def test_resolve_connection_spark_home_not_set_connection(self):
# Given
hook = SparkSubmitHook(conn_id='spark_home_not_set')
# When
connection = hook._resolve_connection()
cmd = hook._build_spark_submit_command(self._spark_job_file)
# Then
expected_spark_connection = {
"master": "yarn://yarn-master",
"spark_binary": "spark-submit",
"deploy_mode": None,
"queue": None,
"spark_home": None,
"namespace": None,
}
assert connection == expected_spark_connection
assert cmd[0] == 'spark-submit'
def test_resolve_connection_spark_k8s_cluster_connection(self):
# Given
hook = SparkSubmitHook(conn_id='spark_k8s_cluster')
# When
connection = hook._resolve_connection()
cmd = hook._build_spark_submit_command(self._spark_job_file)
# Then
dict_cmd = self.cmd_args_to_dict(cmd)
expected_spark_connection = {"spark_home": "/opt/spark",
"queue": None,
"spark_binary": "spark-submit",
"master": "k8s://https://k8s-master",
"deploy_mode": "cluster",
"namespace": "mynamespace"}
self.assertEqual(connection, expected_spark_connection)
self.assertEqual(dict_cmd["--master"], "k8s://https://k8s-master")
self.assertEqual(dict_cmd["--deploy-mode"], "cluster")
def test_resolve_connection_yarn_default_connection(self):
# Given
hook = SparkSubmitHook(conn_id='spark_default')
# When
connection = hook._resolve_connection()
cmd = hook._build_spark_submit_command(self._spark_job_file)
# Then
dict_cmd = self.cmd_args_to_dict(cmd)
expected_spark_connection = {"master": "yarn",
"spark_binary": "spark-submit",
"deploy_mode": None,
"queue": "root.default",
"spark_home": None,
"namespace": None}
self.assertEqual(connection, expected_spark_connection)
self.assertEqual(dict_cmd["--master"], "yarn")
self.assertEqual(dict_cmd["--queue"], "root.default")
def test_resolve_connection_spark_standalone_cluster_connection(self):
# Given
hook = SparkSubmitHook(conn_id='spark_standalone_cluster')
# When
connection = hook._resolve_connection()
cmd = hook._build_spark_submit_command(self._spark_job_file)
# Then
expected_spark_connection = {
"master": "spark://spark-standalone-master:6066",
"spark_binary": "spark-submit",
"deploy_mode": "cluster",
"queue": None,
"spark_home": "/path/to/spark_home",
"namespace": None
}
self.assertEqual(connection, expected_spark_connection)
self.assertEqual(cmd[0], '/path/to/spark_home/bin/spark-submit')
def test_resolve_connection_spark_binary_default_value(self):
# Given
hook = SparkSubmitHook(conn_id='spark_default')
# When
connection = hook._resolve_connection()
cmd = hook._build_spark_submit_command(self._spark_job_file)
# Then
expected_spark_connection = {
"master": "yarn",
"spark_binary": "spark-submit",
"deploy_mode": None,
"queue": 'root.default',
"spark_home": None,
"namespace": None
}
self.assertEqual(connection, expected_spark_connection)
self.assertEqual(cmd[0], 'spark-submit')
def test_resolve_connection_spark_home_set_connection(self):
# Given
hook = SparkSubmitHook(conn_id='spark_home_set')
# When
connection = hook._resolve_connection()
cmd = hook._build_spark_submit_command(self._spark_job_file)
# Then
expected_spark_connection = {
"master": "yarn://yarn-master",
"spark_binary": "spark-submit",
"deploy_mode": None,
"queue": None,
"spark_home": "/opt/myspark",
"namespace": None
}
self.assertEqual(connection, expected_spark_connection)
self.assertEqual(cmd[0], '/opt/myspark/bin/spark-submit')
def test_resolve_connection_spark_binary_default_value_override(self):
# Given
hook = SparkSubmitHook(conn_id='spark_binary_set',
spark_binary='another-custom-spark-submit')
# When
connection = hook._resolve_connection()
cmd = hook._build_spark_submit_command(self._spark_job_file)
# Then
expected_spark_connection = {
"master": "yarn",
"spark_binary": "another-custom-spark-submit",
"deploy_mode": None,
"queue": None,
"spark_home": None,
"namespace": None,
}
assert connection == expected_spark_connection
assert cmd[0] == 'another-custom-spark-submit'
def test_resolve_connection_mesos_default_connection(self):
# Given
hook = SparkSubmitHook(conn_id='spark_default_mesos')
# When
connection = hook._resolve_connection()
cmd = hook._build_spark_submit_command(self._spark_job_file)
# Then
dict_cmd = self.cmd_args_to_dict(cmd)
expected_spark_connection = {
"master": "mesos://host:5050",
"spark_binary": "spark-submit",
"deploy_mode": None,
"queue": None,
"spark_home": None,
"namespace": None,
}
assert connection == expected_spark_connection
assert dict_cmd["--master"] == "mesos://host:5050"
def test_resolve_connection_spark_yarn_cluster_connection(self):
# Given
hook = SparkSubmitHook(conn_id='spark_yarn_cluster')
# When
connection = hook._resolve_connection()
cmd = hook._build_spark_submit_command(self._spark_job_file)
# Then
dict_cmd = self.cmd_args_to_dict(cmd)
expected_spark_connection = {
"master": "yarn://yarn-master",
"spark_binary": "spark-submit",
"deploy_mode": "cluster",
"queue": "root.etl",
"spark_home": None,
"namespace": None,
}
assert connection == expected_spark_connection
assert dict_cmd["--master"] == "yarn://yarn-master"
assert dict_cmd["--queue"] == "root.etl"
assert dict_cmd["--deploy-mode"] == "cluster"
def test_resolve_connection_spark_k8s_cluster_ns_conf(self):
# Given we specify the config option directly
conf = {
'spark.kubernetes.namespace': 'airflow',
}
hook = SparkSubmitHook(conn_id='spark_k8s_cluster', conf=conf)
# When
connection = hook._resolve_connection()
cmd = hook._build_spark_submit_command(self._spark_job_file)
# Then
dict_cmd = self.cmd_args_to_dict(cmd)
expected_spark_connection = {"spark_home": "/opt/spark",
"queue": None,
"spark_binary": "spark-submit",
"master": "k8s://https://k8s-master",
"deploy_mode": "cluster",
"namespace": "airflow"}
self.assertEqual(connection, expected_spark_connection)
self.assertEqual(dict_cmd["--master"], "k8s://https://k8s-master")
self.assertEqual(dict_cmd["--deploy-mode"], "cluster")
self.assertEqual(dict_cmd["--conf"], "spark.kubernetes.namespace=airflow")
def test_build_spark_submit_command(self):
# Given
hook = SparkSubmitHook(**self._config)
# When
cmd = hook._build_spark_submit_command(self._spark_job_file)
# Then
expected_build_cmd = [
'spark-submit',
'--master', 'yarn',
'--conf', 'parquet.compression=SNAPPY',
'--files', 'hive-site.xml',
'--py-files', 'sample_library.py',
'--archives', 'sample_archive.zip#SAMPLE',
'--jars', 'parquet.jar',
'--packages', 'com.databricks:spark-avro_2.11:3.2.0',
'--exclude-packages', 'org.bad.dependency:1.0.0',
'--repositories', 'http://myrepo.org',
'--num-executors', '10',
'--total-executor-cores', '4',
'--executor-cores', '4',
'--executor-memory', '22g',
'--driver-memory', '3g',
'--keytab', 'privileged_user.keytab',
'--principal', 'user/[email protected]',
'--proxy-user', 'sample_user',
'--name', 'spark-job',
'--class', 'com.foo.bar.AppMain',
'--verbose',
'test_application.py',
'-f', 'foo',
'--bar', 'bar',
'--with-spaces', 'args should keep embdedded spaces',
'baz'
]
self.assertEqual(expected_build_cmd, cmd)
