Add a Verify API Server handler that waits for the API server to become

available

playbooks/common/openshift-node/config.yml

@@ -215,6 +215,15 @@
                          | oo_collect('openshift.common.hostname') }}"
     openshift_node_vars: "{{ hostvars | oo_select_keys(groups['oo_nodes_to_config']) }}"
   pre_tasks:
-
+  # Necessary because when you're on a node that's also a master the master will be
+  # restarted after the node restarts docker and it will take up to 60 seconds for
+  # systemd to start the master again
+  - name: Wait for master to become available before proceeding
+    wait_for:
+      host: "{{ hostvars[groups.oo_first_master.0].openshift.common.ip }}"
+      port: "{{ hostvars[groups.oo_first_master.0].openshift.master.api_port }}"
+      state: started
+      timeout: 180
+    when: openshift.common.is_containerized | bool
   roles:
   - openshift_manage_node

roles/openshift_master/handlers/main.yml

@@ -2,11 +2,20 @@
 - name: restart master
   service: name={{ openshift.common.service_type }}-master state=restarted
   when: (not openshift_master_ha | bool) and (not (master_service_status_changed | default(false) | bool))
+  notify: Verify API Server
 
 - name: restart master api
   service: name={{ openshift.common.service_type }}-master-api state=restarted
   when: (openshift_master_ha | bool) and (not (master_api_service_status_changed | default(false) | bool)) and openshift.master.cluster_method == 'native'
+  notify: Verify API Server
 
 - name: restart master controllers
   service: name={{ openshift.common.service_type }}-master-controllers state=restarted
   when: (openshift_master_ha | bool) and (not (master_controllers_service_status_changed | default(false) | bool)) and openshift.master.cluster_method == 'native'
+
+- name: Verify API Server
+  wait_for:
+    host: "{{ openshift.common.ip }}"
+    port: "{{ openshift.master.api_port }}"
+    state: started
+    timeout: 180

roles/openshift_master/tasks/main.yml

@@ -269,6 +269,7 @@
   service: name={{ openshift.common.service_type }}-master enabled=yes state=started
   when: not openshift_master_ha | bool
   register: start_result
+  notify: Verify API Server
 
 - name: Stop and disable non HA master when running HA
   service: name={{ openshift.common.service_type }}-master enabled=no state=stopped
@@ -282,6 +283,7 @@
   service: name={{ openshift.common.service_type }}-master-api enabled=yes state=started
   when: openshift_master_ha | bool and openshift.master.cluster_method == 'native'
   register: start_result
+  notify: Verify API Server
 
 - set_fact:
     master_api_service_status_changed: "{{ start_result | changed }}"