doctor_tests/main.py

   1 ##############################################################################
   2 # Copyright (c) 2017 ZTE Corporation and others.
   3 #
   4 # All rights reserved. This program and the accompanying materials
   5 # are made available under the terms of the Apache License, Version 2.0
   6 # which accompanies this distribution, and is available at
   7 # http://www.apache.org/licenses/LICENSE-2.0
   8 ##############################################################################
   9 import os
  10 from os.path import isfile, join
  11 import random
  12 import sys
  13 import time
  14
  15 from doctor_tests.alarm import Alarm
  16 from doctor_tests.common.constants import Host
  17 from doctor_tests.common.utils import match_rep_in_file
  18 from doctor_tests import config
  19 from doctor_tests.consumer import get_consumer
  20 from doctor_tests.identity_auth import get_identity_auth
  21 from doctor_tests.identity_auth import get_session
  22 from doctor_tests.image import Image
  23 from doctor_tests.instance import Instance
  24 from doctor_tests.inspector import get_inspector
  25 from doctor_tests.installer import get_installer
  26 import doctor_tests.logger as doctor_log
  27 from doctor_tests.network import Network
  28 from doctor_tests.monitor import get_monitor
  29 from doctor_tests.os_clients import nova_client
  30 from doctor_tests.profiler_poc import main as profiler_main
  31 from doctor_tests.scenario.common import calculate_notification_time
  32 from doctor_tests.scenario.network_failure import NetworkFault
  33 from doctor_tests.user import User
  34
  35
  36 Logger = doctor_log.Logger('doctor')
  37 LOG = Logger.getLogger()
  38 LogFile = Logger.getLogFilename()
  39
  40
  41 class DoctorTest(object):
  42
  43     def __init__(self, conf):
  44         self.conf = conf
  45         self.image = Image(self.conf, LOG)
  46         self.user = User(self.conf, LOG)
  47         self.network = Network(self.conf, LOG)
  48         self.instance = Instance(self.conf, LOG)
  49         self.alarm = Alarm(self.conf, LOG)
  50         self.installer = get_installer(self.conf, LOG)
  51         self.inspector = get_inspector(self.conf, LOG)
  52         self.monitor = get_monitor(self.conf,
  53                                    self.inspector.get_inspector_url(),
  54                                    LOG)
  55         self.consumer = get_consumer(self.conf, LOG)
  56         self.fault = NetworkFault(self.conf, self.installer, LOG)
  57         auth = get_identity_auth(project=self.conf.doctor_project)
  58         self.nova = nova_client(self.conf.nova_version,
  59                                 get_session(auth=auth))
  60         self.down_host = None
  61
  62     def setup(self):
  63         # prepare the cloud env
  64         self.installer.setup()
  65
  66         # preparing VM image...
  67         self.image.create()
  68
  69         # creating test user...
  70         self.user.create()
  71
  72     def setup_fault_management(self):
  73         # user settings...
  74         self.user.update_quota()
  75
  76         # creating VM...
  77         self.network.create()
  78         self.instance.create()
  79         self.instance.wait_for_vm_launch()
  80
  81         # creating alarm...
  82         self.alarm.create()
  83
  84         # starting doctor sample components...
  85         # tbd tojuvone: move inspector and consumer to common setup
  86         # when they support updating VMs via instance.create and
  87         # instance.delete alarm
  88
  89         self.inspector.start()
  90         self.consumer.start()
  91         self.down_host = self.get_host_info_for_random_vm()
  92         self.monitor.start(self.down_host)
  93
  94     def test_fault_management(self):
  95         try:
  96             LOG.info('doctor fault management test starting.......')
  97
  98             # prepare test env
  99             self.setup_fault_management()
 100
 101             # wait for aodh alarms are updated in caches for event evaluator,
 102             # sleep time should be larger than event_alarm_cache_ttl
 103             # (default 60)
 104             time.sleep(60)
 105
 106             # injecting host failure...
 107             # NOTE (umar) add INTERFACE_NAME logic to host injection
 108
 109             self.fault.start(self.down_host)
 110             time.sleep(10)
 111
 112             # verify the test results
 113             # NOTE (umar) copy remote monitor.log file when monitor=collectd
 114             self.check_host_status(self.down_host.name, 'down')
 115
 116             notification_time = calculate_notification_time(LogFile)
 117             if notification_time < 1 and notification_time > 0:
 118                 LOG.info('doctor fault management test successfully, '
 119                          'notification_time=%s' % notification_time)
 120             else:
 121                 LOG.error('doctor fault management test failed, '
 122                           'notification_time=%s' % notification_time)
 123                 sys.exit(1)
 124
 125             if self.conf.profiler_type:
 126                 LOG.info('doctor fault management test begin to run '
 127                          'profile.......')
 128                 self.collect_logs()
 129                 self.run_profiler()
 130         except Exception as e:
 131             LOG.error('doctor fault management test failed, '
 132                       'Exception=%s' % e)
 133             sys.exit(1)
 134         finally:
 135             self.cleanup_fault_management()
 136
 137     def _amount_compute_nodes(self):
 138         services = self.nova.services.list(binary='nova-compute')
 139         return len(services)
 140
 141     def test_maintenance(self):
 142         cnodes = self._amount_compute_nodes()
 143         if cnodes < 3:
 144             # need 2 compute for redundancy and one spare to migrate
 145             LOG.info('not enough compute nodes, skipping doctor '
 146                      'maintenance test')
 147             return
 148         try:
 149             LOG.info('doctor maintenance test starting.......')
 150             # TODO (tojuvone) test setup and actual test
 151         except Exception as e:
 152             LOG.error('doctor maintenance test failed, Exception=%s' % e)
 153             sys.exit(1)
 154         # TODO (tojuvone) finally: test case specific cleanup
 155
 156     def run(self):
 157         """run doctor tests"""
 158         try:
 159             LOG.info('doctor test starting.......')
 160             # prepare common test env
 161             self.setup()
 162             if self.conf.test_case == 'all':
 163                 self.test_fault_management()
 164                 self.test_maintenance()
 165             else:
 166                 getattr(self, self.conf.test_case)()
 167         except Exception as e:
 168             LOG.error('doctor test failed, Exception=%s' % e)
 169             sys.exit(1)
 170         finally:
 171             self.cleanup()
 172
 173     def get_host_info_for_random_vm(self):
 174         num = random.randint(0, self.conf.instance_count - 1)
 175         vm_name = "%s%d" % (self.conf.instance_basename, num)
 176
 177         servers = \
 178             {getattr(server, 'name'): server
 179              for server in self.nova.servers.list()}
 180         server = servers.get(vm_name)
 181         if not server:
 182             raise \
 183                 Exception('Can not find instance: vm_name(%s)' % vm_name)
 184         host_name = server.__dict__.get('OS-EXT-SRV-ATTR:hypervisor_hostname')
 185         host_ip = self.installer.get_host_ip_from_hostname(host_name)
 186
 187         LOG.info('Get host info(name:%s, ip:%s) which vm(%s) launched at'
 188                  % (host_name, host_ip, vm_name))
 189         return Host(host_name, host_ip)
 190
 191     def check_host_status(self, hostname, state):
 192         service = self.nova.services.list(host=hostname,
 193                                           binary='nova-compute')
 194         host_state = service[0].__dict__.get('state')
 195         assert host_state == state
 196
 197     def unset_forced_down_hosts(self):
 198         if self.down_host:
 199             self.nova.services.force_down(self.down_host.name,
 200                                           'nova-compute', False)
 201             time.sleep(2)
 202             self.check_host_status(self.down_host.name, 'up')
 203
 204     def collect_logs(self):
 205         self.fault.get_disable_network_log()
 206
 207     def run_profiler(self):
 208
 209         net_down_log_file = self.fault.get_disable_network_log()
 210         reg = '(?<=doctor set link down at )\d+.\d+'
 211         linkdown = float(match_rep_in_file(reg, net_down_log_file).group(0))
 212
 213         reg = '(.* doctor mark vm.* error at )(\d+.\d+)'
 214         vmdown = float(match_rep_in_file(reg, LogFile).group(2))
 215
 216         reg = '(.* doctor mark host.* down at )(\d+.\d+)'
 217         hostdown = float(match_rep_in_file(reg, LogFile).group(2))
 218
 219         reg = '(?<=doctor monitor detected at )\d+.\d+'
 220         detected = float(match_rep_in_file(reg, LogFile).group(0))
 221
 222         reg = '(?<=doctor consumer notified at )\d+.\d+'
 223         notified = float(match_rep_in_file(reg, LogFile).group(0))
 224
 225         # TODO(yujunz) check the actual delay to verify time sync status
 226         # expected ~1s delay from $trigger to $linkdown
 227         relative_start = linkdown
 228         os.environ['DOCTOR_PROFILER_T00'] = \
 229             str(int((linkdown - relative_start) * 1000))
 230         os.environ['DOCTOR_PROFILER_T01'] = \
 231             str(int((detected - relative_start) * 1000))
 232         os.environ['DOCTOR_PROFILER_T03'] = \
 233             str(int((vmdown - relative_start) * 1000))
 234         os.environ['DOCTOR_PROFILER_T04'] = \
 235             str(int((hostdown - relative_start) * 1000))
 236         os.environ['DOCTOR_PROFILER_T09'] = \
 237             str(int((notified - relative_start) * 1000))
 238
 239         profiler_main(log=LOG)
 240
 241     def cleanup_fault_management(self):
 242         self.unset_forced_down_hosts()
 243         self.inspector.stop()
 244         self.monitor.stop()
 245         self.consumer.stop()
 246         self.alarm.delete()
 247         self.instance.delete()
 248         self.network.delete()
 249         self.fault.cleanup()
 250
 251     def cleanup(self):
 252         self.installer.cleanup()
 253         self.image.delete()
 254         self.user.delete()
 255
 256
 257 def main():
 258     """doctor main"""
 259     test_dir = os.path.split(os.path.realpath(__file__))[0]
 260     doctor_root_dir = os.path.dirname(test_dir)
 261
 262     config_file_dir = '{0}/{1}'.format(doctor_root_dir, 'etc/')
 263     config_files = [join(config_file_dir, f)
 264                     for f in os.listdir(config_file_dir)
 265                     if isfile(join(config_file_dir, f))]
 266
 267     conf = config.prepare_conf(args=sys.argv[1:],
 268                                config_files=config_files)
 269
 270     doctor = DoctorTest(conf)
 271     doctor.run()