From e1f780ca8f17b8b9615356796093dc7a75f5ff8a Mon Sep 17 00:00:00 2001
From: Atin Mukherjee <amukherj@redhat.com>
Date: Thu, 9 Nov 2017 22:42:22 +0530
Subject: tests: fix bug-1483058-replace-brick-quorum-validation.t spurious
 failure

>mainline patch : https://review.gluster.org/#/c/18710/

Change-Id: I04c35305bfb663eabbf715eee78695adfd4a2d20
BUG: 1512435
Signed-off-by: Atin Mukherjee <amukherj@redhat.com>
(cherry picked from commit 76a83f98b78a0bdf29bbb0f8e4c9ab74dae52be4)
---
 .../glusterd/bug-1483058-replace-brick-quorum-validation.t     |  9 ++++++++-
 tests/cluster.rc                                               | 10 ++++++++++
 tests/include.rc                                               |  1 +
 3 files changed, 19 insertions(+), 1 deletion(-)

diff --git a/tests/bugs/glusterd/bug-1483058-replace-brick-quorum-validation.t b/tests/bugs/glusterd/bug-1483058-replace-brick-quorum-validation.t
index 3dbe28a7917..2d9e5287818 100644
--- a/tests/bugs/glusterd/bug-1483058-replace-brick-quorum-validation.t
+++ b/tests/bugs/glusterd/bug-1483058-replace-brick-quorum-validation.t
@@ -45,7 +45,14 @@ TEST start_glusterd 2
 
 EXPECT_WITHIN $PROBE_TIMEOUT 2 peer_count
 
+# checking peer_count is not enough to call that quorum is regained as
+# peer_count is based on peerinfo->connected where as quorum is calculated based
+# on peerinfo->quorum_contrib. To avoid this spurious race of replace brick
+# commit force to execute and fail before the quorum is regained run the command
+# in EXPECT_WITHIN to ensure that with multiple attempts the command goes
+# through once the quorum is regained.
+
 # Now quorum is met. replace-brick will execute successfuly
-TEST  $CLI_1 volume replace-brick $V0 $H2:$B2/${V0}1 $H1:$B1/${V0}1_new commit force
+EXPECT_WITHIN $PEER_SYNC_TIMEOUT 0 attempt_replace_brick 1 $V0 $H2:$B2/${V0}1 $H1:$B1/${V0}1_new
 
 #cleanup;
diff --git a/tests/cluster.rc b/tests/cluster.rc
index c1ff8ab5b74..e258b586847 100644
--- a/tests/cluster.rc
+++ b/tests/cluster.rc
@@ -142,6 +142,16 @@ function peer_count() {
     $CLI_1 peer status | grep 'Peer in Cluster (Connected)' | wc -l
 }
 
+function attempt_replace_brick {
+    local cli_no=$1
+    local vol=$2;
+    local src_brick=$3;
+    local dst_brick=$4;
+
+    eval \$CLI_$cli_no volume replace-brick $vol $src_brick $dst_brick commit force;
+    echo $?
+}
+
 function cluster_rebalance_status_field {
         #The rebalance status can be up to 3 words, (e.g.:'fix-layout in progress'), hence the awk-print $7 thru $9.
         #But if the status is less than 3 words, it also prints the next field i.e the run_time_in_secs.(e.g.:'completed 3.00').
diff --git a/tests/include.rc b/tests/include.rc
index 7cd15a2d249..f3b94336179 100644
--- a/tests/include.rc
+++ b/tests/include.rc
@@ -74,6 +74,7 @@ PROCESS_UP_TIMEOUT=30
 NFS_EXPORT_TIMEOUT=20
 CHILD_UP_TIMEOUT=20
 PROBE_TIMEOUT=60
+PEER_SYNC_TIMEOUT=20
 REBALANCE_TIMEOUT=360
 REOPEN_TIMEOUT=20
 HEAL_TIMEOUT=80
-- 
cgit