afr: allow readdir to proceed for directories in split-brain

Problem:
afr_read_txn() bails out if read_subvol==-1. This meant that for
directories that were in entry split-brain, FOPS like readdir, access,
stat etc were not allowed.

Fix:
Except for getxattr, all other FOPS are wound on the first up child
of afr.

Change-Id: Iacec8fbb1e75c4d2094baa304f62331c81a6f670
BUG: 1221481
Signed-off-by: Ravishankar N <ravishankar@redhat.com>
Reviewed-on: http://review.gluster.org/10776
Reviewed-by: Pranith Kumar Karampuri <pkarampu@redhat.com>
Reviewed-by: Anuradha Talur <atalur@redhat.com>
Tested-by: NetBSD Build System

2 files changed, 59 insertions, 18 deletions

diff --git a/tests/bugs/replicate/bug-1221481-allow-fops-on-dir-split-brain.t b/tests/bugs/replicate/bug-1221481-allow-fops-on-dir-split-brain.t
new file mode 100644
index 00000000000..cd4a9bad87b
--- /dev/null
+++ b/tests/bugs/replicate/bug-1221481-allow-fops-on-dir-split-brain.t
@@ -0,0 +1,37 @@
+#!/bin/bash
+. $(dirname $0)/../../include.rc
+. $(dirname $0)/../../volume.rc
+. $(dirname $0)/../../afr.rc
+cleanup;
+
+#Allow readdirs to proceed on directories that are in split-brain
+
+TEST glusterd;
+TEST pidof glusterd;
+TEST $CLI volume create $V0 replica 2 $H0:$B0/${V0}{0,1};
+TEST $CLI volume set $V0 cluster.self-heal-daemon off
+TEST $CLI volume start $V0;
+TEST glusterfs --volfile-id=/$V0 --volfile-server=$H0 $M0 --attribute-timeout=0 --entry-timeout=0
+TEST mkdir $M0/dir
+TEST touch $M0/dir/file{1..5}
+
+#Create metadata and entry split-brain
+TEST kill_brick $V0 $H0 $B0/$V0"1"
+TEST setfattr -n user.attribute -v value1 $M0/dir
+TEST touch $M0/dir/FILE
+TEST $CLI volume start $V0 force
+EXPECT_WITHIN $CHILD_UP_TIMEOUT '1' afr_child_up_status_meta $M0 $V0-replicate-0 1
+TEST kill_brick $V0 $H0 $B0/$V0"0"
+TEST setfattr -n user.attribute -v value2 $M0/dir
+TEST touch $M0/dir/FILE
+TEST $CLI volume start $V0 force
+EXPECT_WITHIN $CHILD_UP_TIMEOUT '1' afr_child_up_status_meta $M0 $V0-replicate-0 0
+
+TEST ! getfattr $M0/dir
+cd $M0/dir
+EXPECT "6" echo $(ls | wc -l)
+TEST ! cat FILE
+TEST `echo hello>hello.txt`
+cd -
+TEST umount $M0
+cleanup
diff --git a/xlators/cluster/afr/src/afr-read-txn.c b/xlators/cluster/afr/src/afr-read-txn.c
index eaa73d9be20..0ec1d912675 100644
--- a/xlators/cluster/afr/src/afr-read-txn.c
+++ b/xlators/cluster/afr/src/afr-read-txn.c
@@ -47,11 +47,19 @@ afr_read_txn_next_subvol (call_frame_t *frame, xlator_t *this)
 	return 0;
 }
 
+#define AFR_READ_TXN_SET_ERROR_AND_GOTO(ret, errnum, index, label) \
+        do {                                                      \
+                local->op_ret = ret;                              \
+                local->op_errno = errnum;                          \
+                read_subvol = index;                              \
+                goto label;                                       \
+        } while (0)
 
 int
 afr_read_txn_refresh_done (call_frame_t *frame, xlator_t *this, int err)
 {
 	afr_local_t *local = NULL;
+        afr_private_t *priv = NULL;
 	int read_subvol = 0;
 	int event_generation = 0;
 	inode_t *inode = NULL;
@@ -60,35 +68,31 @@ afr_read_txn_refresh_done (call_frame_t *frame, xlator_t *this, int err)
 
 	local = frame->local;
 	inode = local->inode;
+        priv  = frame->this->private;
 
-	if (err) {
-		local->op_errno = -err;
-		local->op_ret = -1;
-		read_subvol = -1;
-		goto readfn;
-	}
+	if (err)
+                AFR_READ_TXN_SET_ERROR_AND_GOTO (-1, -err, -1, readfn);
 
 	ret = afr_inode_read_subvol_type_get (inode, this, local->readable,
 					      &event_generation,
 					      local->transaction.type);
 
-	if (ret == -1 || !event_generation) {
+	if (ret == -1 || !event_generation)
 		/* Even after refresh, we don't have a good
 		   read subvolume. Time to bail */
-		local->op_ret = -1;
-		local->op_errno = EIO;
-		read_subvol = -1;
-		goto readfn;
-	}
+                AFR_READ_TXN_SET_ERROR_AND_GOTO (-1, EIO, -1, readfn);
+
+         /* For directories in split-brain, we need to allow all fops
+          * except (f)getxattr and access. */
+        if (!AFR_COUNT(local->readable, priv->child_count) &&
+            local->transaction.type == AFR_DATA_TRANSACTION &&
+            inode->ia_type == IA_IFDIR)
+                memcpy (local->readable, local->child_up, priv->child_count);
 
 	read_subvol = afr_read_subvol_select_by_policy (inode, this,
 							local->readable);
-
-	if (read_subvol == -1) {
-		local->op_ret = -1;
-		local->op_errno = EIO;
-		goto readfn;
-	}
+	if (read_subvol == -1)
+                AFR_READ_TXN_SET_ERROR_AND_GOTO (-1, EIO, -1, readfn);
 
 	if (local->read_attempted[read_subvol]) {
 		afr_read_txn_next_subvol (frame, this);