cli/rebalance: remove brick status is incorrect

If a remove brick operation is preceded by a fix-layout,
running remove-brick status on a node which does not
contain any of the bricks that were removed displays
fix-layout status.

The defrag_cmd variable was not updated in glusterd
for the nodes not hosting removed bricks causing the
status parsing to go wrong. This is now updated.
Also made minor modifications to the spacing in
the fix-layout status output.

Change-Id: Ib735ce26be7434cd71b76e4c33d9b0648d0530db
BUG: 1389697
Signed-off-by: N Balachandran <nbalacha@redhat.com>
Reviewed-on: http://review.gluster.org/15749
Smoke: Gluster Build System <jenkins@build.gluster.org>
NetBSD-regression: NetBSD Build System <jenkins@build.gluster.org>
CentOS-regression: Gluster Build System <jenkins@build.gluster.org>
Reviewed-by: Atin Mukherjee <amukherj@redhat.com>

3 files changed, 61 insertions, 5 deletions

diff --git a/cli/src/cli-rpc-ops.c b/cli/src/cli-rpc-ops.c
index f217e13abdd..1f641f4a2fd 100644
--- a/cli/src/cli-rpc-ops.c
+++ b/cli/src/cli-rpc-ops.c
@@ -1609,6 +1609,7 @@ gf_cli_print_rebalance_status (dict_t *dict, enum gf_task_types task_type,
         int                min          = 0;
         int                sec          = 0;
         gf_boolean_t       down         = _gf_false;
+	gf_boolean_t       fix_layout   = _gf_false;
 
         ret = dict_get_int32 (dict, "count", &count);
         if (ret) {
@@ -1626,10 +1627,20 @@ gf_cli_print_rebalance_status (dict_t *dict, enum gf_task_types task_type,
                 goto out;
         }
 
-        if (status_rcd >= GF_DEFRAG_STATUS_LAYOUT_FIX_STARTED) {
-                cli_out ("%10s %40s %18s", "Node", "status",
+        /* Fix layout will be sent to all nodes for the volume
+           so every status should be of type
+           GF_DEFRAG_STATUS_LAYOUT_FIX*
+        */
+
+        if ((task_type == GF_TASK_TYPE_REBALANCE)
+           && (status_rcd >= GF_DEFRAG_STATUS_LAYOUT_FIX_STARTED)) {
+                fix_layout = _gf_true;
+        }
+
+        if (fix_layout) {
+                cli_out ("%20s %40s %18s", "Node", "status",
                          "run time in h:m:s");
-                cli_out ("%10s %40s %18s", "---------", "-----------",
+                cli_out ("%20s %40s %18s", "---------", "-----------",
                          "------------");
         } else {
                 cli_out ("%40s %16s %13s %13s %13s %13s %20s %18s",
@@ -1738,8 +1749,8 @@ gf_cli_print_rebalance_status (dict_t *dict, enum gf_task_types task_type,
                 min = ((int) elapsed % 3600) / 60;
                 sec = ((int) elapsed % 3600) % 60;
 
-                if (status_rcd >= GF_DEFRAG_STATUS_LAYOUT_FIX_STARTED) {
-                        cli_out ("%10s %40s %8d:%d:%d", node_name, status_str,
+                if (fix_layout) {
+                        cli_out ("%20s %40s %8d:%d:%d", node_name, status_str,
                                  hrs, min, sec);
                 } else {
                         if (size_str) {
diff --git a/tests/bugs/distribute/bug-1389697.t b/tests/bugs/distribute/bug-1389697.t
new file mode 100644
index 00000000000..0d428b8d9d2
--- /dev/null
+++ b/tests/bugs/distribute/bug-1389697.t
@@ -0,0 +1,42 @@
+#!/bin/bash
+
+. $(dirname $0)/../../include.rc
+. $(dirname $0)/../../cluster.rc
+
+
+cleanup
+
+TEST launch_cluster 2;
+TEST $CLI_1 peer probe $H2;
+EXPECT_WITHIN $PROBE_TIMEOUT 1 peer_count
+
+TEST $CLI_1 volume create $V0 $H1:$B1/b1 $H1:$B1/b2 $H2:$B2/b3
+TEST $CLI_1 volume start $V0
+
+#Start a fix-layout
+TEST $CLI_1 volume rebalance $V0 fix-layout start
+
+#volume rebalance status should work
+TEST $CLI_1 volume rebalance $V0 status
+$CLI_1 volume rebalance $V0 status
+
+val=$($CLI_1 volume rebalance $V0 status |grep "fix-layout" 2>&1)
+val=$?
+TEST [ $val -eq 0 ];
+
+#Start a remove brick for the brick on H2
+TEST $CLI_1 volume remove-brick $V0 $H2:$B2/b3 start
+TEST $CLI_1 volume remove-brick $V0 $H2:$B2/b3 status
+
+#Check remove brick status from H1
+$CLI_1 volume remove-brick $V0 $H2:$B2/b3 status |grep "fix-layout" 2>&1
+val=$?
+TEST [ $val -eq 1 ];
+
+$CLI_1 volume remove-brick $V0 $H2:$B2/b3 status
+$CLI_2 volume remove-brick $V0 $H2:$B2/b3 status
+
+
+TEST $CLI_1 volume remove-brick $V0 $H2:$B2/b3 stop
+
+cleanup
diff --git a/xlators/mgmt/glusterd/src/glusterd-brick-ops.c b/xlators/mgmt/glusterd/src/glusterd-brick-ops.c
index 276c6df0aee..7bed110dce9 100644
--- a/xlators/mgmt/glusterd/src/glusterd-brick-ops.c
+++ b/xlators/mgmt/glusterd/src/glusterd-brick-ops.c
@@ -2652,7 +2652,10 @@ glusterd_op_remove_brick (dict_t *dict, char **op_errstr)
                 /* Reset defrag status to 'NOT STARTED' whenever a
                  * remove-brick/rebalance command is issued to remove
                  * stale information from previous run.
+                 * Update defrag_cmd as well or it will only be done
+                 * for nodes on which the brick to be removed exists.
                  */
+                volinfo->rebal.defrag_cmd = cmd;
                 volinfo->rebal.defrag_status = GF_DEFRAG_STATUS_NOT_STARTED;
                 ret = dict_get_str (dict, GF_REMOVE_BRICK_TID_KEY, &task_id_str);
                 if (ret) {