100644
--- a/xlators/cluster/afr/src/afr-self-heal-common.c
+++ b/xlators/cluster/afr/src/afr-self-heal-common.c
@@ -1455,6 +1455,36 @@ afr_does_witness_exist (xlator_t *this, uint64_t *witness)
         return _gf_false;
 }
 
+unsigned int
+afr_get_quorum_count (afr_private_t *priv)
+{
+        if (priv->quorum_count == AFR_QUORUM_AUTO) {
+                return priv->child_count/2 + 1;
+        } else {
+                return priv->quorum_count;
+        }
+}
+
+void
+afr_selfheal_post_op_failure_accounting (afr_private_t *priv, char *accused,
+                                         unsigned char *sources,
+                                         unsigned char *locked_on)
+{
+        int i = 0;
+        unsigned int  quorum_count = 0;
+
+        if (AFR_COUNT (sources, priv->child_count) != 0)
+                return;
+
+        quorum_count = afr_get_quorum_count (priv);
+        for (i = 0; i < priv->child_count; i++) {
+                if ((accused[i] < quorum_count) && locked_on[i]) {
+                        sources[i] = 1;
+                }
+        }
+        return;
+}
+
 /*
  * This function determines if a self-heal is required for a given inode,
  * and if needed, in what direction.
@@ -1490,6 +1520,7 @@ afr_selfheal_find_direction (call_frame_t *frame, xlator_t *this,
         char *accused = NULL;/* Accused others without any self-accusal */
         char *pending = NULL;/* Have pending operations on others */
         char *self_accused = NULL; /* Accused itself */
+        int min_participants = -1;
 
 	priv = this->private;
 
@@ -1513,8 +1544,13 @@ afr_selfheal_find_direction (call_frame_t *frame, xlator_t *this,
                 }
         }
 
+        if (type == AFR_DATA_TRANSACTION) {
+                min_participants = priv->child_count;
+        } else {
+                min_participants = AFR_SH_MIN_PARTICIPANTS;
+        }
         if (afr_success_count (replies,
-                               priv->child_count) < AFR_SH_MIN_PARTICIPANTS) {
+                               priv->child_count) < min_participants) {
                 /* Treat this just like locks not being acquired */
                 return -ENOTCONN;
         }
@@ -1530,11 +1566,10 @@ afr_selfheal_find_direction (call_frame_t *frame, xlator_t *this,
 	for (i = 0; i < priv->child_count; i++) {
 		for (j = 0; j < priv->child_count; j++) {
                         if (matrix[i][j]) {
-                                 if (!self_accused[i])
-                                         accused[j] = 1;
-
-                                 if (i != j)
-                                         pending[i] = 1;
+                                if (!self_accused[i])
+                                        accused[j] += 1;
+                                if (i != j)
+                                        pending[i] += 1;
                          }
 		}
 	}
@@ -1575,6 +1610,10 @@ afr_selfheal_find_direction (call_frame_t *frame, xlator_t *this,
                 }
         }
 
+        if (type == AFR_DATA_TRANSACTION)
+                afr_selfheal_post_op_failure_accounting (priv, accused,
+                                                         sources, locked_on);
+
          /* If no sources, all locked nodes are sinks - split brain */
          if (AFR_COUNT (sources, priv->child_count) == 0) {
                 for (i = 0; i < priv->child_count; i++) {
diff --git a/xlators/cluster/afr/src/afr-self-heal-data.c b/xlators/cluster/afr/src/afr-self-heal-data.c
index 7880a611592..c033a8afc07 100644
--- a/xlators/cluster/afr/src/afr-self-heal-data.c
+++ b/xlators/cluster/afr/src/afr-self-heal-data.c
@@ -684,7 +684,7 @@ __afr_selfheal_data (call_frame_t *frame, xlator_t *this, fd_t *fd,
 	ret = afr_selfheal_inodelk (frame, this, fd->inode, this->name, 0, 0,
 				    data_lock);
 	{
-		if (ret < AFR_SH_MIN_PARTICIPANTS) {
+		if (ret < priv->child_count) {
                         gf_msg_debug (this->name, 0, "%s: Skipping "
                                       "self-heal as only %d number "
                                       "of subvolumes "
@@ -749,7 +749,7 @@ restore_time:
         if (!is_arbiter_the_only_sink) {
                 ret = afr_selfheal_inodelk (frame, this, fd->inode, this->name,
                                             0, 0, data_lock);
-                if (ret < AFR_SH_MIN_PARTICIPANTS) {
+                if (ret < priv->child_count) {
                         ret = -ENOTCONN;
                         did_sh = _gf_false;
                         goto skip_undo_pending;
@@ -878,7 +878,7 @@ afr_selfheal_data (call_frame_t *frame, xlator_t *this, inode_t *inode)
 	                                        priv->sh_domain, 0, 0,
 				                locked_on);
 	{
-		if (ret < AFR_SH_MIN_PARTICIPANTS) {
+		if (ret < priv->child_count) {
                         gf_msg_debug (this->name, 0, "%s: Skipping "
                                       "self-heal as only %d number of "
                                       "subvolumes could be locked",