Merge "Support Fenix and sample implementation accordingly"

diff --git a/doctor_tests/admin_tool/__init__.py b/doctor_tests/admin_tool/__init__.py
index e8b1281..3417a33 100644 (file)
--- a/doctor_tests/admin_tool/__init__.py
+++ b/doctor_tests/admin_tool/__init__.py
@@ -8,16 +8,16 @@
 ##############################################################################
 from oslo_config import cfg
 from oslo_utils import importutils
-
+import os
 
 OPTS = [
     cfg.StrOpt('type',
-               default='sample',
-               choices=['sample'],
+               default=os.environ.get('ADMIN_TOOL_TYPE', 'sample'),
+               choices=['sample', 'fenix'],
                help='the component of doctor admin_tool',
                required=True),
     cfg.StrOpt('ip',
-               default='127.0.0.1',
+               default='0.0.0.0',
                help='the ip of admin_tool',
                required=True),
     cfg.IntOpt('port',

diff --git a/doctor_tests/admin_tool/sample.py b/doctor_tests/admin_tool/sample.py
index 892a4c8..a71f43a 100644 (file)
--- a/doctor_tests/admin_tool/sample.py
+++ b/doctor_tests/admin_tool/sample.py
@@ -59,7 +59,7 @@ class AdminMain(Thread):
         self.parent = parent
         self.log = log
         self.conf = conf
-        self.url = 'http://0.0.0.0:%s' % conf.admin_tool.port
+        self.url = 'http://%s:%s' % (conf.admin_tool.ip, conf.admin_tool.port)
         self.projects_state = dict()  # current state for each project
         self.proj_server_actions = dict()  # actions for each project server
         self.projects_servers = dict()  # servers processed in current state
@@ -86,6 +86,7 @@ class AdminMain(Thread):
                                               driver='messaging',
                                               topics=['notifications'])
         self.notif_admin = self.notif_admin.prepare(publisher_id='admin_tool')
+        self.stopped = False
         self.log.info('Admin tool session %s initialized' % self.session_id)
 
     def cleanup(self):
@@ -116,14 +117,15 @@ class AdminMain(Thread):
         if self._projects_not_in_wanted_states(wanted_states):
             self.log.error('Admin tool session %s: projects in invalid states '
                            '%s' % (self.session_id, self.projects_state))
-            raise Exception('Admin tool session %s: not all projects in states'
-                            ' %s' % (self.session_id, wanted_states))
+            return False
         else:
             self.log.info('all projects replied')
+            return True
 
     def _project_notify(self, project_id, instance_ids, allowed_actions,
                         actions_at, state, metadata):
-        reply_url = '%s/%s/maintenance' % (self.url, project_id)
+        reply_url = '%s/maintenance/%s/%s' % (self.url, self.session_id,
+                                              project_id)
 
         payload = dict(project_id=project_id,
                        instance_ids=instance_ids,
@@ -148,11 +150,12 @@ class AdminMain(Thread):
 
         self.notif_admin.info({'some': 'context'}, 'maintenance.host', payload)
 
-    def down_scale(self):
+    def in_scale(self):
         for project in self.projects_servers:
-            self.log.info('DOWN_SCALE to project %s' % project)
+            self.log.info('SCALE_IN to project %s' % project)
             self.log.debug('instance_ids %s' % self.projects_servers[project])
-            instance_ids = '%s/%s/maintenance' % (self.url, project)
+            instance_ids = '%s/maintenance/%s/%s' % (self.url, self.session_id,
+                                                     project)
             allowed_actions = []
             wait_seconds = 120
             actions_at = (datetime.datetime.utcnow() +
@@ -163,18 +166,20 @@ class AdminMain(Thread):
             self._project_notify(project, instance_ids,
                                  allowed_actions, actions_at, state,
                                  metadata)
-        allowed_states = ['ACK_DOWN_SCALE', 'NACK_DOWN_SCALE']
-        self.wait_projects_state(allowed_states, wait_seconds)
-        if self.projects_not_in_state('ACK_DOWN_SCALE'):
-            raise Exception('Admin tool session %s: all states not '
-                            'ACK_DOWN_SCALE %s' %
-                            (self.session_id, self.projects_state))
+        allowed_states = ['ACK_SCALE_IN', 'NACK_SCALE_IN']
+        if not self.wait_projects_state(allowed_states, wait_seconds):
+            self.state = 'MAINTENANCE_FAILED'
+        if self.projects_not_in_state('ACK_SCALE_IN'):
+            self.log.error('%s: all states not ACK_SCALE_IN' %
+                           self.session_id)
+            self.state = 'MAINTENANCE_FAILED'
 
     def maintenance(self):
         for project in self.projects_servers:
             self.log.info('\nMAINTENANCE to project %s\n' % project)
             self.log.debug('instance_ids %s' % self.projects_servers[project])
-            instance_ids = '%s/%s/maintenance' % (self.url, project)
+            instance_ids = '%s/maintenance/%s/%s' % (self.url, self.session_id,
+                                                     project)
             allowed_actions = []
             actions_at = self.maintenance_at
             state = self.state
@@ -190,16 +195,18 @@ class AdminMain(Thread):
                                  allowed_actions, actions_at, state,
                                  metadata)
         allowed_states = ['ACK_MAINTENANCE', 'NACK_MAINTENANCE']
-        self.wait_projects_state(allowed_states, wait_seconds)
+        if not self.wait_projects_state(allowed_states, wait_seconds):
+            self.state = 'MAINTENANCE_FAILED'
         if self.projects_not_in_state('ACK_MAINTENANCE'):
-            raise Exception('Admin tool session %s: all states not '
-                            'ACK_MAINTENANCE %s' %
-                            (self.session_id, self.projects_state))
+            self.log.error('%s: all states not ACK_MAINTENANCE' %
+                           self.session_id)
+            self.state = 'MAINTENANCE_FAILED'
 
     def maintenance_complete(self):
         for project in self.projects_servers:
             self.log.info('MAINTENANCE_COMPLETE to project %s' % project)
-            instance_ids = '%s/%s/maintenance' % (self.url, project)
+            instance_ids = '%s/maintenance/%s/%s' % (self.url, self.session_id,
+                                                     project)
             allowed_actions = []
             wait_seconds = 120
             actions_at = (datetime.datetime.utcnow() +
@@ -212,13 +219,14 @@ class AdminMain(Thread):
                                  metadata)
         allowed_states = ['ACK_MAINTENANCE_COMPLETE',
                           'NACK_MAINTENANCE_COMPLETE']
-        self.wait_projects_state(allowed_states, wait_seconds)
+        if not self.wait_projects_state(allowed_states, wait_seconds):
+            self.state = 'MAINTENANCE_FAILED'
         if self.projects_not_in_state('ACK_MAINTENANCE_COMPLETE'):
-            raise Exception('Admin tool session %s: all states not '
-                            'ACK_MAINTENANCE_COMPLETE %s' %
-                            (self.session_id, self.projects_state))
+            self.log.error('%s: all states not ACK_MAINTENANCE_COMPLETE' %
+                           self.session_id)
+            self.state = 'MAINTENANCE_FAILED'
 
-    def need_down_scale(self, host_servers):
+    def need_in_scale(self, host_servers):
         room_for_instances = 0
         for host in host_servers:
             instances = 0
@@ -267,7 +275,8 @@ class AdminMain(Thread):
             self.projects_servers[project] = projects_servers[project].copy()
             self.log.info('%s to project %s' % (state, project))
             self.project_servers_log_info(project, projects_servers)
-            instance_ids = '%s/%s/maintenance' % (self.url, project)
+            instance_ids = '%s/maintenance/%s/%s' % (self.url, self.session_id,
+                                                     project)
             allowed_actions = ['MIGRATE', 'LIVE_MIGRATE', 'OWN_ACTION']
             wait_seconds = 120
             actions_at = (datetime.datetime.utcnow() +
@@ -278,11 +287,14 @@ class AdminMain(Thread):
                                  allowed_actions, actions_at, state,
                                  metadata)
         allowed_states = [state_ack, state_nack]
-        self.wait_projects_state(allowed_states, wait_seconds)
-        if self.projects_not_in_state(state_ack):
-            raise Exception('Admin tool session %s: all states not %s %s' %
-                            (self.session_id, state_ack, self.projects_state))
-        self.actions_to_have_empty_host(host)
+        if not self.wait_projects_state(allowed_states, wait_seconds):
+            self.state = 'MAINTENANCE_FAILED'
+        elif self.projects_not_in_state(state_ack):
+            self.log.error('%s: all states not %s' %
+                           (self.session_id, state_ack))
+            self.state = 'MAINTENANCE_FAILED'
+        else:
+            self.actions_to_have_empty_host(host)
 
     def notify_action_done(self, project, instance_id):
         instance_ids = instance_id
@@ -463,7 +475,8 @@ class AdminMain(Thread):
         time.sleep(5)
 
     def run(self):
-        while self.state != 'MAINTENANCE_COMPLETE':
+        while (self.state not in ['MAINTENANCE_DONE', 'MAINTENANCE_FAILED'] and
+               not self.stopped):
             self.log.info('--==session %s: processing state %s==--' %
                           (self.session_id, self.state))
             if self.state == 'MAINTENANCE':
@@ -474,7 +487,8 @@ class AdminMain(Thread):
                     raise Exception('all projects do not listen maintenance '
                                     'alarm')
                 self.maintenance()
-
+                if self.state == 'MAINTENANCE_FAILED':
+                    continue
                 maint_at = self.str_to_datetime(self.maintenance_at)
                 if maint_at > datetime.datetime.utcnow():
                     time_now = (datetime.datetime.utcnow().strftime(
@@ -492,14 +506,14 @@ class AdminMain(Thread):
                 # True -> PLANNED_MAINTENANCE
                 # False -> check if we can migrate VMs to get empty host
                 # True -> PREPARE_MAINTENANCE
-                # False -> DOWN_SCALE
+                # False -> SCALE_IN
                 maintenance_empty_hosts = ([h for h in self.hosts if h not in
                                            host_servers])
 
                 if len(maintenance_empty_hosts) == 0:
-                    if self.need_down_scale(host_servers):
+                    if self.need_in_scale(host_servers):
                         self.log.info('Need to down scale')
-                        self.state = 'DOWN_SCALE'
+                        self.state = 'SCALE_IN'
                     else:
                         self.log.info('Free capacity, but need empty host')
                         self.state = 'PREPARE_MAINTENANCE'
@@ -508,14 +522,17 @@ class AdminMain(Thread):
                     self.state = 'PLANNED_MAINTENANCE'
                 self.log.info('--==State change from MAINTENANCE to %s==--'
                               % self.state)
-            elif self.state == 'DOWN_SCALE':
+            elif self.state == 'SCALE_IN':
                 # Test case is hard coded to have all compute capacity used
                 # We need to down scale to have one empty compute host
-                self.down_scale()
+                self.update_server_info()
+                self.in_scale()
+                if self.state == 'MAINTENANCE_FAILED':
+                    continue
                 self.state = 'PREPARE_MAINTENANCE'
                 host_servers = self.update_server_info()
                 self.servers_log_info(host_servers)
-                self.log.info('--==State change from DOWN_SCALE to'
+                self.log.info('--==State change from SCALE_IN to'
                               ' %s==--' % self.state)
 
             elif self.state == 'PREPARE_MAINTENANCE':
@@ -527,7 +544,7 @@ class AdminMain(Thread):
                                            host_servers])
                 if len(maintenance_empty_hosts) == 0:
                     self.log.info('no empty hosts for maintenance')
-                    if self.need_down_scale(host_servers):
+                    if self.need_in_scale(host_servers):
                         raise Exception('Admin tool session %s: Not enough '
                                         'free capacity for maintenance' %
                                         self.session_id)
@@ -535,6 +552,8 @@ class AdminMain(Thread):
                     if host:
                         self.make_compute_host_empty(host, host_servers[host],
                                                      'PREPARE_MAINTENANCE')
+                        if self.state == 'MAINTENANCE_FAILED':
+                            continue
                     else:
                         # We do not currently support another down scale if
                         # first was not enough
@@ -566,6 +585,7 @@ class AdminMain(Thread):
                         maintenance_empty_hosts.append(host)
                 self.log.info('--==Start to maintain empty hosts==--\n%s' %
                               maintenance_empty_hosts)
+                self.update_server_info()
                 for host in maintenance_empty_hosts:
                     # scheduler has problems, let's see if just down scaled
                     # host is really empty
@@ -586,6 +606,8 @@ class AdminMain(Thread):
                     self.log.info('PLANNED_MAINTENANCE host %s' % host)
                     self.make_compute_host_empty(host, host_servers[host],
                                                  'PLANNED_MAINTENANCE')
+                    if self.state == 'MAINTENANCE_FAILED':
+                        continue
                     self.log.info('IN_MAINTENANCE  host %s' % host)
                     self._admin_notify(admin_project, host, 'IN_MAINTENANCE',
                                        self.session_id)
@@ -603,14 +625,16 @@ class AdminMain(Thread):
                 self.log.info('Projects still need to up scale back to full '
                               'capcity')
                 self.maintenance_complete()
+                if self.state == 'MAINTENANCE_FAILED':
+                    continue
                 host_servers = self.update_server_info()
                 self.servers_log_info(host_servers)
-                self.state = 'MAINTENANCE_COMPLETE'
+                self.state = 'MAINTENANCE_DONE'
             else:
                 raise Exception('Admin tool session %s: session in invalid '
                                 'state %s' % (self.session_id, self.state))
-        self.log.info('--==Maintenance session %s: '
-                      'MAINTENANCE SESSION COMPLETE==--' % self.session_id)
+        self.log.info('--==Maintenance session %s: %s==--' %
+                      (self.session_id, self.state))
 
     def project_input(self, project_id, data):
         self.log.debug('Admin tool session %s: project %s input' %
@@ -637,7 +661,6 @@ class AdminTool(Thread):
         self.admin_tool = admin_tool
         self.log = log
         self.conf = conf
-        self.port = self.conf.admin_tool.port
         self.maint_sessions = {}
         self.projects = {}
         self.maintenance_hosts = []
@@ -650,63 +673,55 @@ class AdminTool(Thread):
         def admin_maintenance_api_post():
             data = json.loads(request.data.decode('utf8'))
             self.log.info('maintenance message: %s' % data)
-            if 'session_id' in data:
-                if data['state'] == 'REMOVE_MAINTENANCE_SESSION':
-                    session_id = data['session_id']
-                    self.log.info('remove session %s'
-                                  % session_id)
-                    self.maint_sessions[session_id].cleanup()
-                    self.maint_sessions[session_id].stop()
-                    del self.maint_sessions[session_id]
-            else:
-                session_id = str(generate_uuid())
-                self.log.info('creating session: %s' % session_id)
-                self.maint_sessions[session_id] = (
-                    AdminMain(self.trasport_url,
-                              session_id,
-                              data,
-                              self,
-                              self.conf,
-                              self.log))
-                self.maint_sessions[session_id].start()
+            session_id = str(generate_uuid())
+            self.log.info('creating session: %s' % session_id)
+            self.maint_sessions[session_id] = (
+                AdminMain(self.trasport_url,
+                          session_id,
+                          data,
+                          self,
+                          self.conf,
+                          self.log))
+            self.maint_sessions[session_id].start()
             reply = json.dumps({'session_id': session_id,
                                 'state': 'ACK_%s' % data['state']})
             self.log.debug('reply: %s' % reply)
             return reply, 200, None
 
-        @app.route('/maintenance', methods=['GET'])
-        def admin_maintenance_api_get():
-            data = json.loads(request.data.decode('utf8'))
-            self.log.debug('Admin get maintenance: %s' % data)
-            session_id = data['session_id']
+        @app.route('/maintenance/<session_id>', methods=['GET'])
+        def admin_maintenance_api_get(session_id=None):
+            self.log.debug('Admin get maintenance')
             reply = json.dumps({'state':
                                self.maint_sessions[session_id].state})
-            self.log.debug('reply: %s' % reply)
+            self.log.info('reply: %s' % reply)
             return reply, 200, None
 
-        @app.route('/<projet_id>/maintenance', methods=['PUT'])
-        def project_maintenance_api_put(projet_id=None):
+        @app.route('/maintenance/<session_id>/<projet_id>', methods=['PUT'])
+        def project_maintenance_api_put(session_id=None, projet_id=None):
             data = json.loads(request.data.decode('utf8'))
             self.log.debug('%s project put: %s' % (projet_id, data))
-            self.project_input(projet_id, data)
+            self.project_input(session_id, projet_id, data)
             return 'OK'
 
-        @app.route('/<projet_id>/maintenance', methods=['GET'])
-        def project_maintenance_api_get(projet_id=None):
-            data = json.loads(request.data.decode('utf8'))
-            self.log.debug('%s project get %s' % (projet_id, data))
-            instances = self.project_get_instances(projet_id, data)
+        @app.route('/maintenance/<session_id>/<projet_id>', methods=['GET'])
+        def project_maintenance_api_get(session_id=None, projet_id=None):
+            self.log.debug('%s project get %s' % (projet_id, session_id))
+            instances = self.project_get_instances(session_id, projet_id)
             reply = json.dumps({'instance_ids': instances})
             self.log.debug('%s reply: %s' % (projet_id, reply))
             return reply, 200, None
 
+        @app.route('/maintenance/<session_id>', methods=['DELETE'])
+        def remove_session(session_id=None):
+            self.log.info('remove session %s'
+                          % session_id)
+            self.maint_sessions[session_id].cleanup()
+            self.maint_sessions[session_id].stop()
+            del self.maint_sessions[session_id]
+            return 'OK'
+
         @app.route('/shutdown', methods=['POST'])
         def shutdown():
-            for session in self.maint_sessions:
-                self.log.info('shutdown admin tool session %s thread' %
-                              session)
-                self.maint_sessions[session].cleanup()
-                self.maint_sessions[session].stop()
             self.log.info('shutdown admin_tool server at %s' % time.time())
             func = request.environ.get('werkzeug.server.shutdown')
             if func is None:
@@ -714,13 +729,11 @@ class AdminTool(Thread):
             func()
             return 'admin_tool app shutting down...'
 
-        app.run(host='0.0.0.0', port=self.port)
+        app.run(host=self.conf.admin_tool.ip, port=self.conf.admin_tool.port)
 
-    def project_input(self, project_id, data):
-        session_id = data['session_id']
+    def project_input(self, session_id, project_id, data):
         self.maint_sessions[session_id].project_input(project_id, data)
 
-    def project_get_instances(self, project_id, data):
-        session_id = data['session_id']
+    def project_get_instances(self, session_id, project_id):
         return self.maint_sessions[session_id].project_get_instances(
             project_id)

diff --git a/doctor_tests/app_manager/sample.py b/doctor_tests/app_manager/sample.py
index 94926ee..a7bc412 100644 (file)
--- a/doctor_tests/app_manager/sample.py
+++ b/doctor_tests/app_manager/sample.py
@@ -114,8 +114,7 @@ class AppManager(Thread):
                 for t in data['reason_data']['event']['traits']})
 
     def get_session_instance_ids(self, url, session_id):
-        data = {'session_id': session_id}
-        ret = requests.get(url, data=json.dumps(data), headers=self.headers)
+        ret = requests.get(url, data=None, headers=self.headers)
         if ret.status_code != 200:
             raise Exception(ret.text)
         self.log.info('get_instance_ids %s' % ret.json())
@@ -177,12 +176,12 @@ class AppManager(Thread):
                 reply['instance_ids'] = instance_ids
                 reply_state = 'ACK_MAINTENANCE'
 
-            elif state == 'DOWN_SCALE':
+            elif state == 'SCALE_IN':
                 # scale down 2 isntances that is VCPUS equaling to single
                 # compute node
                 self.scale_instances(-2)
                 reply['instance_ids'] = self.get_instance_ids()
-                reply_state = 'ACK_DOWN_SCALE'
+                reply_state = 'ACK_SCALE_IN'
 
             elif state == 'MAINTENANCE_COMPLETE':
                 # possibly need to upscale

diff --git a/doctor_tests/installer/apex.py b/doctor_tests/installer/apex.py
index 2aa81ff..3c97378 100644 (file)
--- a/doctor_tests/installer/apex.py
+++ b/doctor_tests/installer/apex.py
@@ -114,6 +114,22 @@ class ApexInstaller(BaseInstaller):
     def set_apply_patches(self):
         self.log.info('Set apply patches start......')
 
+        if self.conf.test_case != 'fault_management':
+            if self.use_containers:
+                restart_cmd = self._set_docker_restart_cmd("nova-compute")
+            else:
+                restart_cmd = 'sudo systemctl restart' \
+                              ' openstack-nova-compute.service'
+            for node_ip in self.computes:
+                client = SSHClient(node_ip, self.node_user_name,
+                                   key_filename=self.key_file)
+                self.compute_clients.append(client)
+                self._run_apply_patches(client,
+                                        restart_cmd,
+                                        [self.nc_set_compute_script],
+                                        python=self.python)
+            time.sleep(10)
+
         set_scripts = [self.cm_set_script]
 
         if self.use_containers:
@@ -147,24 +163,6 @@ class ApexInstaller(BaseInstaller):
                                     set_scripts,
                                     python=self.python)
 
-        if self.conf.test_case != 'fault_management':
-            if self.use_containers:
-                restart_cmd = self._set_docker_restart_cmd("nova-compute")
-            else:
-                restart_cmd = 'sudo systemctl restart' \
-                              ' openstack-nova-compute.service'
-            for node_ip in self.computes:
-                client = SSHClient(node_ip, self.node_user_name,
-                                   key_filename=self.key_file)
-                self.compute_clients.append(client)
-                self._run_apply_patches(client,
-                                        restart_cmd,
-                                        [self.nc_set_compute_script],
-                                        python=self.python)
-
-        if self.conf.test_case != 'fault_management':
-            time.sleep(10)
-
     def restore_apply_patches(self):
         self.log.info('restore apply patches start......')
 

diff --git a/doctor_tests/installer/base.py b/doctor_tests/installer/base.py
index 3043593..124b191 100644 (file)
--- a/doctor_tests/installer/base.py
+++ b/doctor_tests/installer/base.py
@@ -14,8 +14,9 @@ import pwd
 import six
 import stat
 import subprocess
+import time
 
-from doctor_tests.common.utils import get_doctor_test_root_dir
+from doctor_tests.common import utils
 from doctor_tests.identity_auth import get_session
 from doctor_tests.os_clients import nova_client
 
@@ -75,7 +76,7 @@ class BaseInstaller(object):
                 cmd = ("ssh -o UserKnownHostsFile=/dev/null"
                        " -o StrictHostKeyChecking=no"
                        " -i %s %s@%s -R %s:localhost:%s"
-                       " sleep %s > ssh_tunnel.%s"
+                       " sleep %s > ssh_tunnel.%s.%s"
                        " 2>&1 < /dev/null "
                        % (self.key_file,
                           self.node_user_name,
@@ -83,9 +84,28 @@ class BaseInstaller(object):
                           port,
                           port,
                           tunnel_uptime,
-                          node_ip))
+                          node_ip,
+                          port))
                 server = subprocess.Popen('exec ' + cmd, shell=True)
                 self.servers.append(server)
+        if self.conf.admin_tool.type == 'fenix':
+            port = self.conf.admin_tool.port
+            self.log.info('tunnel for port %s' % port)
+            cmd = ("ssh -o UserKnownHostsFile=/dev/null"
+                   " -o StrictHostKeyChecking=no"
+                   " -i %s %s@%s -L %s:localhost:%s"
+                   " sleep %s > ssh_tunnel.%s.%s"
+                   " 2>&1 < /dev/null "
+                   % (self.key_file,
+                      self.node_user_name,
+                      node_ip,
+                      port,
+                      port,
+                      tunnel_uptime,
+                      node_ip,
+                      port))
+            server = subprocess.Popen('exec ' + cmd, shell=True)
+            self.servers.append(server)
 
     def _get_ssh_key(self, client, key_path):
         self.log.info('Get SSH keys from %s installer......'
@@ -96,7 +116,8 @@ class BaseInstaller(object):
                           % self.conf.installer.type)
             return self.key_file
 
-        ssh_key = '{0}/{1}'.format(get_doctor_test_root_dir(), 'instack_key')
+        ssh_key = '{0}/{1}'.format(utils.get_doctor_test_root_dir(),
+                                   'instack_key')
         client.scp(key_path, ssh_key, method='get')
         user = getpass.getuser()
         uid = pwd.getpwnam(user).pw_uid
@@ -131,6 +152,7 @@ class BaseInstaller(object):
             ret = False
         return ret
 
+    @utils.run_async
     def _run_apply_patches(self, client, restart_cmd, script_names,
                            python='python3'):
         installer_dir = os.path.dirname(os.path.realpath(__file__))
@@ -146,4 +168,7 @@ class BaseInstaller(object):
                     raise Exception('Do the command in remote'
                                     ' node failed, ret=%s, cmd=%s, output=%s'
                                     % (ret, cmd, output))
+            if 'nova-scheduler' in restart_cmd:
+                # Make sure scheduler has proper cpu_allocation_ratio
+                time.sleep(5)
             client.ssh(restart_cmd)

diff --git a/doctor_tests/installer/common/set_config.py b/doctor_tests/installer/common/set_config.py
index 3dc6cd9..e66d4c2 100644 (file)
--- a/doctor_tests/installer/common/set_config.py
+++ b/doctor_tests/installer/common/set_config.py
@@ -125,6 +125,7 @@ def set_event_definitions():
                 'reply_url': {'fields': 'payload.reply_url'},
                 'actions_at': {'fields': 'payload.actions_at',
                                'type': 'datetime'},
+                'reply_at': {'fields': 'payload.reply_at', 'type': 'datetime'},
                 'state': {'fields': 'payload.state'},
                 'session_id': {'fields': 'payload.session_id'},
                 'project_id': {'fields': 'payload.project_id'},

diff --git a/doctor_tests/scenario/maintenance.py b/doctor_tests/scenario/maintenance.py
index 9fcd412..09795c2 100644 (file)
--- a/doctor_tests/scenario/maintenance.py
+++ b/doctor_tests/scenario/maintenance.py
@@ -34,7 +34,11 @@ class Maintenance(object):
         auth = get_identity_auth(project=self.conf.doctor_project)
         self.neutron = neutron_client(get_session(auth=auth))
         self.stack = Stack(self.conf, self.log)
-        self.admin_tool = get_admin_tool(trasport_url, self.conf, self.log)
+        if self.conf.admin_tool.type == 'sample':
+            self.admin_tool = get_admin_tool(trasport_url, self.conf, self.log)
+            self.endpoint = 'maintenance'
+        else:
+            self.endpoint = 'v1/maintenance'
         self.app_manager = get_app_manager(self.stack, self.conf, self.log)
         self.inspector = get_inspector(self.conf, self.log)
 
@@ -110,7 +114,11 @@ class Maintenance(object):
                           parameters=parameters,
                           files=files)
 
-        self.admin_tool.start()
+        if self.conf.admin_tool.type == 'sample':
+            self.admin_tool.start()
+        else:
+            # TBD Now we expect Fenix is running in self.conf.admin_tool.port
+            pass
         self.app_manager.start()
         self.inspector.start()
 
@@ -122,16 +130,21 @@ class Maintenance(object):
             hostname = hvisor.__getattr__('hypervisor_hostname')
             maintenance_hosts.append(hostname)
 
-        url = 'http://0.0.0.0:%s/maintenance' % self.conf.admin_tool.port
+        url = ('http://%s:%s/%s' %
+               (self.conf.admin_tool.ip,
+                self.conf.admin_tool.port,
+                self.endpoint))
+
         # let's start maintenance 20sec from now, so projects will have
         # time to ACK to it before that
         maintenance_at = (datetime.datetime.utcnow() +
-                          datetime.timedelta(seconds=20)
+                          datetime.timedelta(seconds=30)
                           ).strftime('%Y-%m-%d %H:%M:%S')
         data = {'hosts': maintenance_hosts,
                 'state': 'MAINTENANCE',
                 'maintenance_at': maintenance_at,
-                'metadata': {'openstack_version': 'Pike'}}
+                'metadata': {'openstack_version': 'Rocky'},
+                'workflow': 'default'}
         headers = {
             'Content-Type': 'application/json',
             'Accept': 'application/json'}
@@ -143,49 +156,56 @@ class Maintenance(object):
 
     def remove_maintenance_session(self, session_id):
         self.log.info('remove maintenance session %s.......' % session_id)
+        url = ('http://%s:%s/%s/%s' %
+               (self.conf.admin_tool.ip,
+                self.conf.admin_tool.port,
+                self.endpoint,
+                session_id))
 
-        url = 'http://0.0.0.0:%s/maintenance' % self.conf.admin_tool.port
-
-        data = {'state': 'REMOVE_MAINTENANCE_SESSION',
-                'session_id': session_id}
         headers = {
             'Content-Type': 'application/json',
             'Accept': 'application/json'}
 
-        ret = requests.post(url, data=json.dumps(data), headers=headers)
+        ret = requests.delete(url, data=None, headers=headers)
         if ret.status_code != 200:
             raise Exception(ret.text)
 
     def get_maintenance_state(self, session_id):
-        url = 'http://0.0.0.0:%s/maintenance' % self.conf.admin_tool.port
-        data = {'session_id': session_id}
+        url = ('http://%s:%s/%s/%s' %
+               (self.conf.admin_tool.ip,
+                self.conf.admin_tool.port,
+                self.endpoint,
+                session_id))
+
         headers = {
             'Content-Type': 'application/json',
             'Accept': 'application/json'}
-        ret = requests.get(url, data=json.dumps(data), headers=headers)
+        ret = requests.get(url, data=None, headers=headers)
         if ret.status_code != 200:
             raise Exception(ret.text)
         return ret.json()['state']
 
     def wait_maintenance_complete(self, session_id):
-        retries = 66
+        retries = 90
         state = None
-        time.sleep(540)
-        while state != 'MAINTENANCE_COMPLETE' and retries > 0:
+        time.sleep(300)
+        while (state not in ['MAINTENANCE_DONE', 'MAINTENANCE_FAILED'] and
+               retries > 0):
             time.sleep(10)
             state = self.get_maintenance_state(session_id)
             retries = retries - 1
-        if retries == 0 and state != 'MAINTENANCE_COMPLETE':
-            raise Exception('maintenance %s not completed within 20min, status'
-                            ' %s' % (session_id, state))
-        elif state == 'MAINTENANCE_COMPLETE':
-            self.log.info('maintenance %s %s' % (session_id, state))
-            self.remove_maintenance_session(session_id)
-        elif state == 'MAINTENANCE_FAILED':
+        self.remove_maintenance_session(session_id)
+        self.log.info('maintenance %s ended with state %s' %
+                      (session_id, state))
+        if state == 'MAINTENANCE_FAILED':
             raise Exception('maintenance %s failed' % session_id)
+        elif retries == 0:
+            raise Exception('maintenance %s not completed within 20min' %
+                            session_id)
 
     def cleanup_maintenance(self):
-        self.admin_tool.stop()
+        if self.conf.admin_tool.type == 'sample':
+            self.admin_tool.stop()
         self.app_manager.stop()
         self.inspector.stop()
         self.log.info('stack delete start.......')

diff --git a/doctor_tests/stack.py b/doctor_tests/stack.py
index ee586fa..8a921be 100644 (file)
--- a/doctor_tests/stack.py
+++ b/doctor_tests/stack.py
@@ -94,7 +94,7 @@ class Stack(object):
             # It might not always work at first
             self.log.info('retry creating maintenance stack.......')
             self.delete()
-            time.sleep(3)
+            time.sleep(5)
             stack = self.heat.stacks.create(stack_name=self.stack_name,
                                             files=files,
                                             template=template,

diff --git a/tox.ini b/tox.ini
index 6e0d8b4..2242637 100644 (file)
--- a/tox.ini
+++ b/tox.ini
@@ -29,6 +29,7 @@ passenv =
     INSTALLER_TYPE
     INSTALLER_IP
     INSPECTOR_TYPE
+    ADMIN_TOOL_TYPE
     TEST_CASE
     SSH_KEY
 changedir = {toxinidir}/doctor_tests