scheduler/monitor_db_unittest.py

   1 #!/usr/bin/python
   2
   3 import gc, logging, time
   4 import common
   5 from autotest_lib.frontend import setup_django_environment
   6 from autotest_lib.frontend.afe import frontend_test_utils
   7 from autotest_lib.client.common_lib.test_utils import mock
   8 from autotest_lib.client.common_lib.test_utils import unittest
   9 from autotest_lib.database import database_connection
  10 from autotest_lib.frontend.afe import models
  11 from autotest_lib.scheduler import monitor_db, drone_manager, email_manager
  12 from autotest_lib.scheduler import scheduler_config, gc_stats, host_scheduler
  13 from autotest_lib.scheduler import monitor_db_functional_test
  14 from autotest_lib.scheduler import scheduler_models
  15
  16 _DEBUG = False
  17
  18
  19 class DummyAgentTask(object):
  20     num_processes = 1
  21     owner_username = 'my_user'
  22
  23     def get_drone_hostnames_allowed(self):
  24         return None
  25
  26
  27 class DummyAgent(object):
  28     started = False
  29     _is_done = False
  30     host_ids = ()
  31     queue_entry_ids = ()
  32
  33     def __init__(self):
  34         self.task = DummyAgentTask()
  35
  36
  37     def tick(self):
  38         self.started = True
  39
  40
  41     def is_done(self):
  42         return self._is_done
  43
  44
  45     def set_done(self, done):
  46         self._is_done = done
  47
  48
  49 class IsRow(mock.argument_comparator):
  50     def __init__(self, row_id):
  51         self.row_id = row_id
  52
  53
  54     def is_satisfied_by(self, parameter):
  55         return list(parameter)[0] == self.row_id
  56
  57
  58     def __str__(self):
  59         return 'row with id %s' % self.row_id
  60
  61
  62 class IsAgentWithTask(mock.argument_comparator):
  63     def __init__(self, task):
  64         self._task = task
  65
  66
  67     def is_satisfied_by(self, parameter):
  68         if not isinstance(parameter, monitor_db.Agent):
  69             return False
  70         tasks = list(parameter.queue.queue)
  71         if len(tasks) != 1:
  72             return False
  73         return tasks[0] == self._task
  74
  75
  76 def _set_host_and_qe_ids(agent_or_task, id_list=None):
  77     if id_list is None:
  78         id_list = []
  79     agent_or_task.host_ids = agent_or_task.queue_entry_ids = id_list
  80
  81
  82 class BaseSchedulerTest(unittest.TestCase,
  83                         frontend_test_utils.FrontendTestMixin):
  84     _config_section = 'AUTOTEST_WEB'
  85
  86     def _do_query(self, sql):
  87         self._database.execute(sql)
  88
  89
  90     def _set_monitor_stubs(self):
  91         # Clear the instance cache as this is a brand new database.
  92         scheduler_models.DBObject._clear_instance_cache()
  93
  94         self._database = (
  95             database_connection.TranslatingDatabase.get_test_database(
  96                 translators=monitor_db_functional_test._DB_TRANSLATORS))
  97         self._database.connect(db_type='django')
  98         self._database.debug = _DEBUG
  99
 100         self.god.stub_with(monitor_db, '_db', self._database)
 101         self.god.stub_with(scheduler_models, '_db', self._database)
 102         self.god.stub_with(drone_manager.instance(), '_results_dir',
 103                            '/test/path')
 104         self.god.stub_with(drone_manager.instance(), '_temporary_directory',
 105                            '/test/path/tmp')
 106
 107         monitor_db.initialize_globals()
 108         scheduler_models.initialize_globals()
 109
 110
 111     def setUp(self):
 112         self._frontend_common_setup()
 113         self._set_monitor_stubs()
 114         self._dispatcher = monitor_db.Dispatcher()
 115
 116
 117     def tearDown(self):
 118         self._database.disconnect()
 119         self._frontend_common_teardown()
 120
 121
 122     def _update_hqe(self, set, where=''):
 123         query = 'UPDATE afe_host_queue_entries SET ' + set
 124         if where:
 125             query += ' WHERE ' + where
 126         self._do_query(query)
 127
 128
 129 class DispatcherSchedulingTest(BaseSchedulerTest):
 130     _jobs_scheduled = []
 131
 132
 133     def tearDown(self):
 134         super(DispatcherSchedulingTest, self).tearDown()
 135
 136
 137     def _set_monitor_stubs(self):
 138         super(DispatcherSchedulingTest, self)._set_monitor_stubs()
 139
 140         def hqe__do_schedule_pre_job_tasks_stub(queue_entry):
 141             """Called by HostQueueEntry.run()."""
 142             self._record_job_scheduled(queue_entry.job.id, queue_entry.host.id)
 143             queue_entry.set_status('Starting')
 144
 145         self.god.stub_with(scheduler_models.HostQueueEntry,
 146                            '_do_schedule_pre_job_tasks',
 147                            hqe__do_schedule_pre_job_tasks_stub)
 148
 149         def hqe_queue_log_record_stub(self, log_line):
 150             """No-Op to avoid calls down to the _drone_manager during tests."""
 151
 152         self.god.stub_with(scheduler_models.HostQueueEntry, 'queue_log_record',
 153                            hqe_queue_log_record_stub)
 154
 155
 156     def _record_job_scheduled(self, job_id, host_id):
 157         record = (job_id, host_id)
 158         self.assert_(record not in self._jobs_scheduled,
 159                      'Job %d scheduled on host %d twice' %
 160                      (job_id, host_id))
 161         self._jobs_scheduled.append(record)
 162
 163
 164     def _assert_job_scheduled_on(self, job_id, host_id):
 165         record = (job_id, host_id)
 166         self.assert_(record in self._jobs_scheduled,
 167                      'Job %d not scheduled on host %d as expected\n'
 168                      'Jobs scheduled: %s' %
 169                      (job_id, host_id, self._jobs_scheduled))
 170         self._jobs_scheduled.remove(record)
 171
 172
 173     def _assert_job_scheduled_on_number_of(self, job_id, host_ids, number):
 174         """Assert job was scheduled on exactly number hosts out of a set."""
 175         found = []
 176         for host_id in host_ids:
 177             record = (job_id, host_id)
 178             if record in self._jobs_scheduled:
 179                 found.append(record)
 180                 self._jobs_scheduled.remove(record)
 181         if len(found) < number:
 182             self.fail('Job %d scheduled on fewer than %d hosts in %s.\n'
 183                       'Jobs scheduled: %s' % (job_id, number, host_ids, found))
 184         elif len(found) > number:
 185             self.fail('Job %d scheduled on more than %d hosts in %s.\n'
 186                       'Jobs scheduled: %s' % (job_id, number, host_ids, found))
 187
 188
 189     def _check_for_extra_schedulings(self):
 190         if len(self._jobs_scheduled) != 0:
 191             self.fail('Extra jobs scheduled: ' +
 192                       str(self._jobs_scheduled))
 193
 194
 195     def _convert_jobs_to_metahosts(self, *job_ids):
 196         sql_tuple = '(' + ','.join(str(i) for i in job_ids) + ')'
 197         self._do_query('UPDATE afe_host_queue_entries SET '
 198                        'meta_host=host_id, host_id=NULL '
 199                        'WHERE job_id IN ' + sql_tuple)
 200
 201
 202     def _lock_host(self, host_id):
 203         self._do_query('UPDATE afe_hosts SET locked=1 WHERE id=' +
 204                        str(host_id))
 205
 206
 207     def setUp(self):
 208         super(DispatcherSchedulingTest, self).setUp()
 209         self._jobs_scheduled = []
 210
 211
 212     def _run_scheduler(self):
 213         for _ in xrange(2): # metahost scheduling can take two cycles
 214             self._dispatcher._schedule_new_jobs()
 215
 216
 217     def _test_basic_scheduling_helper(self, use_metahosts):
 218         'Basic nonmetahost scheduling'
 219         self._create_job_simple([1], use_metahosts)
 220         self._create_job_simple([2], use_metahosts)
 221         self._run_scheduler()
 222         self._assert_job_scheduled_on(1, 1)
 223         self._assert_job_scheduled_on(2, 2)
 224         self._check_for_extra_schedulings()
 225
 226
 227     def _test_priorities_helper(self, use_metahosts):
 228         'Test prioritization ordering'
 229         self._create_job_simple([1], use_metahosts)
 230         self._create_job_simple([2], use_metahosts)
 231         self._create_job_simple([1,2], use_metahosts)
 232         self._create_job_simple([1], use_metahosts, priority=1)
 233         self._run_scheduler()
 234         self._assert_job_scheduled_on(4, 1) # higher priority
 235         self._assert_job_scheduled_on(2, 2) # earlier job over later
 236         self._check_for_extra_schedulings()
 237
 238
 239     def _test_hosts_ready_helper(self, use_metahosts):
 240         """
 241         Only hosts that are status=Ready, unlocked and not invalid get
 242         scheduled.
 243         """
 244         self._create_job_simple([1], use_metahosts)
 245         self._do_query('UPDATE afe_hosts SET status="Running" WHERE id=1')
 246         self._run_scheduler()
 247         self._check_for_extra_schedulings()
 248
 249         self._do_query('UPDATE afe_hosts SET status="Ready", locked=1 '
 250                        'WHERE id=1')
 251         self._run_scheduler()
 252         self._check_for_extra_schedulings()
 253
 254         self._do_query('UPDATE afe_hosts SET locked=0, invalid=1 '
 255                        'WHERE id=1')
 256         self._run_scheduler()
 257         if not use_metahosts:
 258             self._assert_job_scheduled_on(1, 1)
 259         self._check_for_extra_schedulings()
 260
 261
 262     def _test_hosts_idle_helper(self, use_metahosts):
 263         'Only idle hosts get scheduled'
 264         self._create_job(hosts=[1], active=True)
 265         self._create_job_simple([1], use_metahosts)
 266         self._run_scheduler()
 267         self._check_for_extra_schedulings()
 268
 269
 270     def _test_obey_ACLs_helper(self, use_metahosts):
 271         self._do_query('DELETE FROM afe_acl_groups_hosts WHERE host_id=1')
 272         self._create_job_simple([1], use_metahosts)
 273         self._run_scheduler()
 274         self._check_for_extra_schedulings()
 275
 276
 277     def test_basic_scheduling(self):
 278         self._test_basic_scheduling_helper(False)
 279
 280
 281     def test_priorities(self):
 282         self._test_priorities_helper(False)
 283
 284
 285     def test_hosts_ready(self):
 286         self._test_hosts_ready_helper(False)
 287
 288
 289     def test_hosts_idle(self):
 290         self._test_hosts_idle_helper(False)
 291
 292
 293     def test_obey_ACLs(self):
 294         self._test_obey_ACLs_helper(False)
 295
 296
 297     def test_one_time_hosts_ignore_ACLs(self):
 298         self._do_query('DELETE FROM afe_acl_groups_hosts WHERE host_id=1')
 299         self._do_query('UPDATE afe_hosts SET invalid=1 WHERE id=1')
 300         self._create_job_simple([1])
 301         self._run_scheduler()
 302         self._assert_job_scheduled_on(1, 1)
 303         self._check_for_extra_schedulings()
 304
 305
 306     def test_non_metahost_on_invalid_host(self):
 307         """
 308         Non-metahost entries can get scheduled on invalid hosts (this is how
 309         one-time hosts work).
 310         """
 311         self._do_query('UPDATE afe_hosts SET invalid=1')
 312         self._test_basic_scheduling_helper(False)
 313
 314
 315     def test_metahost_scheduling(self):
 316         """
 317         Basic metahost scheduling
 318         """
 319         self._test_basic_scheduling_helper(True)
 320
 321
 322     def test_metahost_priorities(self):
 323         self._test_priorities_helper(True)
 324
 325
 326     def test_metahost_hosts_ready(self):
 327         self._test_hosts_ready_helper(True)
 328
 329
 330     def test_metahost_hosts_idle(self):
 331         self._test_hosts_idle_helper(True)
 332
 333
 334     def test_metahost_obey_ACLs(self):
 335         self._test_obey_ACLs_helper(True)
 336
 337
 338     def _setup_test_only_if_needed_labels(self):
 339         # apply only_if_needed label3 to host1
 340         models.Host.smart_get('host1').labels.add(self.label3)
 341         return self._create_job_simple([1], use_metahost=True)
 342
 343
 344     def test_only_if_needed_labels_avoids_host(self):
 345         job = self._setup_test_only_if_needed_labels()
 346         # if the job doesn't depend on label3, there should be no scheduling
 347         self._run_scheduler()
 348         self._check_for_extra_schedulings()
 349
 350
 351     def test_only_if_needed_labels_schedules(self):
 352         job = self._setup_test_only_if_needed_labels()
 353         job.dependency_labels.add(self.label3)
 354         self._run_scheduler()
 355         self._assert_job_scheduled_on(1, 1)
 356         self._check_for_extra_schedulings()
 357
 358
 359     def test_only_if_needed_labels_via_metahost(self):
 360         job = self._setup_test_only_if_needed_labels()
 361         job.dependency_labels.add(self.label3)
 362         # should also work if the metahost is the only_if_needed label
 363         self._do_query('DELETE FROM afe_jobs_dependency_labels')
 364         self._create_job(metahosts=[3])
 365         self._run_scheduler()
 366         self._assert_job_scheduled_on(2, 1)
 367         self._check_for_extra_schedulings()
 368
 369
 370     def test_nonmetahost_over_metahost(self):
 371         """
 372         Non-metahost entries should take priority over metahost entries
 373         for the same host
 374         """
 375         self._create_job(metahosts=[1])
 376         self._create_job(hosts=[1])
 377         self._run_scheduler()
 378         self._assert_job_scheduled_on(2, 1)
 379         self._check_for_extra_schedulings()
 380
 381
 382     def test_metahosts_obey_blocks(self):
 383         """
 384         Metahosts can't get scheduled on hosts already scheduled for
 385         that job.
 386         """
 387         self._create_job(metahosts=[1], hosts=[1])
 388         # make the nonmetahost entry complete, so the metahost can try
 389         # to get scheduled
 390         self._update_hqe(set='complete = 1', where='host_id=1')
 391         self._run_scheduler()
 392         self._check_for_extra_schedulings()
 393
 394
 395     # TODO(gps): These should probably live in their own TestCase class
 396     # specific to testing HostScheduler methods directly.  It was convenient
 397     # to put it here for now to share existing test environment setup code.
 398     def test_HostScheduler_check_atomic_group_labels(self):
 399         normal_job = self._create_job(metahosts=[0])
 400         atomic_job = self._create_job(atomic_group=1)
 401         # Indirectly initialize the internal state of the host scheduler.
 402         self._dispatcher._refresh_pending_queue_entries()
 403
 404         atomic_hqe = scheduler_models.HostQueueEntry.fetch(where='job_id=%d' %
 405                                                      atomic_job.id)[0]
 406         normal_hqe = scheduler_models.HostQueueEntry.fetch(where='job_id=%d' %
 407                                                      normal_job.id)[0]
 408
 409         host_scheduler = self._dispatcher._host_scheduler
 410         self.assertTrue(host_scheduler._check_atomic_group_labels(
 411                 [self.label4.id], atomic_hqe))
 412         self.assertFalse(host_scheduler._check_atomic_group_labels(
 413                 [self.label4.id], normal_hqe))
 414         self.assertFalse(host_scheduler._check_atomic_group_labels(
 415                 [self.label5.id, self.label6.id, self.label7.id], normal_hqe))
 416         self.assertTrue(host_scheduler._check_atomic_group_labels(
 417                 [self.label4.id, self.label6.id], atomic_hqe))
 418         self.assertTrue(host_scheduler._check_atomic_group_labels(
 419                         [self.label4.id, self.label5.id],
 420                         atomic_hqe))
 421
 422
 423     def test_HostScheduler_get_host_atomic_group_id(self):
 424         job = self._create_job(metahosts=[self.label6.id])
 425         queue_entry = scheduler_models.HostQueueEntry.fetch(
 426                 where='job_id=%d' % job.id)[0]
 427         # Indirectly initialize the internal state of the host scheduler.
 428         self._dispatcher._refresh_pending_queue_entries()
 429
 430         # Test the host scheduler
 431         host_scheduler = self._dispatcher._host_scheduler
 432
 433         # Two labels each in a different atomic group.  This should log an
 434         # error and continue.
 435         orig_logging_error = logging.error
 436         def mock_logging_error(message, *args):
 437             mock_logging_error._num_calls += 1
 438             # Test the logging call itself, we just wrapped it to count it.
 439             orig_logging_error(message, *args)
 440         mock_logging_error._num_calls = 0
 441         self.god.stub_with(logging, 'error', mock_logging_error)
 442         self.assertNotEquals(None, host_scheduler._get_host_atomic_group_id(
 443                 [self.label4.id, self.label8.id], queue_entry))
 444         self.assertTrue(mock_logging_error._num_calls > 0)
 445         self.god.unstub(logging, 'error')
 446
 447         # Two labels both in the same atomic group, this should not raise an
 448         # error, it will merely cause the job to schedule on the intersection.
 449         self.assertEquals(1, host_scheduler._get_host_atomic_group_id(
 450                 [self.label4.id, self.label5.id]))
 451
 452         self.assertEquals(None, host_scheduler._get_host_atomic_group_id([]))
 453         self.assertEquals(None, host_scheduler._get_host_atomic_group_id(
 454                 [self.label3.id, self.label7.id, self.label6.id]))
 455         self.assertEquals(1, host_scheduler._get_host_atomic_group_id(
 456                 [self.label4.id, self.label7.id, self.label6.id]))
 457         self.assertEquals(1, host_scheduler._get_host_atomic_group_id(
 458                 [self.label7.id, self.label5.id]))
 459
 460
 461     def test_atomic_group_hosts_blocked_from_non_atomic_jobs(self):
 462         # Create a job scheduled to run on label6.
 463         self._create_job(metahosts=[self.label6.id])
 464         self._run_scheduler()
 465         # label6 only has hosts that are in atomic groups associated with it,
 466         # there should be no scheduling.
 467         self._check_for_extra_schedulings()
 468
 469
 470     def test_atomic_group_hosts_blocked_from_non_atomic_jobs_explicit(self):
 471         # Create a job scheduled to run on label5.  This is an atomic group
 472         # label but this job does not request atomic group scheduling.
 473         self._create_job(metahosts=[self.label5.id])
 474         self._run_scheduler()
 475         # label6 only has hosts that are in atomic groups associated with it,
 476         # there should be no scheduling.
 477         self._check_for_extra_schedulings()
 478
 479
 480     def test_atomic_group_scheduling_basics(self):
 481         # Create jobs scheduled to run on an atomic group.
 482         job_a = self._create_job(synchronous=True, metahosts=[self.label4.id],
 483                          atomic_group=1)
 484         job_b = self._create_job(synchronous=True, metahosts=[self.label5.id],
 485                          atomic_group=1)
 486         self._run_scheduler()
 487         # atomic_group.max_number_of_machines was 2 so we should run on 2.
 488         self._assert_job_scheduled_on_number_of(job_a.id, (5, 6, 7), 2)
 489         self._assert_job_scheduled_on(job_b.id, 8)  # label5
 490         self._assert_job_scheduled_on(job_b.id, 9)  # label5
 491         self._check_for_extra_schedulings()
 492
 493         # The three host label4 atomic group still has one host available.
 494         # That means a job with a synch_count of 1 asking to be scheduled on
 495         # the atomic group can still use the final machine.
 496         #
 497         # This may seem like a somewhat odd use case.  It allows the use of an
 498         # atomic group as a set of machines to run smaller jobs within (a set
 499         # of hosts configured for use in network tests with eachother perhaps?)
 500         onehost_job = self._create_job(atomic_group=1)
 501         self._run_scheduler()
 502         self._assert_job_scheduled_on_number_of(onehost_job.id, (5, 6, 7), 1)
 503         self._check_for_extra_schedulings()
 504
 505         # No more atomic groups have hosts available, no more jobs should
 506         # be scheduled.
 507         self._create_job(atomic_group=1)
 508         self._run_scheduler()
 509         self._check_for_extra_schedulings()
 510
 511
 512     def test_atomic_group_scheduling_obeys_acls(self):
 513         # Request scheduling on a specific atomic label but be denied by ACLs.
 514         self._do_query('DELETE FROM afe_acl_groups_hosts '
 515                        'WHERE host_id in (8,9)')
 516         job = self._create_job(metahosts=[self.label5.id], atomic_group=1)
 517         self._run_scheduler()
 518         self._check_for_extra_schedulings()
 519
 520
 521     def test_atomic_group_scheduling_dependency_label_exclude(self):
 522         # A dependency label that matches no hosts in the atomic group.
 523         job_a = self._create_job(atomic_group=1)
 524         job_a.dependency_labels.add(self.label3)
 525         self._run_scheduler()
 526         self._check_for_extra_schedulings()
 527
 528
 529     def test_atomic_group_scheduling_metahost_dependency_label_exclude(self):
 530         # A metahost and dependency label that excludes too many hosts.
 531         job_b = self._create_job(synchronous=True, metahosts=[self.label4.id],
 532                                  atomic_group=1)
 533         job_b.dependency_labels.add(self.label7)
 534         self._run_scheduler()
 535         self._check_for_extra_schedulings()
 536
 537
 538     def test_atomic_group_scheduling_dependency_label_match(self):
 539         # A dependency label that exists on enough atomic group hosts in only
 540         # one of the two atomic group labels.
 541         job_c = self._create_job(synchronous=True, atomic_group=1)
 542         job_c.dependency_labels.add(self.label7)
 543         self._run_scheduler()
 544         self._assert_job_scheduled_on_number_of(job_c.id, (8, 9), 2)
 545         self._check_for_extra_schedulings()
 546
 547
 548     def test_atomic_group_scheduling_no_metahost(self):
 549         # Force it to schedule on the other group for a reliable test.
 550         self._do_query('UPDATE afe_hosts SET invalid=1 WHERE id=9')
 551         # An atomic job without a metahost.
 552         job = self._create_job(synchronous=True, atomic_group=1)
 553         self._run_scheduler()
 554         self._assert_job_scheduled_on_number_of(job.id, (5, 6, 7), 2)
 555         self._check_for_extra_schedulings()
 556
 557
 558     def test_atomic_group_scheduling_partial_group(self):
 559         # Make one host in labels[3] unavailable so that there are only two
 560         # hosts left in the group.
 561         self._do_query('UPDATE afe_hosts SET status="Repair Failed" WHERE id=5')
 562         job = self._create_job(synchronous=True, metahosts=[self.label4.id],
 563                          atomic_group=1)
 564         self._run_scheduler()
 565         # Verify that it was scheduled on the 2 ready hosts in that group.
 566         self._assert_job_scheduled_on(job.id, 6)
 567         self._assert_job_scheduled_on(job.id, 7)
 568         self._check_for_extra_schedulings()
 569
 570
 571     def test_atomic_group_scheduling_not_enough_available(self):
 572         # Mark some hosts in each atomic group label as not usable.
 573         # One host running, another invalid in the first group label.
 574         self._do_query('UPDATE afe_hosts SET status="Running" WHERE id=5')
 575         self._do_query('UPDATE afe_hosts SET invalid=1 WHERE id=6')
 576         # One host invalid in the second group label.
 577         self._do_query('UPDATE afe_hosts SET invalid=1 WHERE id=9')
 578         # Nothing to schedule when no group label has enough (2) good hosts..
 579         self._create_job(atomic_group=1, synchronous=True)
 580         self._run_scheduler()
 581         # There are not enough hosts in either atomic group,
 582         # No more scheduling should occur.
 583         self._check_for_extra_schedulings()
 584
 585         # Now create an atomic job that has a synch count of 1.  It should
 586         # schedule on exactly one of the hosts.
 587         onehost_job = self._create_job(atomic_group=1)
 588         self._run_scheduler()
 589         self._assert_job_scheduled_on_number_of(onehost_job.id, (7, 8), 1)
 590
 591
 592     def test_atomic_group_scheduling_no_valid_hosts(self):
 593         self._do_query('UPDATE afe_hosts SET invalid=1 WHERE id in (8,9)')
 594         self._create_job(synchronous=True, metahosts=[self.label5.id],
 595                          atomic_group=1)
 596         self._run_scheduler()
 597         # no hosts in the selected group and label are valid.  no schedulings.
 598         self._check_for_extra_schedulings()
 599
 600
 601     def test_atomic_group_scheduling_metahost_works(self):
 602         # Test that atomic group scheduling also obeys metahosts.
 603         self._create_job(metahosts=[0], atomic_group=1)
 604         self._run_scheduler()
 605         # There are no atomic group hosts that also have that metahost.
 606         self._check_for_extra_schedulings()
 607
 608         job_b = self._create_job(metahosts=[self.label5.id], atomic_group=1)
 609         self._run_scheduler()
 610         self._assert_job_scheduled_on(job_b.id, 8)
 611         self._assert_job_scheduled_on(job_b.id, 9)
 612         self._check_for_extra_schedulings()
 613
 614
 615     def test_atomic_group_skips_ineligible_hosts(self):
 616         # Test hosts marked ineligible for this job are not eligible.
 617         # How would this ever happen anyways?
 618         job = self._create_job(metahosts=[self.label4.id], atomic_group=1)
 619         models.IneligibleHostQueue.objects.create(job=job, host_id=5)
 620         models.IneligibleHostQueue.objects.create(job=job, host_id=6)
 621         models.IneligibleHostQueue.objects.create(job=job, host_id=7)
 622         self._run_scheduler()
 623         # No scheduling should occur as all desired hosts were ineligible.
 624         self._check_for_extra_schedulings()
 625
 626
 627     def test_atomic_group_scheduling_fail(self):
 628         # If synch_count is > the atomic group number of machines, the job
 629         # should be aborted immediately.
 630         model_job = self._create_job(synchronous=True, atomic_group=1)
 631         model_job.synch_count = 4
 632         model_job.save()
 633         job = scheduler_models.Job(id=model_job.id)
 634         self._run_scheduler()
 635         self._check_for_extra_schedulings()
 636         queue_entries = job.get_host_queue_entries()
 637         self.assertEqual(1, len(queue_entries))
 638         self.assertEqual(queue_entries[0].status,
 639                          models.HostQueueEntry.Status.ABORTED)
 640
 641
 642     def test_atomic_group_no_labels_no_scheduling(self):
 643         # Never schedule on atomic groups marked invalid.
 644         job = self._create_job(metahosts=[self.label5.id], synchronous=True,
 645                                atomic_group=1)
 646         # Deleting an atomic group via the frontend marks it invalid and
 647         # removes all label references to the group.  The job now references
 648         # an invalid atomic group with no labels associated with it.
 649         self.label5.atomic_group.invalid = True
 650         self.label5.atomic_group.save()
 651         self.label5.atomic_group = None
 652         self.label5.save()
 653
 654         self._run_scheduler()
 655         self._check_for_extra_schedulings()
 656
 657
 658     def test_schedule_directly_on_atomic_group_host_fail(self):
 659         # Scheduling a job directly on hosts in an atomic group must
 660         # fail to avoid users inadvertently holding up the use of an
 661         # entire atomic group by using the machines individually.
 662         job = self._create_job(hosts=[5])
 663         self._run_scheduler()
 664         self._check_for_extra_schedulings()
 665
 666
 667     def test_schedule_directly_on_atomic_group_host(self):
 668         # Scheduling a job directly on one host in an atomic group will
 669         # work when the atomic group is listed on the HQE in addition
 670         # to the host (assuming the sync count is 1).
 671         job = self._create_job(hosts=[5], atomic_group=1)
 672         self._run_scheduler()
 673         self._assert_job_scheduled_on(job.id, 5)
 674         self._check_for_extra_schedulings()
 675
 676
 677     def test_schedule_directly_on_atomic_group_hosts_sync2(self):
 678         job = self._create_job(hosts=[5,8], atomic_group=1, synchronous=True)
 679         self._run_scheduler()
 680         self._assert_job_scheduled_on(job.id, 5)
 681         self._assert_job_scheduled_on(job.id, 8)
 682         self._check_for_extra_schedulings()
 683
 684
 685     def test_schedule_directly_on_atomic_group_hosts_wrong_group(self):
 686         job = self._create_job(hosts=[5,8], atomic_group=2, synchronous=True)
 687         self._run_scheduler()
 688         self._check_for_extra_schedulings()
 689
 690
 691     def test_only_schedule_queued_entries(self):
 692         self._create_job(metahosts=[1])
 693         self._update_hqe(set='active=1, host_id=2')
 694         self._run_scheduler()
 695         self._check_for_extra_schedulings()
 696
 697
 698     def test_no_ready_hosts(self):
 699         self._create_job(hosts=[1])
 700         self._do_query('UPDATE afe_hosts SET status="Repair Failed"')
 701         self._run_scheduler()
 702         self._check_for_extra_schedulings()
 703
 704
 705     def test_garbage_collection(self):
 706         self.god.stub_with(self._dispatcher, '_seconds_between_garbage_stats',
 707                            999999)
 708         self.god.stub_function(gc, 'collect')
 709         self.god.stub_function(gc_stats, '_log_garbage_collector_stats')
 710         gc.collect.expect_call().and_return(0)
 711         gc_stats._log_garbage_collector_stats.expect_call()
 712         # Force a garbage collection run
 713         self._dispatcher._last_garbage_stats_time = 0
 714         self._dispatcher._garbage_collection()
 715         # The previous call should have reset the time, it won't do anything
 716         # the second time.  If it does, we'll get an unexpected call.
 717         self._dispatcher._garbage_collection()
 718
 719
 720
 721 class DispatcherThrottlingTest(BaseSchedulerTest):
 722     """
 723     Test that the dispatcher throttles:
 724      * total number of running processes
 725      * number of processes started per cycle
 726     """
 727     _MAX_RUNNING = 3
 728     _MAX_STARTED = 2
 729
 730     def setUp(self):
 731         super(DispatcherThrottlingTest, self).setUp()
 732         scheduler_config.config.max_processes_per_drone = self._MAX_RUNNING
 733         scheduler_config.config.max_processes_started_per_cycle = (
 734             self._MAX_STARTED)
 735
 736         def fake_max_runnable_processes(fake_self, username,
 737                                         drone_hostnames_allowed):
 738             running = sum(agent.task.num_processes
 739                           for agent in self._agents
 740                           if agent.started and not agent.is_done())
 741             return self._MAX_RUNNING - running
 742         self.god.stub_with(drone_manager.DroneManager, 'max_runnable_processes',
 743                            fake_max_runnable_processes)
 744
 745
 746     def _setup_some_agents(self, num_agents):
 747         self._agents = [DummyAgent() for i in xrange(num_agents)]
 748         self._dispatcher._agents = list(self._agents)
 749
 750
 751     def _run_a_few_cycles(self):
 752         for i in xrange(4):
 753             self._dispatcher._handle_agents()
 754
 755
 756     def _assert_agents_started(self, indexes, is_started=True):
 757         for i in indexes:
 758             self.assert_(self._agents[i].started == is_started,
 759                          'Agent %d %sstarted' %
 760                          (i, is_started and 'not ' or ''))
 761
 762
 763     def _assert_agents_not_started(self, indexes):
 764         self._assert_agents_started(indexes, False)
 765
 766
 767     def test_throttle_total(self):
 768         self._setup_some_agents(4)
 769         self._run_a_few_cycles()
 770         self._assert_agents_started([0, 1, 2])
 771         self._assert_agents_not_started([3])
 772
 773
 774     def test_throttle_per_cycle(self):
 775         self._setup_some_agents(3)
 776         self._dispatcher._handle_agents()
 777         self._assert_agents_started([0, 1])
 778         self._assert_agents_not_started([2])
 779
 780
 781     def test_throttle_with_synchronous(self):
 782         self._setup_some_agents(2)
 783         self._agents[0].task.num_processes = 3
 784         self._run_a_few_cycles()
 785         self._assert_agents_started([0])
 786         self._assert_agents_not_started([1])
 787
 788
 789     def test_large_agent_starvation(self):
 790         """
 791         Ensure large agents don't get starved by lower-priority agents.
 792         """
 793         self._setup_some_agents(3)
 794         self._agents[1].task.num_processes = 3
 795         self._run_a_few_cycles()
 796         self._assert_agents_started([0])
 797         self._assert_agents_not_started([1, 2])
 798
 799         self._agents[0].set_done(True)
 800         self._run_a_few_cycles()
 801         self._assert_agents_started([1])
 802         self._assert_agents_not_started([2])
 803
 804
 805     def test_zero_process_agent(self):
 806         self._setup_some_agents(5)
 807         self._agents[4].task.num_processes = 0
 808         self._run_a_few_cycles()
 809         self._assert_agents_started([0, 1, 2, 4])
 810         self._assert_agents_not_started([3])
 811
 812
 813 class PidfileRunMonitorTest(unittest.TestCase):
 814     execution_tag = 'test_tag'
 815     pid = 12345
 816     process = drone_manager.Process('myhost', pid)
 817     num_tests_failed = 1
 818
 819     def setUp(self):
 820         self.god = mock.mock_god()
 821         self.mock_drone_manager = self.god.create_mock_class(
 822             drone_manager.DroneManager, 'drone_manager')
 823         self.god.stub_with(monitor_db, '_drone_manager',
 824                            self.mock_drone_manager)
 825         self.god.stub_function(email_manager.manager, 'enqueue_notify_email')
 826         self.god.stub_with(monitor_db, '_get_pidfile_timeout_secs',
 827                            self._mock_get_pidfile_timeout_secs)
 828
 829         self.pidfile_id = object()
 830
 831         (self.mock_drone_manager.get_pidfile_id_from
 832              .expect_call(self.execution_tag,
 833                           pidfile_name=drone_manager.AUTOSERV_PID_FILE)
 834              .and_return(self.pidfile_id))
 835
 836         self.monitor = monitor_db.PidfileRunMonitor()
 837         self.monitor.attach_to_existing_process(self.execution_tag)
 838
 839     def tearDown(self):
 840         self.god.unstub_all()
 841
 842
 843     def _mock_get_pidfile_timeout_secs(self):
 844         return 300
 845
 846
 847     def setup_pidfile(self, pid=None, exit_code=None, tests_failed=None,
 848                       use_second_read=False):
 849         contents = drone_manager.PidfileContents()
 850         if pid is not None:
 851             contents.process = drone_manager.Process('myhost', pid)
 852         contents.exit_status = exit_code
 853         contents.num_tests_failed = tests_failed
 854         self.mock_drone_manager.get_pidfile_contents.expect_call(
 855             self.pidfile_id, use_second_read=use_second_read).and_return(
 856             contents)
 857
 858
 859     def set_not_yet_run(self):
 860         self.setup_pidfile()
 861
 862
 863     def set_empty_pidfile(self):
 864         self.setup_pidfile()
 865
 866
 867     def set_running(self, use_second_read=False):
 868         self.setup_pidfile(self.pid, use_second_read=use_second_read)
 869
 870
 871     def set_complete(self, error_code, use_second_read=False):
 872         self.setup_pidfile(self.pid, error_code, self.num_tests_failed,
 873                            use_second_read=use_second_read)
 874
 875
 876     def _check_monitor(self, expected_pid, expected_exit_status,
 877                        expected_num_tests_failed):
 878         if expected_pid is None:
 879             self.assertEquals(self.monitor._state.process, None)
 880         else:
 881             self.assertEquals(self.monitor._state.process.pid, expected_pid)
 882         self.assertEquals(self.monitor._state.exit_status, expected_exit_status)
 883         self.assertEquals(self.monitor._state.num_tests_failed,
 884                           expected_num_tests_failed)
 885
 886
 887         self.god.check_playback()
 888
 889
 890     def _test_read_pidfile_helper(self, expected_pid, expected_exit_status,
 891                                   expected_num_tests_failed):
 892         self.monitor._read_pidfile()
 893         self._check_monitor(expected_pid, expected_exit_status,
 894                             expected_num_tests_failed)
 895
 896
 897     def _get_expected_tests_failed(self, expected_exit_status):
 898         if expected_exit_status is None:
 899             expected_tests_failed = None
 900         else:
 901             expected_tests_failed = self.num_tests_failed
 902         return expected_tests_failed
 903
 904
 905     def test_read_pidfile(self):
 906         self.set_not_yet_run()
 907         self._test_read_pidfile_helper(None, None, None)
 908
 909         self.set_empty_pidfile()
 910         self._test_read_pidfile_helper(None, None, None)
 911
 912         self.set_running()
 913         self._test_read_pidfile_helper(self.pid, None, None)
 914
 915         self.set_complete(123)
 916         self._test_read_pidfile_helper(self.pid, 123, self.num_tests_failed)
 917
 918
 919     def test_read_pidfile_error(self):
 920         self.mock_drone_manager.get_pidfile_contents.expect_call(
 921             self.pidfile_id, use_second_read=False).and_return(
 922             drone_manager.InvalidPidfile('error'))
 923         self.assertRaises(monitor_db.PidfileRunMonitor._PidfileException,
 924                           self.monitor._read_pidfile)
 925         self.god.check_playback()
 926
 927
 928     def setup_is_running(self, is_running):
 929         self.mock_drone_manager.is_process_running.expect_call(
 930             self.process).and_return(is_running)
 931
 932
 933     def _test_get_pidfile_info_helper(self, expected_pid, expected_exit_status,
 934                                       expected_num_tests_failed):
 935         self.monitor._get_pidfile_info()
 936         self._check_monitor(expected_pid, expected_exit_status,
 937                             expected_num_tests_failed)
 938
 939
 940     def test_get_pidfile_info(self):
 941         """
 942         normal cases for get_pidfile_info
 943         """
 944         # running
 945         self.set_running()
 946         self.setup_is_running(True)
 947         self._test_get_pidfile_info_helper(self.pid, None, None)
 948
 949         # exited during check
 950         self.set_running()
 951         self.setup_is_running(False)
 952         self.set_complete(123, use_second_read=True) # pidfile gets read again
 953         self._test_get_pidfile_info_helper(self.pid, 123, self.num_tests_failed)
 954
 955         # completed
 956         self.set_complete(123)
 957         self._test_get_pidfile_info_helper(self.pid, 123, self.num_tests_failed)
 958
 959
 960     def test_get_pidfile_info_running_no_proc(self):
 961         """
 962         pidfile shows process running, but no proc exists
 963         """
 964         # running but no proc
 965         self.set_running()
 966         self.setup_is_running(False)
 967         self.set_running(use_second_read=True)
 968         email_manager.manager.enqueue_notify_email.expect_call(
 969             mock.is_string_comparator(), mock.is_string_comparator())
 970         self._test_get_pidfile_info_helper(self.pid, 1, 0)
 971         self.assertTrue(self.monitor.lost_process)
 972
 973
 974     def test_get_pidfile_info_not_yet_run(self):
 975         """
 976         pidfile hasn't been written yet
 977         """
 978         self.set_not_yet_run()
 979         self._test_get_pidfile_info_helper(None, None, None)
 980
 981
 982     def test_process_failed_to_write_pidfile(self):
 983         self.set_not_yet_run()
 984         email_manager.manager.enqueue_notify_email.expect_call(
 985             mock.is_string_comparator(), mock.is_string_comparator())
 986         self.monitor._start_time = (time.time() -
 987                                     monitor_db._get_pidfile_timeout_secs() - 1)
 988         self._test_get_pidfile_info_helper(None, 1, 0)
 989         self.assertTrue(self.monitor.lost_process)
 990
 991
 992 class AgentTest(unittest.TestCase):
 993     def setUp(self):
 994         self.god = mock.mock_god()
 995         self._dispatcher = self.god.create_mock_class(monitor_db.Dispatcher,
 996                                                       'dispatcher')
 997
 998
 999     def tearDown(self):
1000         self.god.unstub_all()
1001
1002
1003     def _create_mock_task(self, name):
1004         task = self.god.create_mock_class(monitor_db.AgentTask, name)
1005         task.num_processes = 1
1006         _set_host_and_qe_ids(task)
1007         return task
1008
1009     def _create_agent(self, task):
1010         agent = monitor_db.Agent(task)
1011         agent.dispatcher = self._dispatcher
1012         return agent
1013
1014
1015     def _finish_agent(self, agent):
1016         while not agent.is_done():
1017             agent.tick()
1018
1019
1020     def test_agent_abort(self):
1021         task = self._create_mock_task('task')
1022         task.poll.expect_call()
1023         task.is_done.expect_call().and_return(False)
1024         task.abort.expect_call()
1025         task.aborted = True
1026
1027         agent = self._create_agent(task)
1028         agent.tick()
1029         agent.abort()
1030         self._finish_agent(agent)
1031         self.god.check_playback()
1032
1033
1034     def _test_agent_abort_before_started_helper(self, ignore_abort=False):
1035         task = self._create_mock_task('task')
1036         task.abort.expect_call()
1037         if ignore_abort:
1038             task.aborted = False
1039             task.poll.expect_call()
1040             task.is_done.expect_call().and_return(True)
1041             task.success = True
1042         else:
1043             task.aborted = True
1044
1045         agent = self._create_agent(task)
1046         agent.abort()
1047         self._finish_agent(agent)
1048         self.god.check_playback()
1049
1050
1051     def test_agent_abort_before_started(self):
1052         self._test_agent_abort_before_started_helper()
1053         self._test_agent_abort_before_started_helper(True)
1054
1055
1056 class JobSchedulingTest(BaseSchedulerTest):
1057     def _test_run_helper(self, expect_agent=True, expect_starting=False,
1058                          expect_pending=False):
1059         if expect_starting:
1060             expected_status = models.HostQueueEntry.Status.STARTING
1061         elif expect_pending:
1062             expected_status = models.HostQueueEntry.Status.PENDING
1063         else:
1064             expected_status = models.HostQueueEntry.Status.VERIFYING
1065         job = scheduler_models.Job.fetch('id = 1')[0]
1066         queue_entry = scheduler_models.HostQueueEntry.fetch('id = 1')[0]
1067         assert queue_entry.job is job
1068         job.run_if_ready(queue_entry)
1069
1070         self.god.check_playback()
1071
1072         self._dispatcher._schedule_delay_tasks()
1073         self._dispatcher._schedule_running_host_queue_entries()
1074         agent = self._dispatcher._agents[0]
1075
1076         actual_status = models.HostQueueEntry.smart_get(1).status
1077         self.assertEquals(expected_status, actual_status)
1078
1079         if not expect_agent:
1080             self.assertEquals(agent, None)
1081             return
1082
1083         self.assert_(isinstance(agent, monitor_db.Agent))
1084         self.assert_(agent.task)
1085         return agent.task
1086
1087
1088     def test_run_if_ready_delays(self):
1089         # Also tests Job.run_with_ready_delay() on atomic group jobs.
1090         django_job = self._create_job(hosts=[5, 6], atomic_group=1)
1091         job = scheduler_models.Job(django_job.id)
1092         self.assertEqual(1, job.synch_count)
1093         django_hqes = list(models.HostQueueEntry.objects.filter(job=job.id))
1094         self.assertEqual(2, len(django_hqes))
1095         self.assertEqual(2, django_hqes[0].atomic_group.max_number_of_machines)
1096
1097         def set_hqe_status(django_hqe, status):
1098             django_hqe.status = status
1099             django_hqe.save()
1100             scheduler_models.HostQueueEntry(django_hqe.id).host.set_status(status)
1101
1102         # An initial state, our synch_count is 1
1103         set_hqe_status(django_hqes[0], models.HostQueueEntry.Status.VERIFYING)
1104         set_hqe_status(django_hqes[1], models.HostQueueEntry.Status.PENDING)
1105
1106         # So that we don't depend on the config file value during the test.
1107         self.assert_(scheduler_config.config
1108                      .secs_to_wait_for_atomic_group_hosts is not None)
1109         self.god.stub_with(scheduler_config.config,
1110                            'secs_to_wait_for_atomic_group_hosts', 123456)
1111
1112         # Get the pending one as a scheduler_models.HostQueueEntry object.
1113         hqe = scheduler_models.HostQueueEntry(django_hqes[1].id)
1114         self.assert_(not job._delay_ready_task)
1115         self.assertTrue(job.is_ready())
1116
1117         # Ready with one pending, one verifying and an atomic group should
1118         # result in a DelayCallTask to re-check if we're ready a while later.
1119         job.run_if_ready(hqe)
1120         self.assertEquals('Waiting', hqe.status)
1121         self._dispatcher._schedule_delay_tasks()
1122         self.assertEquals('Pending', hqe.status)
1123         agent = self._dispatcher._agents[0]
1124         self.assert_(job._delay_ready_task)
1125         self.assert_(isinstance(agent, monitor_db.Agent))
1126         self.assert_(agent.task)
1127         delay_task = agent.task
1128         self.assert_(isinstance(delay_task, scheduler_models.DelayedCallTask))
1129         self.assert_(not delay_task.is_done())
1130
1131         self.god.stub_function(delay_task, 'abort')
1132
1133         self.god.stub_function(job, 'run')
1134
1135         self.god.stub_function(job, '_pending_count')
1136         self.god.stub_with(job, 'synch_count', 9)
1137         self.god.stub_function(job, 'request_abort')
1138
1139         # Test that the DelayedCallTask's callback queued up above does the
1140         # correct thing and does not call run if there are not enough hosts
1141         # in pending after the delay.
1142         job._pending_count.expect_call().and_return(0)
1143         job.request_abort.expect_call()
1144         delay_task._callback()
1145         self.god.check_playback()
1146
1147         # Test that the DelayedCallTask's callback queued up above does the
1148         # correct thing and returns the Agent returned by job.run() if
1149         # there are still enough hosts pending after the delay.
1150         job.synch_count = 4
1151         job._pending_count.expect_call().and_return(4)
1152         job.run.expect_call(hqe)
1153         delay_task._callback()
1154         self.god.check_playback()
1155
1156         job._pending_count.expect_call().and_return(4)
1157
1158         # Adjust the delay deadline so that enough time has passed.
1159         job._delay_ready_task.end_time = time.time() - 111111
1160         job.run.expect_call(hqe)
1161         # ...the delay_expired condition should cause us to call run()
1162         self._dispatcher._handle_agents()
1163         self.god.check_playback()
1164         delay_task.success = False
1165
1166         # Adjust the delay deadline back so that enough time has not passed.
1167         job._delay_ready_task.end_time = time.time() + 111111
1168         self._dispatcher._handle_agents()
1169         self.god.check_playback()
1170
1171         # Now max_number_of_machines HQEs are in pending state.  Remaining
1172         # delay will now be ignored.
1173         other_hqe = scheduler_models.HostQueueEntry(django_hqes[0].id)
1174         self.god.unstub(job, 'run')
1175         self.god.unstub(job, '_pending_count')
1176         self.god.unstub(job, 'synch_count')
1177         self.god.unstub(job, 'request_abort')
1178         # ...the over_max_threshold test should cause us to call run()
1179         delay_task.abort.expect_call()
1180         other_hqe.on_pending()
1181         self.assertEquals('Starting', other_hqe.status)
1182         self.assertEquals('Starting', hqe.status)
1183         self.god.stub_function(job, 'run')
1184         self.god.unstub(delay_task, 'abort')
1185
1186         hqe.set_status('Pending')
1187         other_hqe.set_status('Pending')
1188         # Now we're not over the max for the atomic group.  But all assigned
1189         # hosts are in pending state.  over_max_threshold should make us run().
1190         hqe.atomic_group.max_number_of_machines += 1
1191         hqe.atomic_group.save()
1192         job.run.expect_call(hqe)
1193         hqe.on_pending()
1194         self.god.check_playback()
1195         hqe.atomic_group.max_number_of_machines -= 1
1196         hqe.atomic_group.save()
1197
1198         other_hqe = scheduler_models.HostQueueEntry(django_hqes[0].id)
1199         self.assertTrue(hqe.job is other_hqe.job)
1200         # DBObject classes should reuse instances so these should be the same.
1201         self.assertEqual(job, other_hqe.job)
1202         self.assertEqual(other_hqe.job, hqe.job)
1203         # Be sure our delay was not lost during the other_hqe construction.
1204         self.assertEqual(job._delay_ready_task, delay_task)
1205         self.assert_(job._delay_ready_task)
1206         self.assertFalse(job._delay_ready_task.is_done())
1207         self.assertFalse(job._delay_ready_task.aborted)
1208
1209         # We want the real run() to be called below.
1210         self.god.unstub(job, 'run')
1211
1212         # We pass in the other HQE this time the same way it would happen
1213         # for real when one host finishes verifying and enters pending.
1214         job.run_if_ready(other_hqe)
1215
1216         # The delayed task must be aborted by the actual run() call above.
1217         self.assertTrue(job._delay_ready_task.aborted)
1218         self.assertFalse(job._delay_ready_task.success)
1219         self.assertTrue(job._delay_ready_task.is_done())
1220
1221         # Check that job run() and _finish_run() were called by the above:
1222         self._dispatcher._schedule_running_host_queue_entries()
1223         agent = self._dispatcher._agents[0]
1224         self.assert_(agent.task)
1225         task = agent.task
1226         self.assert_(isinstance(task, monitor_db.QueueTask))
1227         # Requery these hqes in order to verify the status from the DB.
1228         django_hqes = list(models.HostQueueEntry.objects.filter(job=job.id))
1229         for entry in django_hqes:
1230             self.assertEqual(models.HostQueueEntry.Status.STARTING,
1231                              entry.status)
1232
1233         # We're already running, but more calls to run_with_ready_delay can
1234         # continue to come in due to straggler hosts enter Pending.  Make
1235         # sure we don't do anything.
1236         self.god.stub_function(job, 'run')
1237         job.run_with_ready_delay(hqe)
1238         self.god.check_playback()
1239         self.god.unstub(job, 'run')
1240
1241
1242     def test_run_synchronous_atomic_group_ready(self):
1243         self._create_job(hosts=[5, 6], atomic_group=1, synchronous=True)
1244         self._update_hqe("status='Pending', execution_subdir=''")
1245
1246         queue_task = self._test_run_helper(expect_starting=True)
1247
1248         self.assert_(isinstance(queue_task, monitor_db.QueueTask))
1249         # Atomic group jobs that do not depend on a specific label in the
1250         # atomic group will use the atomic group name as their group name.
1251         self.assertEquals(queue_task.queue_entries[0].get_group_name(),
1252                           'atomic1')
1253
1254
1255     def test_run_synchronous_atomic_group_with_label_ready(self):
1256         job = self._create_job(hosts=[5, 6], atomic_group=1, synchronous=True)
1257         job.dependency_labels.add(self.label4)
1258         self._update_hqe("status='Pending', execution_subdir=''")
1259
1260         queue_task = self._test_run_helper(expect_starting=True)
1261
1262         self.assert_(isinstance(queue_task, monitor_db.QueueTask))
1263         # Atomic group jobs that also specify a label in the atomic group
1264         # will use the label name as their group name.
1265         self.assertEquals(queue_task.queue_entries[0].get_group_name(),
1266                           'label4')
1267
1268
1269     def test_run_synchronous_ready(self):
1270         self._create_job(hosts=[1, 2], synchronous=True)
1271         self._update_hqe("status='Pending', execution_subdir=''")
1272
1273         queue_task = self._test_run_helper(expect_starting=True)
1274
1275         self.assert_(isinstance(queue_task, monitor_db.QueueTask))
1276         self.assertEquals(queue_task.job.id, 1)
1277         hqe_ids = [hqe.id for hqe in queue_task.queue_entries]
1278         self.assertEquals(hqe_ids, [1, 2])
1279
1280
1281     def test_schedule_running_host_queue_entries_fail(self):
1282         self._create_job(hosts=[2])
1283         self._update_hqe("status='%s', execution_subdir=''" %
1284                          models.HostQueueEntry.Status.PENDING)
1285         job = scheduler_models.Job.fetch('id = 1')[0]
1286         queue_entry = scheduler_models.HostQueueEntry.fetch('id = 1')[0]
1287         assert queue_entry.job is job
1288         job.run_if_ready(queue_entry)
1289         self.assertEqual(queue_entry.status,
1290                          models.HostQueueEntry.Status.STARTING)
1291         self.assert_(queue_entry.execution_subdir)
1292         self.god.check_playback()
1293
1294         class dummy_test_agent(object):
1295             task = 'dummy_test_agent'
1296         self._dispatcher._register_agent_for_ids(
1297                 self._dispatcher._host_agents, [queue_entry.host.id],
1298                 dummy_test_agent)
1299
1300         # Attempted to schedule on a host that already has an agent.
1301         self.assertRaises(host_scheduler.SchedulerError,
1302                           self._dispatcher._schedule_running_host_queue_entries)
1303
1304
1305     def test_schedule_hostless_job(self):
1306         job = self._create_job(hostless=True)
1307         self.assertEqual(1, job.hostqueueentry_set.count())
1308         hqe_query = scheduler_models.HostQueueEntry.fetch(
1309                 'id = %s' % job.hostqueueentry_set.all()[0].id)
1310         self.assertEqual(1, len(hqe_query))
1311         hqe = hqe_query[0]
1312
1313         self.assertEqual(models.HostQueueEntry.Status.QUEUED, hqe.status)
1314         self.assertEqual(0, len(self._dispatcher._agents))
1315
1316         self._dispatcher._schedule_new_jobs()
1317
1318         self.assertEqual(models.HostQueueEntry.Status.STARTING, hqe.status)
1319         self.assertEqual(1, len(self._dispatcher._agents))
1320
1321         self._dispatcher._schedule_new_jobs()
1322
1323         # No change to previously schedule hostless job, and no additional agent
1324         self.assertEqual(models.HostQueueEntry.Status.STARTING, hqe.status)
1325         self.assertEqual(1, len(self._dispatcher._agents))
1326
1327
1328 class TopLevelFunctionsTest(unittest.TestCase):
1329     def setUp(self):
1330         self.god = mock.mock_god()
1331
1332
1333     def tearDown(self):
1334         self.god.unstub_all()
1335
1336
1337     def test_autoserv_command_line(self):
1338         machines = 'abcd12,efgh34'
1339         extra_args = ['-Z', 'hello']
1340         expected_command_line_base = set((monitor_db._autoserv_path, '-p',
1341                                           '-m', machines, '-r',
1342                                           drone_manager.WORKING_DIRECTORY))
1343
1344         expected_command_line = expected_command_line_base.union(
1345                 ['--verbose']).union(extra_args)
1346         command_line = set(
1347                 monitor_db._autoserv_command_line(machines, extra_args))
1348         self.assertEqual(expected_command_line, command_line)
1349
1350         class FakeJob(object):
1351             owner = 'Bob'
1352             name = 'fake job name'
1353             id = 1337
1354
1355         class FakeHQE(object):
1356             job = FakeJob
1357
1358         expected_command_line = expected_command_line_base.union(
1359                 ['-u', FakeJob.owner, '-l', FakeJob.name])
1360         command_line = set(monitor_db._autoserv_command_line(
1361                 machines, extra_args=[], queue_entry=FakeHQE, verbose=False))
1362         self.assertEqual(expected_command_line, command_line)
1363
1364
1365 class AgentTaskTest(unittest.TestCase,
1366                     frontend_test_utils.FrontendTestMixin):
1367     def setUp(self):
1368         self._frontend_common_setup()
1369
1370
1371     def tearDown(self):
1372         self._frontend_common_teardown()
1373
1374
1375     def _setup_drones(self):
1376         self.god.stub_function(models.DroneSet, 'drone_sets_enabled')
1377         models.DroneSet.drone_sets_enabled.expect_call().and_return(True)
1378
1379         drones = []
1380         for x in xrange(4):
1381             drones.append(models.Drone.objects.create(hostname=str(x)))
1382
1383         drone_set_1 = models.DroneSet.objects.create(name='1')
1384         drone_set_1.drones.add(*drones[0:2])
1385         drone_set_2 = models.DroneSet.objects.create(name='2')
1386         drone_set_2.drones.add(*drones[2:4])
1387         drone_set_3 = models.DroneSet.objects.create(name='3')
1388
1389         job_1 = self._create_job_simple([self.hosts[0].id],
1390                                         drone_set=drone_set_1)
1391         job_2 = self._create_job_simple([self.hosts[0].id],
1392                                         drone_set=drone_set_2)
1393         job_3 = self._create_job_simple([self.hosts[0].id],
1394                                         drone_set=drone_set_3)
1395
1396         job_4 = self._create_job_simple([self.hosts[0].id])
1397         job_4.drone_set = None
1398         job_4.save()
1399
1400         hqe_1 = job_1.hostqueueentry_set.all()[0]
1401         hqe_2 = job_2.hostqueueentry_set.all()[0]
1402         hqe_3 = job_3.hostqueueentry_set.all()[0]
1403         hqe_4 = job_4.hostqueueentry_set.all()[0]
1404
1405         return (hqe_1, hqe_2, hqe_3, hqe_4), monitor_db.AgentTask()
1406
1407
1408     def test_get_drone_hostnames_allowed_no_drones_in_set(self):
1409         hqes, task = self._setup_drones()
1410         task.queue_entry_ids = (hqes[2].id,)
1411         self.assertEqual(set(), task.get_drone_hostnames_allowed())
1412         self.god.check_playback()
1413
1414
1415     def test_get_drone_hostnames_allowed_no_drone_set(self):
1416         hqes, task = self._setup_drones()
1417         hqe = hqes[3]
1418         task.queue_entry_ids = (hqe.id,)
1419
1420         result = object()
1421
1422         self.god.stub_function(task, '_user_or_global_default_drone_set')
1423         task._user_or_global_default_drone_set.expect_call(
1424                 hqe.job, hqe.job.user()).and_return(result)
1425
1426         self.assertEqual(result, task.get_drone_hostnames_allowed())
1427         self.god.check_playback()
1428
1429
1430     def test_get_drone_hostnames_allowed_success(self):
1431         hqes, task = self._setup_drones()
1432         task.queue_entry_ids = (hqes[0].id,)
1433         self.assertEqual(set(('0','1')), task.get_drone_hostnames_allowed())
1434         self.god.check_playback()
1435
1436
1437     def test_get_drone_hostnames_allowed_multiple_jobs(self):
1438         hqes, task = self._setup_drones()
1439         task.queue_entry_ids = (hqes[0].id, hqes[1].id)
1440         self.assertRaises(AssertionError,
1441                           task.get_drone_hostnames_allowed)
1442         self.god.check_playback()
1443
1444
1445     def test_get_drone_hostnames_allowed_no_hqe(self):
1446         class MockSpecialTask(object):
1447             requested_by = object()
1448
1449         class MockSpecialAgentTask(monitor_db.SpecialAgentTask):
1450             task = MockSpecialTask()
1451             queue_entry_ids = []
1452             def __init__(self, *args, **kwargs):
1453                 pass
1454
1455         task = MockSpecialAgentTask()
1456         self.god.stub_function(models.DroneSet, 'drone_sets_enabled')
1457         self.god.stub_function(task, '_user_or_global_default_drone_set')
1458
1459         result = object()
1460         models.DroneSet.drone_sets_enabled.expect_call().and_return(True)
1461         task._user_or_global_default_drone_set.expect_call(
1462                 task.task, MockSpecialTask.requested_by).and_return(result)
1463
1464         self.assertEqual(result, task.get_drone_hostnames_allowed())
1465         self.god.check_playback()
1466
1467
1468     def _setup_test_user_or_global_default_drone_set(self):
1469         result = object()
1470         class MockDroneSet(object):
1471             def get_drone_hostnames(self):
1472                 return result
1473
1474         self.god.stub_function(models.DroneSet, 'get_default')
1475         models.DroneSet.get_default.expect_call().and_return(MockDroneSet())
1476         return result
1477
1478
1479     def test_user_or_global_default_drone_set(self):
1480         expected = object()
1481         class MockDroneSet(object):
1482             def get_drone_hostnames(self):
1483                 return expected
1484         class MockUser(object):
1485             drone_set = MockDroneSet()
1486
1487         self._setup_test_user_or_global_default_drone_set()
1488
1489         actual = monitor_db.AgentTask()._user_or_global_default_drone_set(
1490                 None, MockUser())
1491
1492         self.assertEqual(expected, actual)
1493         self.god.check_playback()
1494
1495
1496     def test_user_or_global_default_drone_set_no_user(self):
1497         expected = self._setup_test_user_or_global_default_drone_set()
1498         actual = monitor_db.AgentTask()._user_or_global_default_drone_set(
1499                 None, None)
1500
1501         self.assertEqual(expected, actual)
1502         self.god.check_playback()
1503
1504
1505     def test_user_or_global_default_drone_set_no_user_drone_set(self):
1506         class MockUser(object):
1507             drone_set = None
1508             login = None
1509
1510         expected = self._setup_test_user_or_global_default_drone_set()
1511         actual = monitor_db.AgentTask()._user_or_global_default_drone_set(
1512                 None, MockUser())
1513
1514         self.assertEqual(expected, actual)
1515         self.god.check_playback()
1516
1517
1518 if __name__ == '__main__':
1519     unittest.main()