def test_explicit_group(self):
group_spec = self.minimal_group()
job_a = util.minimal_job(group=group_spec["uuid"])
job_b = util.minimal_job(group=group_spec["uuid"])
data = {'jobs': [job_a, job_b], 'groups': [group_spec]}
resp = util.session.post('%s/rawscheduler' % self.cook_url, json=data)
self.assertEqual(resp.status_code, 201)
jobs = util.session.get('%s/rawscheduler?job=%s&job=%s' %
(self.cook_url, job_a['uuid'], job_b['uuid']))
self.assertEqual(200, jobs.status_code)
jobs = jobs.json()
self.assertEqual(group_spec['uuid'], jobs[0]['groups'][0])
self.assertEqual(group_spec['uuid'], jobs[1]['groups'][0])
util.wait_for_job(self.cook_url, job_a['uuid'], 'completed')
util.wait_for_job(self.cook_url, job_b['uuid'], 'completed')
def test_list_jobs_by_time(self):
# schedule two jobs with different submit times
job_specs = [util.minimal_job() for _ in range(2)]
request_body = {'jobs': [job_specs[0]]}
resp = util.session.post('%s/rawscheduler' % self.cook_url, json=request_body)
self.assertEqual(resp.status_code, 201)
time.sleep(1)
request_body = {'jobs': [job_specs[1]]}
resp = util.session.post('%s/rawscheduler' % self.cook_url, json=request_body)
self.assertEqual(resp.status_code, 201)
submit_times = [self.get_job(job_spec['uuid'])['submit_time'] for job_spec in job_specs]
user = self.determine_user()
# start-ms and end-ms are exclusive
# query where start-ms and end-ms are the submit times of jobs 1 & 2 respectively
resp = util.session.get('%s/list?user=%s&state=waiting&start-ms=%s&end-ms=%s'
% (self.cook_url, user, submit_times[0] - 1, submit_times[1] + 1))
self.assertEqual(200, resp.status_code)
jobs = resp.json()
self.assertTrue(any(job for job in jobs if job['uuid'] == job_specs[0]['uuid']))
self.assertTrue(any(job for job in jobs if job['uuid'] == job_specs[1]['uuid']))
# query just for job 1
resp = util.session.get('%s/list?user=%s&state=waiting&start-ms=%s&end-ms=%s'
% (self.cook_url, user, submit_times[0] - 1, submit_times[1]))
self.assertEqual(200, resp.status_code)
jobs = resp.json()
self.assertTrue(any(job for job in jobs if job['uuid'] == job_specs[0]['uuid']))
self.assertFalse(any(job for job in jobs if job['uuid'] == job_specs[1]['uuid']))
# query just for job 2
resp = util.session.get('%s/list?user=%s&state=waiting&start-ms=%s&end-ms=%s'
% (self.cook_url, user, submit_times[0], submit_times[1] + 1))
self.assertEqual(200, resp.status_code)
jobs = resp.json()
self.assertFalse(any(job for job in jobs if job['uuid'] == job_specs[0]['uuid']))
self.assertTrue(any(job for job in jobs if job['uuid'] == job_specs[1]['uuid']))
# query for neither
resp = util.session.get('%s/list?user=%s&state=waiting&start-ms=%s&end-ms=%s'
% (self.cook_url, user, submit_times[0], submit_times[1]))
self.assertEqual(200, resp.status_code)
jobs = resp.json()
self.assertFalse(any(job for job in jobs if job['uuid'] == job_specs[0]['uuid']))
self.assertFalse(any(job for job in jobs if job['uuid'] == job_specs[1]['uuid']))
def test_straggler_handling(self):
straggler_handling = {
'type': 'quantile-deviation',
'parameters': {
'quantile': 0.5,
'multiplier': 2.0
}
}
group_spec = self.minimal_group(straggler_handling=straggler_handling)
job_fast = util.minimal_job(group=group_spec["uuid"])
job_slow = util.minimal_job(group=group_spec["uuid"], command='sleep 120')
data = {'jobs': [job_fast, job_slow], 'groups': [group_spec]}
resp = util.session.post('%s/rawscheduler' % self.cook_url, json=data)
self.assertEqual(resp.status_code, 201)
util.wait_for_job(self.cook_url, job_fast['uuid'], 'completed')
util.wait_for_job(self.cook_url, job_slow['uuid'], 'completed')
jobs = util.session.get('%s/rawscheduler?job=%s&job=%s' %
(self.cook_url, job_fast['uuid'], job_slow['uuid']))
self.assertEqual(200, jobs.status_code)
jobs = jobs.json()
self.logger.debug('Loaded jobs %s', jobs)
self.assertEqual('success', jobs[0]['state'])
self.assertEqual('failed', jobs[1]['state'])
self.assertEqual(2004, jobs[1]['instances'][0]['reason_code'])
def test_list_jobs_by_state(self):
# schedule a bunch of jobs in hopes of getting jobs into different statuses
request_body = {'jobs': [util.minimal_job(command="sleep %s" % i) for i in range(1, 20)]}
resp = util.session.post('%s/rawscheduler' % self.cook_url, json=request_body)
self.assertEqual(resp.status_code, 201)
# let some jobs get scheduled
time.sleep(10)
user = self.determine_user()
for state in ['waiting', 'running', 'completed']:
resp = util.session.get('%s/list?user=%s&state=%s' % (self.cook_url, user, state))
self.assertEqual(200, resp.status_code)
jobs = resp.json()
for job in jobs:
# print "%s %s" % (job['uuid'], job['status'])
self.assertEquals(state, job['status'])
def test_get_job(self):
# schedule a job
job_spec = util.minimal_job()
resp = util.session.post('%s/rawscheduler' % self.cook_url, json={'jobs': [job_spec]})
self.assertEqual(201, resp.status_code)
# query for the same job & ensure the response has what it's supposed to have
job = util.wait_for_job(self.cook_url, job_spec['uuid'], 'completed')
self.assertEquals(job_spec['mem'], job['mem'])
self.assertEquals(job_spec['max_retries'], job['max_retries'])
self.assertEquals(job_spec['name'], job['name'])
self.assertEquals(job_spec['priority'], job['priority'])
self.assertEquals(job_spec['uuid'], job['uuid'])
self.assertEquals(job_spec['cpus'], job['cpus'])
self.assertTrue('labels' in job)
self.assertEquals(9223372036854775807, job['max_runtime'])
# 9223372036854775807 is MAX_LONG(ish), the default value for max_runtime
self.assertEquals('success', job['state'])
self.assertTrue('env' in job)
self.assertTrue('framework_id' in job)
self.assertTrue('ports' in job)
self.assertTrue('instances' in job)
self.assertEquals('completed', job['status'])
self.assertTrue(isinstance(job['submit_time'], int))
self.assertTrue('uris' in job)
self.assertTrue('retries_remaining' in job)
instance = job['instances'][0]
self.assertTrue(isinstance(instance['start_time'], int))
self.assertTrue('executor_id' in instance)
self.assertTrue('hostname' in instance)
self.assertTrue('slave_id' in instance)
self.assertTrue(isinstance(instance['preempted'], bool))
self.assertTrue(isinstance(instance['end_time'], int))
self.assertTrue(isinstance(instance['backfilled'], bool))
self.assertTrue('ports' in instance)
self.assertEquals('completed', job['status'])
self.assertTrue('task_id' in instance)
def test_error_while_creating_job(self):
job1 = util.minimal_job()
job2 = util.minimal_job(uuid=job1['uuid'])
resp = util.session.post('%s/rawscheduler' % self.cook_url,
json={'jobs':[job1, job2]})
self.assertEqual(resp.status_code, 500)
def determine_user(self):
job_spec = util.minimal_job()
request_body = {'jobs': [job_spec]}
resp = util.session.post('%s/rawscheduler' % self.cook_url, json=request_body)
self.assertEqual(resp.status_code, 201)
return self.get_job(job_spec['uuid'])['user']