Merge "Increase default RabbitMQ/Erlang TCP timeout from 5 to 15 seconds"

[apex-tripleo-heat-templates.git] / extraconfig / tasks / pacemaker_resource_restart.sh

diff --git a/extraconfig/tasks/pacemaker_resource_restart.sh b/extraconfig/tasks/pacemaker_resource_restart.sh
index 1637cee..49d39bc 100755 (executable)
--- a/extraconfig/tasks/pacemaker_resource_restart.sh
+++ b/extraconfig/tasks/pacemaker_resource_restart.sh
@@ -2,20 +2,32 @@
 
 set -eux
 
-pacemaker_status=$(systemctl is-active pacemaker)
-
 # Run if pacemaker is running, we're the bootstrap node,
 # and we're updating the deployment (not creating).
-if [ "$pacemaker_status" = "active" -a \
-     "$(hiera bootstrap_nodeid)" = "$(facter hostname)" -a \
-     "$(hiera stack_action)" = "UPDATE" ]; then
 
-    PCMK_RESOURCES="haproxy-clone redis-master rabbitmq-clone galera-master openstack-cinder-volume openstack-cinder-backup"
-    # Ten minutes of timeout to restart each resource, given there are no constraints should be enough
+RESTART_FOLDER="/var/lib/tripleo/pacemaker-restarts"
+
+if [[ -d "$RESTART_FOLDER" && -n $(pcmk_running) && -n $(is_bootstrap_node) ]]; then
+
     TIMEOUT=600
-    for resource in $PCMK_RESOURCES; do
-      if pcs status | grep $resource; then
-        pcs resource restart --wait=$TIMEOUT $resource
-      fi
+    PCS_STATUS_OUTPUT="$(pcs status)"
+    SERVICES_TO_RESTART="$(ls $RESTART_FOLDER)"
+
+    for service in $SERVICES_TO_RESTART; do
+        if ! echo "$PCS_STATUS_OUTPUT" | grep $service; then
+            echo "Service $service not found as a pacemaker resource, cannot restart it."
+            exit 1
+        fi
     done
+
+    for service in $SERVICES_TO_RESTART; do
+        echo "Restarting $service..."
+        pcs resource restart --wait=$TIMEOUT $service
+        rm -f "$RESTART_FOLDER"/$service
+    done
+
+fi
+
+if [ $(systemctl is-active haproxy) = "active" ]; then
+    systemctl reload haproxy
 fi