def test_build_workflow_for_job_pig(self, job_binary):
job, job_exec = u.create_job_exec(edp.JOB_TYPE_PIG, configs={})
job_binary.return_value = {"name": "script.pig"}
input_data = u.create_data_source('swift://ex/i')
output_data = u.create_data_source('swift://ex/o')
data_source_urls = {input_data.id: input_data.url,
output_data.id: output_data.url}
res = workflow_factory.get_workflow_xml(
job, u.create_cluster(), job_exec.job_configs,
input_data, output_data, 'hadoop', data_source_urls)
self.assertIn("""
<param>INPUT=swift://ex.sahara/i</param>
<param>OUTPUT=swift://ex.sahara/o</param>""", res)
self.assertIn("""
<configuration>
<property>
<name>fs.swift.service.sahara.password</name>
<value>admin1</value>
</property>
<property>
<name>fs.swift.service.sahara.username</name>
<value>admin</value>
</property>
</configuration>""", res)
self.assertIn("<script>script.pig</script>", res)
# testing workflow creation with a proxy domain
self.override_config('use_domain_for_proxy_users', True)
self.override_config("proxy_user_domain_name", 'sahara_proxy_domain')
job, job_exec = u.create_job_exec(edp.JOB_TYPE_PIG, proxy=True)
res = workflow_factory.get_workflow_xml(
job, u.create_cluster(), job_exec.job_configs,
input_data, output_data, 'hadoop', data_source_urls)
self.assertIn("""
<configuration>
<property>
<name>fs.swift.service.sahara.domain.name</name>
<value>sahara_proxy_domain</value>
</property>
<property>
<name>fs.swift.service.sahara.password</name>
<value>55555555-6666-7777-8888-999999999999</value>
</property>
<property>
<name>fs.swift.service.sahara.trust.id</name>
<value>0123456789abcdef0123456789abcdef</value>
</property>
<property>
<name>fs.swift.service.sahara.username</name>
<value>job_00000000-1111-2222-3333-4444444444444444</value>
</property>
</configuration>""", res)
def test_build_workflow_swift_configs(self, job_binary):
# Test that swift configs come from either input or output data sources
job, job_exec = u.create_job_exec(edp.JOB_TYPE_PIG, configs={})
job_binary.return_value = {"name": "script.pig"}
input_data = u.create_data_source('swift://ex/i')
output_data = u.create_data_source('hdfs://user/hadoop/out')
res = workflow_factory.get_workflow_xml(
job, u.create_cluster(), job_exec, input_data, output_data,
'hadoop')
self.assertIn("""
<configuration>
<property>
<name>fs.swift.service.sahara.password</name>
<value>admin1</value>
</property>
<property>
<name>fs.swift.service.sahara.username</name>
<value>admin</value>
</property>
</configuration>""", res)
input_data = u.create_data_source('hdfs://user/hadoop/in')
output_data = u.create_data_source('swift://ex/o')
res = workflow_factory.get_workflow_xml(
job, u.create_cluster(), job_exec, input_data, output_data,
'hadoop')
self.assertIn("""
<configuration>
<property>
<name>fs.swift.service.sahara.password</name>
<value>admin1</value>
</property>
<property>
<name>fs.swift.service.sahara.username</name>
<value>admin</value>
</property>
</configuration>""", res)
job, job_exec = u.create_job_exec(
edp.JOB_TYPE_PIG, configs={'configs': {'dummy': 'value'}})
input_data = u.create_data_source('hdfs://user/hadoop/in')
output_data = u.create_data_source('hdfs://user/hadoop/out')
res = workflow_factory.get_workflow_xml(
job, u.create_cluster(), job_exec, input_data, output_data,
'hadoop')
self.assertIn("""
<configuration>
<property>
<name>dummy</name>
<value>value</value>
</property>
</configuration>""", res)
def test_build_workflow_for_job_hive(self, job_binary):
job, job_exec = u.create_job_exec(edp.JOB_TYPE_HIVE, configs={})
job_binary.return_value = {"name": "script.q"}
input_data = u.create_data_source('swift://ex/i')
output_data = u.create_data_source('swift://ex/o')
data_source_urls = {input_data.id: input_data.url,
output_data.id: output_data.url}
res = workflow_factory.get_workflow_xml(
job, u.create_cluster(), job_exec.job_configs,
input_data, output_data, 'hadoop', data_source_urls)
doc = xml.parseString(res)
hive = doc.getElementsByTagName('hive')[0]
self.assertEqual('/user/hadoop/conf/hive-site.xml',
xmlutils.get_text_from_node(hive, 'job-xml'))
configuration = hive.getElementsByTagName('configuration')
properties = xmlutils.get_property_dict(configuration[0])
self.assertEqual({'fs.swift.service.sahara.password': '******',
'fs.swift.service.sahara.username': '******'},
properties)
self.assertEqual('script.q',
xmlutils.get_text_from_node(hive, 'script'))
params = xmlutils.get_param_dict(hive)
self.assertEqual({'INPUT': 'swift://ex.sahara/i',
'OUTPUT': 'swift://ex.sahara/o'}, params)
# testing workflow creation with a proxy domain
self.override_config('use_domain_for_proxy_users', True)
self.override_config("proxy_user_domain_name", 'sahara_proxy_domain')
job, job_exec = u.create_job_exec(edp.JOB_TYPE_HIVE, proxy=True)
res = workflow_factory.get_workflow_xml(
job, u.create_cluster(), job_exec.job_configs,
input_data, output_data, 'hadoop', data_source_urls)
doc = xml.parseString(res)
hive = doc.getElementsByTagName('hive')[0]
configuration = hive.getElementsByTagName('configuration')
properties = xmlutils.get_property_dict(configuration[0])
self.assertEqual({
'fs.swift.service.sahara.domain.name':
'sahara_proxy_domain',
'fs.swift.service.sahara.trust.id':
'0123456789abcdef0123456789abcdef',
'fs.swift.service.sahara.password':
'******',
'fs.swift.service.sahara.username':
'******'}, properties)
def test_build_workflow_for_job_hive(self, job_binary):
job, job_exec = u.create_job_exec(edp.JOB_TYPE_HIVE, configs={})
job_binary.return_value = {"name": "script.q"}
input_data = u.create_data_source("swift://ex/i")
output_data = u.create_data_source("swift://ex/o")
data_source_urls = {input_data.id: input_data.url, output_data.id: output_data.url}
res = workflow_factory.get_workflow_xml(
job, u.create_cluster(), job_exec.job_configs, input_data, output_data, "hadoop", data_source_urls
)
doc = xml.parseString(res)
hive = doc.getElementsByTagName("hive")[0]
self.assertEqual("/user/hadoop/conf/hive-site.xml", xmlutils.get_text_from_node(hive, "job-xml"))
configuration = hive.getElementsByTagName("configuration")
properties = xmlutils.get_property_dict(configuration[0])
self.assertEqual(
{"fs.swift.service.sahara.password": "******", "fs.swift.service.sahara.username": "******"}, properties
)
self.assertEqual("script.q", xmlutils.get_text_from_node(hive, "script"))
params = xmlutils.get_param_dict(hive)
self.assertEqual({"INPUT": "swift://ex.sahara/i", "OUTPUT": "swift://ex.sahara/o"}, params)
# testing workflow creation with a proxy domain
self.override_config("use_domain_for_proxy_users", True)
self.override_config("proxy_user_domain_name", "sahara_proxy_domain")
job, job_exec = u.create_job_exec(edp.JOB_TYPE_HIVE, proxy=True)
res = workflow_factory.get_workflow_xml(
job, u.create_cluster(), job_exec.job_configs, input_data, output_data, "hadoop", data_source_urls
)
doc = xml.parseString(res)
hive = doc.getElementsByTagName("hive")[0]
configuration = hive.getElementsByTagName("configuration")
properties = xmlutils.get_property_dict(configuration[0])
self.assertEqual(
{
"fs.swift.service.sahara.domain.name": "sahara_proxy_domain",
"fs.swift.service.sahara.trust.id": "0123456789abcdef0123456789abcdef",
"fs.swift.service.sahara.password": "******",
"fs.swift.service.sahara.username": "******",
},
properties,
)
def test_hdfs_upload_job_files(self, conductor_raw_data, remote_class,
remote):
remote_class.__exit__.return_value = 'closed'
remote.return_value = remote_class
conductor_raw_data.return_value = 'ok'
oje = FakeOozieJobEngine(u.create_cluster())
job, _ = u.create_job_exec(edp.JOB_TYPE_PIG)
res = oje._upload_job_files_to_hdfs(mock.Mock(), 'job_prefix', job, {})
self.assertEqual(['job_prefix/script.pig'], res)
job, _ = u.create_job_exec(edp.JOB_TYPE_MAPREDUCE)
res = oje._upload_job_files_to_hdfs(mock.Mock(), 'job_prefix', job, {})
self.assertEqual(['job_prefix/lib/main.jar'], res)
def test_prepare_run_job(self, job, data_source, update,
remote, wf_factory, get_ds_urls,
prepare_cluster):
wf_factory.return_value = mock.MagicMock()
remote_class = mock.MagicMock()
remote_class.__exit__.return_value = 'closed'
remote.return_value = remote_class
job_class = mock.MagicMock()
job_class.name = "myJob"
job.return_value = job_class
source = mock.MagicMock()
source.url = "localhost"
get_ds_urls.return_value = ('url', 'url')
data_source.return_value = source
oje = FakeOozieJobEngine(u.create_cluster())
_, job_exec = u.create_job_exec(edp.JOB_TYPE_PIG)
update.return_value = job_exec
res = oje._prepare_run_job(job_exec)
self.assertEqual(ctx.ctx(), res['context'])
self.assertEqual('hadoop', res['hdfs_user'])
self.assertEqual(job_exec, res['job_execution'])
self.assertEqual({}, res['oozie_params'])
def test_job_type_supported(self, job_get):
job, job_exec = u.create_job_exec(edp.JOB_TYPE_PIG)
job_get.return_value = job
self.assertIsNotNone(job_manager._get_job_engine(u.create_cluster(), job_exec))
job.type = "unsupported_type"
self.assertIsNone(job_manager._get_job_engine(u.create_cluster(), job_exec))
def test_build_workflow_for_job_java(self):
# If args include swift paths, user and password values
# will have to be supplied via configs instead of being
# lifted from input or output data sources
configs = {sw.HADOOP_SWIFT_USERNAME: '******',
sw.HADOOP_SWIFT_PASSWORD: '******'}
configs = {
'configs': configs,
'args': ['swift://ex/i',
'output_path']
}
job, job_exec = u.create_job_exec(edp.JOB_TYPE_JAVA, configs)
res = workflow_factory.get_workflow_xml(
job, u.create_cluster(), job_exec)
self.assertIn("""
<configuration>
<property>
<name>fs.swift.service.sahara.password</name>
<value>admin1</value>
</property>
<property>
<name>fs.swift.service.sahara.username</name>
<value>admin</value>
</property>
</configuration>
<main-class>%s</main-class>
<java-opts>%s</java-opts>
<arg>swift://ex.sahara/i</arg>
<arg>output_path</arg>""" % (_java_main_class, _java_opts), res)
def test_build_workflow_for_job_hive(self, job_binary):
job, job_exec = u.create_job_exec(edp.JOB_TYPE_HIVE)
job_binary.return_value = {"name": "script.q"}
input_data = u.create_data_source('swift://ex/i')
output_data = u.create_data_source('swift://ex/o')
res = workflow_factory.get_workflow_xml(
job, u.create_cluster(), job_exec, input_data, output_data,
'hadoop')
self.assertIn("""
<job-xml>/user/hadoop/conf/hive-site.xml</job-xml>
<configuration>
<property>
<name>fs.swift.service.sahara.password</name>
<value>admin1</value>
</property>
<property>
<name>fs.swift.service.sahara.username</name>
<value>admin</value>
</property>
</configuration>
<script>script.q</script>
<param>INPUT=swift://ex.sahara/i</param>
<param>OUTPUT=swift://ex.sahara/o</param>""", res)
def test_run_job(self, exec_get, job, data_source,
update, remote, wf_factory, get_ds_urls,
prepare_cluster):
wf_factory.return_value = mock.MagicMock()
remote_class = mock.MagicMock()
remote_class.__exit__.return_value = 'closed'
remote.return_value = remote_class
job_class = mock.MagicMock()
job.return_value = job_class
job.name = "myJob"
source = mock.MagicMock()
source.url = "localhost"
data_source.return_value = source
get_ds_urls.return_value = ('url', 'url')
oje = FakeOozieJobEngine(u.create_cluster())
client_class = mock.MagicMock()
client_class.add_job = mock.MagicMock(return_value=1)
client_class.get_job_info = mock.MagicMock(
return_value={'status': 'PENDING'})
oje.get_client = mock.MagicMock(return_value=client_class)
_, job_exec = u.create_job_exec(edp.JOB_TYPE_PIG)
update.return_value = job_exec
self.assertEqual((1, 'PENDING', None), oje.run_job(job_exec))
def _build_workflow_with_conf_common(self, job_type):
input_data = u.create_data_source('swift://ex/i')
output_data = u.create_data_source('swift://ex/o')
job, job_exec = u.create_job_exec(job_type,
configs={"configs": {'c': 'f'}})
res = workflow_factory.get_workflow_xml(
job, u.create_cluster(), job_exec, input_data, output_data,
'hadoop')
self.assertIn("""
<property>
<name>c</name>
<value>f</value>
</property>""", res)
self.assertIn("""
<property>
<name>mapred.input.dir</name>
<value>swift://ex.sahara/i</value>
</property>""", res)
self.assertIn("""
<property>
<name>mapred.output.dir</name>
<value>swift://ex.sahara/o</value>
</property>""", res)
def test_run_job_handles_exceptions(self, runjob, job_ex_upd):
runjob.side_effect = ex.SwiftClientException("Unauthorised")
job, job_exec = u.create_job_exec(edp.JOB_TYPE_PIG)
job_manager.run_job(job_exec.id)
self.assertEqual(1, job_ex_upd.call_count)
new_status = job_ex_upd.call_args[0][2]["info"]["status"]
self.assertEqual(edp.JOB_STATUS_FAILED, new_status)
def test_hdfs_create_workflow_dir(self, remote):
remote_class = mock.MagicMock()
remote_class.__exit__.return_value = 'closed'
remote.return_value = remote_class
oje = FakeOozieJobEngine(u.create_cluster())
job, _ = u.create_job_exec(edp.JOB_TYPE_PIG)
res = oje._create_hdfs_workflow_dir(mock.Mock(), job)
self.assertIn('/user/hadoop/special_name/', res)
def test_build_workflow_for_job_java_with_adapter(self, edp_conf_mock):
edp_conf_mock.return_value = True
configs = {"configs": {"edp.java.main_class": "some_main"}}
job, job_exec = u.create_job_exec(edp.JOB_TYPE_JAVA, configs)
res = workflow_factory.get_workflow_xml(job, u.create_cluster(), job_exec.job_configs)
self.assertIn("<main-class>org.openstack.sahara.edp.MainWrapper</main-class>", res)
self.assertNotIn("some_main", res)
def test_run_job_unsupported_type(self, cluster_get, job_exec_get, job_get):
job, job_exec = u.create_job_exec("unsupported_type")
job_exec_get.return_value = job_exec
job_get.return_value = job
cluster = u.create_cluster()
cluster.status = c_u.CLUSTER_STATUS_ACTIVE
cluster_get.return_value = cluster
with testtools.ExpectedException(ex.EDPError):
job_manager._run_job(job_exec.id)
def test_run_job_unsupported_type(self,
cluster_get, job_exec_get, job_get):
job, job_exec = u.create_job_exec("unsupported_type")
job_exec_get.return_value = job_exec
job_get.return_value = job
cluster = u.create_cluster()
cluster.status = c_u.CLUSTER_STATUS_ACTIVE
cluster_get.return_value = cluster
with testtools.ExpectedException(ex.EDPError):
job_manager._run_job(job_exec.id)
def test_build_workflow_for_job_shell(self):
configs = {"configs": {"k1": "v1"}, "params": {"p1": "v1"}, "args": ["a1", "a2"]}
job, job_exec = u.create_job_exec(edp.JOB_TYPE_SHELL, configs)
res = workflow_factory.get_workflow_xml(job, u.create_cluster(), job_exec.job_configs)
self.assertIn("<name>k1</name>", res)
self.assertIn("<value>v1</value>", res)
self.assertIn("<env-var>p1=v1</env-var>", res)
self.assertIn("<argument>a1</argument>", res)
self.assertIn("<argument>a2</argument>", res)
def test_suspend_unsuspendible_job(self, suspend_job_get, cluster_get,
job_exec_get, job_get):
info = {'status': edp.JOB_STATUS_SUCCEEDED}
job, job_exec = u.create_job_exec(edp.JOB_TYPE_PIG, None, False, info)
job_exec_get.return_value = job_exec
job_get.return_value = job
cluster = u.create_cluster()
cluster.status = "Active"
cluster_get.return_value = cluster
self.assertEqual(0, suspend_job_get.call_count)
def test_build_workflow_for_job_java_with_adapter(self, edp_conf_mock):
edp_conf_mock.return_value = True
configs = {"configs": {"edp.java.main_class": "some_main"}}
job, job_exec = u.create_job_exec(edp.JOB_TYPE_JAVA, configs)
res = workflow_factory.get_workflow_xml(
job, u.create_cluster(), job_exec.job_configs)
self.assertIn(
"<main-class>org.openstack.sahara.edp.MainWrapper</main-class>",
res)
self.assertNotIn("some_main", res)
def test_get_job_status(self):
oje = FakeOozieJobEngine(u.create_cluster())
client_class = mock.MagicMock()
client_class.add_job = mock.MagicMock(return_value=1)
client_class.get_job_info = mock.MagicMock(
return_value={'status': 'PENDING'})
oje.get_client = mock.MagicMock(return_value=client_class)
_, job_exec = u.create_job_exec(edp.JOB_TYPE_PIG)
self.assertIsNone(oje.get_job_status(job_exec))
job_exec.engine_job_id = 1
self.assertEqual({'status': 'PENDING'}, oje.get_job_status(job_exec))
def test_scheduled_edp_job_run(self, job_exec_get, cluster_get, job_get, run_scheduled_job):
configs = {"job_execution_info": {"job_execution_type": "scheduled", "start": "2015-5-15T01:00Z"}}
job, job_exec = u.create_job_exec(edp.JOB_TYPE_PIG, configs)
job_exec_get.return_value = job_exec
job_get.return_value = job
cluster = u.create_cluster()
cluster.status = "Active"
cluster_get.return_value = cluster
job_manager._run_job(job_exec.id)
self.assertEqual(1, run_scheduled_job.call_count)
def test_get_job_status(self):
oje = FakeOozieJobEngine(u.create_cluster())
client_class = mock.MagicMock()
client_class.add_job = mock.MagicMock(return_value=1)
client_class.get_job_info = mock.MagicMock(
return_value={'status': 'PENDING'})
oje.get_client = mock.MagicMock(return_value=client_class)
_, job_exec = u.create_job_exec(edp.JOB_TYPE_PIG)
self.assertIsNone(oje.get_job_status(job_exec))
job_exec.engine_job_id = 1
self.assertEqual({'status': 'PENDING'}, oje.get_job_status(job_exec))
def test_run_job_handles_exceptions_with_run_job(self, canceljob, runjob, job_ex_get, job_ex_upd):
runjob.side_effect = ex.OozieException("run_job failed")
job, job_exec = u.create_job_exec(edp.JOB_TYPE_PIG)
job_exec.engine_job_id = "fake_oozie_id"
job_ex_get.return_value = job_exec
job_manager.run_job(job_exec.id)
self.assertEqual(1, job_ex_get.call_count)
self.assertEqual(1, job_ex_upd.call_count)
new_status = job_ex_upd.call_args[0][2]["info"]["status"]
self.assertEqual(edp.JOB_STATUS_FAILED, new_status)
self.assertEqual(1, canceljob.call_count)
def test_cancel_job(self, kill_get, info_get):
info_get.return_value = {}
oje = FakeOozieJobEngine(u.create_cluster())
_, job_exec = u.create_job_exec(edp.JOB_TYPE_PIG)
# test cancel job without engine_job_id
job_exec.engine_job_id = None
oje.cancel_job(job_exec)
self.assertEqual(0, kill_get.call_count)
# test cancel job with engine_job_id
job_exec.engine_job_id = 123
oje.cancel_job(job_exec)
self.assertEqual(1, kill_get.call_count)
def test_suspend_unsuspendible_job(self, suspend_job_get,
cluster_get, job_exec_get, job_get):
info = {
'status': edp.JOB_STATUS_SUCCEEDED
}
job, job_exec = u.create_job_exec(edp.JOB_TYPE_PIG, None, False, info)
job_exec_get.return_value = job_exec
job_get.return_value = job
cluster = u.create_cluster()
cluster.status = "Active"
cluster_get.return_value = cluster
self.assertEqual(0, suspend_job_get.call_count)
def test_get_data_sources(self, ds):
def _conductor_data_source_get(ctx, id):
return mock.Mock(id=id, url="obj_" + id)
job, job_exec = u.create_job_exec(edp.JOB_TYPE_PIG)
job_exec.input_id = "s1"
job_exec.output_id = "s2"
ds.side_effect = _conductor_data_source_get
input_source, output_source = job_utils.get_data_sources(job_exec, job, {})
self.assertEqual("obj_s1", input_source.url)
self.assertEqual("obj_s2", output_source.url)
def test_get_data_sources_with_null_id(self):
configs = {sw.HADOOP_SWIFT_USERNAME: "******", sw.HADOOP_SWIFT_PASSWORD: "******"}
configs = {"configs": configs, "args": ["swift://ex/i", "output_path"]}
job, job_exec = u.create_job_exec(edp.JOB_TYPE_JAVA, configs)
job_exec.input_id = None
job_exec.output_id = None
input_source, output_source = job_utils.get_data_sources(job_exec, job, {})
self.assertIsNone(input_source)
self.assertIsNone(output_source)
def test_cancel_job(self, kill_get, info_get):
info_get.return_value = {}
oje = FakeOozieJobEngine(u.create_cluster())
_, job_exec = u.create_job_exec(edp.JOB_TYPE_PIG)
# test cancel job without engine_job_id
job_exec.engine_job_id = None
oje.cancel_job(job_exec)
self.assertEqual(0, kill_get.call_count)
# test cancel job with engine_job_id
job_exec.engine_job_id = 123
oje.cancel_job(job_exec)
self.assertEqual(1, kill_get.call_count)
def test_get_data_sources_java(self):
configs = {
sw.HADOOP_SWIFT_USERNAME: '******',
sw.HADOOP_SWIFT_PASSWORD: '******'
}
configs = {'configs': configs, 'args': ['swift://ex/i', 'output_path']}
job, job_exec = u.create_job_exec(edp.JOB_TYPE_JAVA, configs)
input_source, output_source = (job_utils.get_data_sources(
job_exec, job))
self.assertEqual(None, input_source)
self.assertEqual(None, output_source)
def test_get_input_output_data_sources(self, ds):
def _conductor_data_source_get(ctx, id):
return mock.Mock(id=id, url="hdfs://obj_" + id, type='hdfs')
job, job_exec = u.create_job_exec(edp.JOB_TYPE_PIG)
job_exec.input_id = 's1'
job_exec.output_id = 's2'
ds.side_effect = _conductor_data_source_get
input_source, output_source = (
job_utils.get_input_output_data_sources(job_exec, job, {}))
self.assertEqual('hdfs://obj_s1', input_source.url)
self.assertEqual('hdfs://obj_s2', output_source.url)
def test_get_data_sources(self, ds):
def _conductor_data_source_get(ctx, id):
return mock.Mock(id=id, url="obj_" + id)
job, job_exec = u.create_job_exec(edp.JOB_TYPE_PIG)
job_exec.input_id = 's1'
job_exec.output_id = 's2'
ds.side_effect = _conductor_data_source_get
input_source, output_source = (
job_utils.get_data_sources(job_exec, job, {}))
self.assertEqual('obj_s1', input_source.url)
self.assertEqual('obj_s2', output_source.url)
def test_run_job_handles_exceptions_with_run_job(self, canceljob, runjob,
job_ex_get, job_ex_upd):
runjob.side_effect = ex.OozieException("run_job failed")
job, job_exec = u.create_job_exec(edp.JOB_TYPE_PIG)
job_exec.engine_job_id = "fake_oozie_id"
job_ex_get.return_value = job_exec
job_manager.run_job(job_exec.id)
self.assertEqual(1, job_ex_get.call_count)
self.assertEqual(1, job_ex_upd.call_count)
new_status = job_ex_upd.call_args[0][2]["info"]["status"]
self.assertEqual(edp.JOB_STATUS_FAILED, new_status)
self.assertEqual(1, canceljob.call_count)
def test_build_workflow_for_job_shell(self):
configs = {"configs": {"k1": "v1"},
"params": {"p1": "v1"},
"args": ["a1", "a2"]}
job, job_exec = u.create_job_exec(edp.JOB_TYPE_SHELL, configs)
res = workflow_factory.get_workflow_xml(
job, u.create_cluster(), job_exec.job_configs)
self.assertIn("<name>k1</name>", res)
self.assertIn("<value>v1</value>", res)
self.assertIn("<env-var>p1=v1</env-var>", res)
self.assertIn("<argument>a1</argument>", res)
self.assertIn("<argument>a2</argument>", res)
def test_get_data_sources(self, ds):
def _conductor_data_source_get(ctx, id):
return "obj_" + id
job, job_exec = u.create_job_exec(edp.JOB_TYPE_PIG)
job_exec.input_id = 's1'
job_exec.output_id = 's2'
ds.side_effect = _conductor_data_source_get
input_source, output_source = (
job_utils.get_data_sources(job_exec, job))
self.assertEqual('obj_s1', input_source)
self.assertEqual('obj_s2', output_source)
def test_run_job_handles_exceptions(self, canceljob, runjob,
job_ex_get, job_ex_upd):
runjob.side_effect = ex.SwiftClientException("Unauthorised")
job, job_exec = u.create_job_exec(edp.JOB_TYPE_PIG)
job_exec.engine_job_id = None
job_ex_get.return_value = job_exec
job_manager.run_job(job_exec.id)
self.assertEqual(1, job_ex_get.call_count)
self.assertEqual(1, job_ex_upd.call_count)
new_status = job_ex_upd.call_args[0][2]["info"]["status"]
self.assertEqual(edp.JOB_STATUS_FAILED, new_status)
self.assertEqual(0, canceljob.call_count)
def _build_workflow_common(self, job_type, streaming=False):
if streaming:
configs = {'edp.streaming.mapper': '/usr/bin/cat',
'edp.streaming.reducer': '/usr/bin/wc'}
configs = {'configs': configs}
else:
configs = {}
job, job_exec = u.create_job_exec(job_type, configs)
input_data = u.create_data_source('swift://ex/i')
output_data = u.create_data_source('swift://ex/o')
res = workflow_factory.get_workflow_xml(
job, u.create_cluster(), job_exec, input_data, output_data,
'hadoop')
if streaming:
self.assertIn("""
<streaming>
<mapper>/usr/bin/cat</mapper>
<reducer>/usr/bin/wc</reducer>
</streaming>""", res)
self.assertIn("""
<property>
<name>mapred.output.dir</name>
<value>swift://ex.sahara/o</value>
</property>""", res)
self.assertIn("""
<property>
<name>mapred.input.dir</name>
<value>swift://ex.sahara/i</value>
</property>""", res)
self.assertIn("""
<property>
<name>fs.swift.service.sahara.password</name>
<value>admin1</value>
</property>""", res)
self.assertIn("""
<property>
<name>fs.swift.service.sahara.username</name>
<value>admin</value>
</property>""", res)
def test_failed_to_cancel_job(self, time_get, cluster_get, job_exec_get, job_get, job_execution_update_get):
info = {"status": edp.JOB_STATUS_RUNNING}
job, job_exec = u.create_job_exec(edp.JOB_TYPE_PIG, None, False, info)
job_exec_get.return_value = job_exec
job_get.return_value = job
cluster = u.create_cluster()
cluster.status = c_u.CLUSTER_STATUS_ACTIVE
cluster_get.return_value = cluster
time_get.return_value = 10000
job_execution_update_get.return_value = job_exec
with testtools.ExpectedException(ex.CancelingFailed):
job_manager.cancel_job(job_exec.id)
def test_get_data_sources_java(self):
configs = {sw.HADOOP_SWIFT_USERNAME: '******',
sw.HADOOP_SWIFT_PASSWORD: '******'}
configs = {
'configs': configs,
'args': ['swift://ex/i',
'output_path']
}
job, job_exec = u.create_job_exec(edp.JOB_TYPE_JAVA, configs)
input_source, output_source = (
job_utils.get_data_sources(job_exec, job))
self.assertEqual(None, input_source)
self.assertEqual(None, output_source)
def test_get_input_output_data_sources_with_null_id(self):
configs = {
sw.HADOOP_SWIFT_USERNAME: '******',
sw.HADOOP_SWIFT_PASSWORD: '******'
}
configs = {'configs': configs, 'args': ['hdfs://ex/i', 'output_path']}
job, job_exec = u.create_job_exec(edp.JOB_TYPE_JAVA, configs)
job_exec.input_id = None
job_exec.output_id = None
input_source, output_source = (job_utils.get_input_output_data_sources(
job_exec, job, {}))
self.assertIsNone(input_source)
self.assertIsNone(output_source)
def test_failed_to_cancel_job(self, time_get, cluster_get, job_exec_get,
job_get, job_execution_update_get):
info = {'status': edp.JOB_STATUS_RUNNING}
job, job_exec = u.create_job_exec(edp.JOB_TYPE_PIG, None, False, info)
job_exec_get.return_value = job_exec
job_get.return_value = job
cluster = u.create_cluster()
cluster.status = c_u.CLUSTER_STATUS_ACTIVE
cluster_get.return_value = cluster
time_get.return_value = 10000
job_execution_update_get.return_value = job_exec
with testtools.ExpectedException(ex.CancelingFailed):
job_manager.cancel_job(job_exec.id)
def test_scheduled_edp_job_run(self, job_exec_get, cluster_get,
job_get, run_scheduled_job):
configs = {
'job_execution_info': {
'job_execution_type': 'scheduled',
'start': '2015-5-15T01:00Z'
}
}
job, job_exec = u.create_job_exec(edp.JOB_TYPE_PIG, configs)
job_exec_get.return_value = job_exec
job_get.return_value = job
cluster = u.create_cluster()
cluster.status = "Active"
cluster_get.return_value = cluster
job_manager._run_job(job_exec.id)
self.assertEqual(1, run_scheduled_job.call_count)
def test_get_data_sources_with_null_id(self):
configs = {sw.HADOOP_SWIFT_USERNAME: '******',
sw.HADOOP_SWIFT_PASSWORD: '******'}
configs = {
'configs': configs,
'args': ['swift://ex/i',
'output_path']
}
job, job_exec = u.create_job_exec(edp.JOB_TYPE_JAVA, configs)
job_exec.input_id = None
job_exec.output_id = None
input_source, output_source = (
job_utils.get_data_sources(job_exec, job, {}))
self.assertIsNone(input_source)
self.assertIsNone(output_source)
def test_build_workflow_for_job_java(self):
# If args include swift paths, user and password values
# will have to be supplied via configs instead of being
# lifted from input or output data sources
configs = {sw.HADOOP_SWIFT_USERNAME: '******',
sw.HADOOP_SWIFT_PASSWORD: '******'}
configs = {
'configs': configs,
'args': ['swift://ex/i',
'output_path']
}
job, job_exec = u.create_job_exec(edp.JOB_TYPE_JAVA, configs)
res = workflow_factory.get_workflow_xml(
job, u.create_cluster(), job_exec)
self.assertIn("""
<configuration>
<property>
<name>fs.swift.service.sahara.password</name>
<value>admin1</value>
</property>
<property>
<name>fs.swift.service.sahara.username</name>
<value>admin</value>
</property>
</configuration>
<main-class>%s</main-class>
<java-opts>%s</java-opts>
<arg>swift://ex.sahara/i</arg>
<arg>output_path</arg>""" % (_java_main_class, _java_opts), res)
# testing workflow creation with a proxy domain
self.override_config('use_domain_for_proxy_users', True)
self.override_config("proxy_user_domain_name", 'sahara_proxy_domain')
configs = {
'configs': {},
'args': ['swift://ex/i',
'output_path']
}
job, job_exec = u.create_job_exec(edp.JOB_TYPE_JAVA, configs,
proxy=True)
res = workflow_factory.get_workflow_xml(job, u.create_cluster(),
job_exec)
self.assertIn("""
<configuration>
<property>
<name>fs.swift.service.sahara.domain.name</name>
<value>sahara_proxy_domain</value>
</property>
<property>
<name>fs.swift.service.sahara.password</name>
<value>55555555-6666-7777-8888-999999999999</value>
</property>
<property>
<name>fs.swift.service.sahara.trust.id</name>
<value>0123456789abcdef0123456789abcdef</value>
</property>
<property>
<name>fs.swift.service.sahara.username</name>
<value>job_00000000-1111-2222-3333-4444444444444444</value>
</property>
</configuration>
<main-class>%s</main-class>
<java-opts>%s</java-opts>
<arg>swift://ex.sahara/i</arg>
<arg>output_path</arg>""" % (_java_main_class, _java_opts), res)
def test_build_workflow_for_job_hive(self, job_binary):
job, job_exec = u.create_job_exec(edp.JOB_TYPE_HIVE, configs={})
job_binary.return_value = {"name": "script.q"}
input_data = u.create_data_source('swift://ex/i')
output_data = u.create_data_source('swift://ex/o')
data_source_urls = {
input_data.id: input_data.url,
output_data.id: output_data.url
}
res = workflow_factory.get_workflow_xml(job, u.create_cluster(),
job_exec.job_configs,
input_data, output_data,
'hadoop', data_source_urls)
doc = xml.parseString(res)
hive = doc.getElementsByTagName('hive')[0]
self.assertEqual('/user/hadoop/conf/hive-site.xml',
xmlutils.get_text_from_node(hive, 'job-xml'))
configuration = hive.getElementsByTagName('configuration')
properties = xmlutils.get_property_dict(configuration[0])
self.assertEqual(
{
'fs.swift.service.sahara.password': '******',
'fs.swift.service.sahara.username': '******'
}, properties)
self.assertEqual('script.q',
xmlutils.get_text_from_node(hive, 'script'))
params = xmlutils.get_param_dict(hive)
self.assertEqual(
{
'INPUT': 'swift://ex.sahara/i',
'OUTPUT': 'swift://ex.sahara/o'
}, params)
# testing workflow creation with a proxy domain
self.override_config('use_domain_for_proxy_users', True)
self.override_config("proxy_user_domain_name", 'sahara_proxy_domain')
job, job_exec = u.create_job_exec(edp.JOB_TYPE_HIVE, proxy=True)
res = workflow_factory.get_workflow_xml(job, u.create_cluster(),
job_exec.job_configs,
input_data, output_data,
'hadoop', data_source_urls)
doc = xml.parseString(res)
hive = doc.getElementsByTagName('hive')[0]
configuration = hive.getElementsByTagName('configuration')
properties = xmlutils.get_property_dict(configuration[0])
self.assertEqual(
{
'fs.swift.service.sahara.domain.name':
'sahara_proxy_domain',
'fs.swift.service.sahara.trust.id':
'0123456789abcdef0123456789abcdef',
'fs.swift.service.sahara.password':
'******',
'fs.swift.service.sahara.username':
'******'
}, properties)
def test_build_workflow_for_job_java(self):
# If args include swift paths, user and password values
# will have to be supplied via configs instead of being
# lifted from input or output data sources
configs = {sw.HADOOP_SWIFT_USERNAME: '******',
sw.HADOOP_SWIFT_PASSWORD: '******'}
configs = {
'configs': configs,
'args': ['swift://ex/i',
'output_path']
}
job, job_exec = u.create_job_exec(edp.JOB_TYPE_JAVA, configs)
res = workflow_factory.get_workflow_xml(
job, u.create_cluster(), job_exec.job_configs)
self.assertIn("""
<configuration>
<property>
<name>fs.swift.service.sahara.password</name>
<value>admin1</value>
</property>
<property>
<name>fs.swift.service.sahara.username</name>
<value>admin</value>
</property>
</configuration>
<main-class>%s</main-class>
<java-opts>%s</java-opts>
<arg>swift://ex.sahara/i</arg>
<arg>output_path</arg>""" % (_java_main_class, _java_opts), res)
# testing workflow creation with a proxy domain
self.override_config('use_domain_for_proxy_users', True)
self.override_config("proxy_user_domain_name", 'sahara_proxy_domain')
configs = {
'configs': {},
'args': ['swift://ex/i',
'output_path']
}
job, job_exec = u.create_job_exec(edp.JOB_TYPE_JAVA, configs,
proxy=True)
res = workflow_factory.get_workflow_xml(job, u.create_cluster(),
job_exec.job_configs)
self.assertIn("""
<configuration>
<property>
<name>fs.swift.service.sahara.domain.name</name>
<value>sahara_proxy_domain</value>
</property>
<property>
<name>fs.swift.service.sahara.password</name>
<value>55555555-6666-7777-8888-999999999999</value>
</property>
<property>
<name>fs.swift.service.sahara.trust.id</name>
<value>0123456789abcdef0123456789abcdef</value>
</property>
<property>
<name>fs.swift.service.sahara.username</name>
<value>job_00000000-1111-2222-3333-4444444444444444</value>
</property>
</configuration>
<main-class>%s</main-class>
<java-opts>%s</java-opts>
<arg>swift://ex.sahara/i</arg>
<arg>output_path</arg>""" % (_java_main_class, _java_opts), res)
def test_build_workflow_swift_configs(self, job_binary):
# Test that swift configs come from either input or output data sources
job, job_exec = u.create_job_exec(edp.JOB_TYPE_PIG, configs={})
job_binary.return_value = {"name": "script.pig"}
input_data = u.create_data_source('swift://ex/i')
output_data = u.create_data_source('hdfs://user/hadoop/out')
data_source_urls = {input_data.id: input_data.url,
output_data.id: output_data.url}
res = workflow_factory.get_workflow_xml(
job, u.create_cluster(), job_exec.job_configs,
input_data, output_data, 'hadoop', data_source_urls)
self.assertIn("""
<configuration>
<property>
<name>fs.swift.service.sahara.password</name>
<value>admin1</value>
</property>
<property>
<name>fs.swift.service.sahara.username</name>
<value>admin</value>
</property>
</configuration>""", res)
input_data = u.create_data_source('hdfs://user/hadoop/in')
output_data = u.create_data_source('swift://ex/o')
data_source_urls = {input_data.id: input_data.url,
output_data.id: output_data.url}
res = workflow_factory.get_workflow_xml(
job, u.create_cluster(), job_exec.job_configs,
input_data, output_data, 'hadoop', data_source_urls)
self.assertIn("""
<configuration>
<property>
<name>fs.swift.service.sahara.password</name>
<value>admin1</value>
</property>
<property>
<name>fs.swift.service.sahara.username</name>
<value>admin</value>
</property>
</configuration>""", res)
job, job_exec = u.create_job_exec(
edp.JOB_TYPE_PIG, configs={'configs': {'dummy': 'value'}})
input_data = u.create_data_source('hdfs://user/hadoop/in')
output_data = u.create_data_source('hdfs://user/hadoop/out')
data_source_urls = {input_data.id: input_data.url,
output_data.id: output_data.url}
res = workflow_factory.get_workflow_xml(
job, u.create_cluster(), job_exec.job_configs,
input_data, output_data, 'hadoop', data_source_urls)
self.assertIn("""
<configuration>
<property>
<name>dummy</name>
<value>value</value>
</property>
</configuration>""", res)
def test_build_workflow_for_job_pig(self, job_binary):
job, job_exec = u.create_job_exec(edp.JOB_TYPE_PIG, configs={})
job_binary.return_value = {"name": "script.pig"}
input_data = u.create_data_source('swift://ex/i')
output_data = u.create_data_source('swift://ex/o')
data_source_urls = {
input_data.id: input_data.url,
output_data.id: output_data.url
}
res = workflow_factory.get_workflow_xml(job, u.create_cluster(),
job_exec.job_configs,
input_data, output_data,
'hadoop', data_source_urls)
self.assertIn(
"""
<param>INPUT=swift://ex.sahara/i</param>
<param>OUTPUT=swift://ex.sahara/o</param>""", res)
self.assertIn(
"""
<configuration>
<property>
<name>fs.swift.service.sahara.password</name>
<value>admin1</value>
</property>
<property>
<name>fs.swift.service.sahara.username</name>
<value>admin</value>
</property>
</configuration>""", res)
self.assertIn("<script>script.pig</script>", res)
# testing workflow creation with a proxy domain
self.override_config('use_domain_for_proxy_users', True)
self.override_config("proxy_user_domain_name", 'sahara_proxy_domain')
job, job_exec = u.create_job_exec(edp.JOB_TYPE_PIG, proxy=True)
res = workflow_factory.get_workflow_xml(job, u.create_cluster(),
job_exec.job_configs,
input_data, output_data,
'hadoop', data_source_urls)
self.assertIn(
"""
<configuration>
<property>
<name>fs.swift.service.sahara.domain.name</name>
<value>sahara_proxy_domain</value>
</property>
<property>
<name>fs.swift.service.sahara.password</name>
<value>55555555-6666-7777-8888-999999999999</value>
</property>
<property>
<name>fs.swift.service.sahara.trust.id</name>
<value>0123456789abcdef0123456789abcdef</value>
</property>
<property>
<name>fs.swift.service.sahara.username</name>
<value>job_00000000-1111-2222-3333-4444444444444444</value>
</property>
</configuration>""", res)
def _build_workflow_common(self, job_type, streaming=False, proxy=False):
if streaming:
configs = {'edp.streaming.mapper': '/usr/bin/cat',
'edp.streaming.reducer': '/usr/bin/wc'}
configs = {'configs': configs}
else:
configs = {}
job, job_exec = u.create_job_exec(job_type, configs)
input_data = u.create_data_source('swift://ex/i')
output_data = u.create_data_source('swift://ex/o')
data_source_urls = {input_data.id: input_data.url,
output_data.id: output_data.url}
res = workflow_factory.get_workflow_xml(
job, u.create_cluster(), job_exec.job_configs,
input_data, output_data, 'hadoop', data_source_urls)
if streaming:
self.assertIn("""
<streaming>
<mapper>/usr/bin/cat</mapper>
<reducer>/usr/bin/wc</reducer>
</streaming>""", res)
self.assertIn("""
<property>
<name>mapred.output.dir</name>
<value>swift://ex.sahara/o</value>
</property>""", res)
self.assertIn("""
<property>
<name>mapred.input.dir</name>
<value>swift://ex.sahara/i</value>
</property>""", res)
if not proxy:
self.assertIn("""
<property>
<name>fs.swift.service.sahara.password</name>
<value>admin1</value>
</property>""", res)
self.assertIn("""
<property>
<name>fs.swift.service.sahara.username</name>
<value>admin</value>
</property>""", res)
else:
# testing workflow creation with a proxy domain
self.override_config('use_domain_for_proxy_users', True)
self.override_config("proxy_user_domain_name",
'sahara_proxy_domain')
job, job_exec = u.create_job_exec(job_type, proxy=True)
res = workflow_factory.get_workflow_xml(
job, u.create_cluster(), job_exec.job_configs,
input_data, output_data, 'hadoop', data_source_urls)
self.assertIn("""
<property>
<name>fs.swift.service.sahara.domain.name</name>
<value>sahara_proxy_domain</value>
</property>
<property>
<name>fs.swift.service.sahara.password</name>
<value>55555555-6666-7777-8888-999999999999</value>
</property>
<property>
<name>fs.swift.service.sahara.trust.id</name>
<value>0123456789abcdef0123456789abcdef</value>
</property>
<property>
<name>fs.swift.service.sahara.username</name>
<value>job_00000000-1111-2222-3333-4444444444444444</value>
</property>""", res)