arbiter: write performance improvement

Problem: The throughput for a 'dd' workload was much less for arbiter
configuration when compared to normal replica-3 volume. There were 2
issues:

i)arbiter_writev was using the request dict as response dict while
unwinding, leading to incorect GLUSTERFS_WRITE_IS_APPEND and
GLUSTERFS_OPEN_FD_COUNT values (=4), leading to immediate post-ops
because is_afr_delayed_changelog_post_op_needed() failed due to
afr_are_multiple_fds_opened() check.

ii) The arbiter code in afr was setting local->transaction.{start and len} =0
to take full file locks. What this meant was even for simultaenous but
non-overlapping writevs, afr_transaction_eager_lock_init() was not
happening because afr_locals_overlap() always stays true. Consequently
is_afr_delayed_changelog_post_op_needed() failed due to
local->delayed_post_op not being set.

Fix:
i) Send appropriate response dict values in arbiter_writev.
ii) Modify flock params instead of local->transaction.{start and len} to
take full file locks in the transaction.

Also changed _fill_writev_xdata() in posix to fill rsp_xdata for
whatever key is requested for.

Change-Id: I1c5fc5e98aba49ade540bb441a022e65b753432a
BUG: 1324004
Signed-off-by: Ravishankar N <ravishankar@redhat.com>
Reported-by: Robert Rauch <robert.rauch@gns-systems.de>
Reported-by: Russel Purinton <russell.purinton@gmail.com>
Reviewed-on: http://review.gluster.org/13906
Reviewed-by: Pranith Kumar Karampuri <pkarampu@redhat.com>
Smoke: Gluster Build System <jenkins@build.gluster.com>
CentOS-regression: Gluster Build System <jenkins@build.gluster.com>
NetBSD-regression: NetBSD Build System <jenkins@build.gluster.org>

2 files changed, 12 insertions, 12 deletions

diff --git a/xlators/cluster/afr/src/afr-inode-write.c b/xlators/cluster/afr/src/afr-inode-write.c
index e0928425c6a..15bae87a4f4 100644
--- a/xlators/cluster/afr/src/afr-inode-write.c
+++ b/xlators/cluster/afr/src/afr-inode-write.c
@@ -412,7 +412,6 @@ afr_do_writev (call_frame_t *frame, xlator_t *this)
 {
         call_frame_t    *transaction_frame = NULL;
         afr_local_t     *local             = NULL;
-        afr_private_t   *priv              = NULL;
         int             ret   = -1;
         int             op_errno = ENOMEM;
 
@@ -421,7 +420,6 @@ afr_do_writev (call_frame_t *frame, xlator_t *this)
                 goto out;
 
         local = frame->local;
-        priv = this->private;
         transaction_frame->local = local;
 	frame->local = NULL;
 
@@ -451,12 +449,6 @@ afr_do_writev (call_frame_t *frame, xlator_t *this)
                 local->transaction.start   = local->cont.writev.offset;
                 local->transaction.len     = iov_length (local->cont.writev.vector,
                                                          local->cont.writev.count);
-
-                /*Lock entire file to avoid network split brains.*/
-                if (priv->arbiter_count == 1) {
-                        local->transaction.start   = 0;
-                        local->transaction.len     = 0;
-                }
         }
 
         ret = afr_transaction (transaction_frame, this, AFR_DATA_TRANSACTION);
diff --git a/xlators/cluster/afr/src/afr-transaction.c b/xlators/cluster/afr/src/afr-transaction.c
index 92f68b91113..9c27f07b5d4 100644
--- a/xlators/cluster/afr/src/afr-transaction.c
+++ b/xlators/cluster/afr/src/afr-transaction.c
@@ -1356,16 +1356,24 @@ afr_post_lower_unlock_cbk (call_frame_t *frame, xlator_t *this)
 
 
 int
-afr_set_transaction_flock (afr_local_t *local)
+afr_set_transaction_flock (xlator_t *this, afr_local_t *local)
 {
         afr_internal_lock_t *int_lock = NULL;
         afr_inodelk_t       *inodelk  = NULL;
+        afr_private_t       *priv     = NULL;
 
         int_lock = &local->internal_lock;
         inodelk = afr_get_inodelk (int_lock, int_lock->domain);
+        priv = this->private;
 
-        inodelk->flock.l_len   = local->transaction.len;
-        inodelk->flock.l_start = local->transaction.start;
+        if (priv->arbiter_count) {
+                /*Lock entire file to avoid network split brains.*/
+                inodelk->flock.l_len   = 0;
+                inodelk->flock.l_start = 0;
+        } else {
+                inodelk->flock.l_len   = local->transaction.len;
+                inodelk->flock.l_start = local->transaction.start;
+        }
         inodelk->flock.l_type  = F_WRLCK;
 
         return 0;
@@ -1386,7 +1394,7 @@ afr_lock_rec (call_frame_t *frame, xlator_t *this)
         switch (local->transaction.type) {
         case AFR_DATA_TRANSACTION:
         case AFR_METADATA_TRANSACTION:
-                afr_set_transaction_flock (local);
+                afr_set_transaction_flock (this, local);
 
                 int_lock->lock_cbk = afr_post_nonblocking_inodelk_cbk;