def test_copy_files():
cluster = pseudo_hdfs4.shared_cluster()
try:
c = make_logged_in_client()
user = User.objects.get(username='******')
prefix = '/tmp/test_copy_files'
if cluster.fs.exists(prefix):
cluster.fs.rmtree(prefix)
# Jars in various locations
deployment_dir = '%s/workspace' % prefix
external_deployment_dir = '%s/deployment' % prefix
jar_1 = '%s/udf1.jar' % prefix
jar_2 = '%s/lib/udf2.jar' % prefix
jar_3 = '%s/udf3.jar' % deployment_dir
jar_4 = '%s/lib/udf4.jar' % deployment_dir # Doesn't move
jar_5 = 'udf5.jar'
jar_6 = 'lib/udf6.jar' # Doesn't move
cluster.fs.mkdir(prefix)
cluster.fs.create(jar_1)
cluster.fs.create(jar_2)
cluster.fs.create(jar_3)
cluster.fs.create(jar_4)
cluster.fs.create(deployment_dir + '/' + jar_5)
cluster.fs.create(deployment_dir + '/' + jar_6)
class MockJob():
XML_FILE_NAME = 'workflow.xml'
def __init__(self):
self.deployment_dir = deployment_dir
self.nodes = [
Node({'id': '1', 'type': 'mapreduce', 'properties': {'jar_path': jar_1}}),
Node({'id': '2', 'type': 'mapreduce', 'properties': {'jar_path': jar_2}}),
Node({'id': '3', 'type': 'java', 'properties': {'jar_path': jar_3}}),
Node({'id': '4', 'type': 'java', 'properties': {'jar_path': jar_4}}),
# Workspace relative paths
Node({'id': '5', 'type': 'java', 'properties': {'jar_path': jar_5}}),
Node({'id': '6', 'type': 'java', 'properties': {'jar_path': jar_6}})
]
submission = Submission(user, job=MockJob(), fs=cluster.fs, jt=cluster.jt)
submission._copy_files(deployment_dir, "<xml>My XML</xml>", {'prop1': 'val1'})
submission._copy_files(external_deployment_dir, "<xml>My XML</xml>", {'prop1': 'val1'})
assert_true(cluster.fs.exists(deployment_dir + '/workflow.xml'), deployment_dir)
assert_true(cluster.fs.exists(deployment_dir + '/job.properties'), deployment_dir)
# All sources still there
assert_true(cluster.fs.exists(jar_1))
assert_true(cluster.fs.exists(jar_2))
assert_true(cluster.fs.exists(jar_3))
assert_true(cluster.fs.exists(jar_4))
assert_true(cluster.fs.exists(deployment_dir + '/' + jar_5))
assert_true(cluster.fs.exists(deployment_dir + '/' + jar_6))
# Lib
deployment_dir = deployment_dir + '/lib'
external_deployment_dir = external_deployment_dir + '/lib'
list_dir_workspace = cluster.fs.listdir(deployment_dir)
list_dir_deployement = cluster.fs.listdir(external_deployment_dir)
# All destinations there
assert_true(cluster.fs.exists(deployment_dir + '/udf1.jar'), list_dir_workspace)
assert_true(cluster.fs.exists(deployment_dir + '/udf2.jar'), list_dir_workspace)
assert_true(cluster.fs.exists(deployment_dir + '/udf3.jar'), list_dir_workspace)
assert_true(cluster.fs.exists(deployment_dir + '/udf4.jar'), list_dir_workspace)
assert_true(cluster.fs.exists(deployment_dir + '/udf5.jar'), list_dir_workspace)
assert_true(cluster.fs.exists(deployment_dir + '/udf6.jar'), list_dir_workspace)
assert_true(cluster.fs.exists(external_deployment_dir + '/udf1.jar'), list_dir_deployement)
assert_true(cluster.fs.exists(external_deployment_dir + '/udf2.jar'), list_dir_deployement)
assert_true(cluster.fs.exists(external_deployment_dir + '/udf3.jar'), list_dir_deployement)
assert_true(cluster.fs.exists(external_deployment_dir + '/udf4.jar'), list_dir_deployement)
assert_true(cluster.fs.exists(external_deployment_dir + '/udf5.jar'), list_dir_deployement)
assert_true(cluster.fs.exists(external_deployment_dir + '/udf6.jar'), list_dir_deployement)
stats_udf1 = cluster.fs.stats(deployment_dir + '/udf1.jar')
stats_udf2 = cluster.fs.stats(deployment_dir + '/udf2.jar')
stats_udf3 = cluster.fs.stats(deployment_dir + '/udf3.jar')
stats_udf4 = cluster.fs.stats(deployment_dir + '/udf4.jar')
stats_udf5 = cluster.fs.stats(deployment_dir + '/udf5.jar')
stats_udf6 = cluster.fs.stats(deployment_dir + '/udf6.jar')
submission._copy_files('%s/workspace' % prefix, "<xml>My XML</xml>", {'prop1': 'val1'})
assert_not_equal(stats_udf1['fileId'], cluster.fs.stats(deployment_dir + '/udf1.jar')['fileId'])
assert_not_equal(stats_udf2['fileId'], cluster.fs.stats(deployment_dir + '/udf2.jar')['fileId'])
assert_not_equal(stats_udf3['fileId'], cluster.fs.stats(deployment_dir + '/udf3.jar')['fileId'])
assert_equal(stats_udf4['fileId'], cluster.fs.stats(deployment_dir + '/udf4.jar')['fileId'])
assert_not_equal(stats_udf5['fileId'], cluster.fs.stats(deployment_dir + '/udf5.jar')['fileId'])
assert_equal(stats_udf6['fileId'], cluster.fs.stats(deployment_dir + '/udf6.jar')['fileId'])
# Test _create_file()
submission._create_file(deployment_dir, 'test.txt', data='Test data')
assert_true(cluster.fs.exists(deployment_dir + '/test.txt'), list_dir_workspace)
finally:
try:
cluster.fs.rmtree(prefix)
except:
LOG.exception('failed to remove %s' % prefix)
def test_copy_files():
cluster = pseudo_hdfs4.shared_cluster()
try:
c = make_logged_in_client()
user = User.objects.get(username='******')
ensure_home_directory(cluster.fs, user)
prefix = '/tmp/test_copy_files'
if cluster.fs.exists(prefix):
cluster.fs.rmtree(prefix)
# Jars in various locations
deployment_dir = '%s/workspace' % prefix
external_deployment_dir = '%s/deployment' % prefix
jar_1 = '%s/udf1.jar' % prefix
jar_2 = '%s/lib/udf2.jar' % prefix
jar_3 = '%s/udf3.jar' % deployment_dir
jar_4 = '%s/lib/udf4.jar' % deployment_dir # Doesn't move
jar_5 = 'udf5.jar'
jar_6 = 'lib/udf6.jar' # Doesn't move
cluster.fs.mkdir(prefix)
cluster.fs.create(jar_1)
cluster.fs.create(jar_2)
cluster.fs.create(jar_3)
cluster.fs.create(jar_4)
cluster.fs.create(deployment_dir + '/' + jar_5)
cluster.fs.create(deployment_dir + '/' + jar_6)
class MockJob(object):
XML_FILE_NAME = 'workflow.xml'
def __init__(self):
self.deployment_dir = deployment_dir
self.nodes = [
Node({
'id': '1',
'type': 'mapreduce',
'properties': {
'jar_path': jar_1
}
}),
Node({
'id': '2',
'type': 'mapreduce',
'properties': {
'jar_path': jar_2
}
}),
Node({
'id': '3',
'type': 'java',
'properties': {
'jar_path': jar_3
}
}),
Node({
'id': '4',
'type': 'java',
'properties': {
'jar_path': jar_4
}
}),
# Workspace relative paths
Node({
'id': '5',
'type': 'java',
'properties': {
'jar_path': jar_5
}
}),
Node({
'id': '6',
'type': 'java',
'properties': {
'jar_path': jar_6
}
})
]
submission = Submission(user,
job=MockJob(),
fs=cluster.fs,
jt=cluster.jt)
submission._copy_files(deployment_dir, "<xml>My XML</xml>",
{'prop1': 'val1'})
submission._copy_files(external_deployment_dir, "<xml>My XML</xml>",
{'prop1': 'val1'})
assert_true(cluster.fs.exists(deployment_dir + '/workflow.xml'),
deployment_dir)
assert_true(cluster.fs.exists(deployment_dir + '/job.properties'),
deployment_dir)
# All sources still there
assert_true(cluster.fs.exists(jar_1))
assert_true(cluster.fs.exists(jar_2))
assert_true(cluster.fs.exists(jar_3))
assert_true(cluster.fs.exists(jar_4))
assert_true(cluster.fs.exists(deployment_dir + '/' + jar_5))
assert_true(cluster.fs.exists(deployment_dir + '/' + jar_6))
# Lib
deployment_dir = deployment_dir + '/lib'
external_deployment_dir = external_deployment_dir + '/lib'
if USE_LIBPATH_FOR_JARS.get():
assert_true(jar_1 in submission.properties['oozie.libpath'])
assert_true(jar_2 in submission.properties['oozie.libpath'])
assert_true(jar_3 in submission.properties['oozie.libpath'])
assert_true(jar_4 in submission.properties['oozie.libpath'])
print(deployment_dir + '/' + jar_5)
assert_true((deployment_dir + '/' + jar_5)
in submission.properties['oozie.libpath'],
submission.properties['oozie.libpath'])
assert_true((deployment_dir + '/' + jar_6)
in submission.properties['oozie.libpath'],
submission.properties['oozie.libpath'])
else:
list_dir_workspace = cluster.fs.listdir(deployment_dir)
list_dir_deployement = cluster.fs.listdir(external_deployment_dir)
# All destinations there
assert_true(cluster.fs.exists(deployment_dir + '/udf1.jar'),
list_dir_workspace)
assert_true(cluster.fs.exists(deployment_dir + '/udf2.jar'),
list_dir_workspace)
assert_true(cluster.fs.exists(deployment_dir + '/udf3.jar'),
list_dir_workspace)
assert_true(cluster.fs.exists(deployment_dir + '/udf4.jar'),
list_dir_workspace)
assert_true(cluster.fs.exists(deployment_dir + '/udf5.jar'),
list_dir_workspace)
assert_true(cluster.fs.exists(deployment_dir + '/udf6.jar'),
list_dir_workspace)
assert_true(
cluster.fs.exists(external_deployment_dir + '/udf1.jar'),
list_dir_deployement)
assert_true(
cluster.fs.exists(external_deployment_dir + '/udf2.jar'),
list_dir_deployement)
assert_true(
cluster.fs.exists(external_deployment_dir + '/udf3.jar'),
list_dir_deployement)
assert_true(
cluster.fs.exists(external_deployment_dir + '/udf4.jar'),
list_dir_deployement)
assert_true(
cluster.fs.exists(external_deployment_dir + '/udf5.jar'),
list_dir_deployement)
assert_true(
cluster.fs.exists(external_deployment_dir + '/udf6.jar'),
list_dir_deployement)
stats_udf1 = cluster.fs.stats(deployment_dir + '/udf1.jar')
stats_udf2 = cluster.fs.stats(deployment_dir + '/udf2.jar')
stats_udf3 = cluster.fs.stats(deployment_dir + '/udf3.jar')
stats_udf4 = cluster.fs.stats(deployment_dir + '/udf4.jar')
stats_udf5 = cluster.fs.stats(deployment_dir + '/udf5.jar')
stats_udf6 = cluster.fs.stats(deployment_dir + '/udf6.jar')
submission._copy_files('%s/workspace' % prefix,
"<xml>My XML</xml>", {'prop1': 'val1'})
assert_not_equal(
stats_udf1['fileId'],
cluster.fs.stats(deployment_dir + '/udf1.jar')['fileId'])
assert_not_equal(
stats_udf2['fileId'],
cluster.fs.stats(deployment_dir + '/udf2.jar')['fileId'])
assert_not_equal(
stats_udf3['fileId'],
cluster.fs.stats(deployment_dir + '/udf3.jar')['fileId'])
assert_equal(
stats_udf4['fileId'],
cluster.fs.stats(deployment_dir + '/udf4.jar')['fileId'])
assert_not_equal(
stats_udf5['fileId'],
cluster.fs.stats(deployment_dir + '/udf5.jar')['fileId'])
assert_equal(
stats_udf6['fileId'],
cluster.fs.stats(deployment_dir + '/udf6.jar')['fileId'])
# Test _create_file()
submission._create_file(deployment_dir, 'test.txt', data='Test data')
assert_true(cluster.fs.exists(deployment_dir + '/test.txt'),
list_dir_workspace)
finally:
try:
cluster.fs.rmtree(prefix)
except:
LOG.exception('failed to remove %s' % prefix)
def test_copy_files():
cluster = pseudo_hdfs4.shared_cluster()
try:
c = make_logged_in_client()
user = User.objects.get(username='******')
prefix = '/tmp/test_copy_files'
if cluster.fs.exists(prefix):
cluster.fs.rmtree(prefix)
# Jars in various locations
deployment_dir = '%s/workspace' % prefix
external_deployment_dir = '%s/deployment' % prefix
jar_1 = '%s/udf1.jar' % prefix
jar_2 = '%s/lib/udf2.jar' % prefix
jar_3 = '%s/udf3.jar' % deployment_dir
jar_4 = '%s/lib/udf4.jar' % deployment_dir # Never move
cluster.fs.mkdir(prefix)
cluster.fs.create(jar_1)
cluster.fs.create(jar_2)
cluster.fs.create(jar_3)
cluster.fs.create(jar_4)
class MockNode():
def __init__(self, jar_path):
self.jar_path = jar_path
class MockJob():
XML_FILE_NAME = 'workflow.xml'
def __init__(self):
self.node_list = [
MockNode(jar_1),
MockNode(jar_2),
MockNode(jar_3),
MockNode(jar_4),
]
submission = Submission(user, job=MockJob(), fs=cluster.fs, jt=cluster.jt)
submission._copy_files(deployment_dir, "<xml>My XML</xml>", {'prop1': 'val1'})
submission._copy_files(external_deployment_dir, "<xml>My XML</xml>", {'prop1': 'val1'})
assert_true(cluster.fs.exists(deployment_dir + '/workflow.xml'), deployment_dir)
assert_true(cluster.fs.exists(deployment_dir + '/job.properties'), deployment_dir)
# All sources still there
assert_true(cluster.fs.exists(jar_1))
assert_true(cluster.fs.exists(jar_2))
assert_true(cluster.fs.exists(jar_3))
assert_true(cluster.fs.exists(jar_4))
deployment_dir = deployment_dir + '/lib'
external_deployment_dir = external_deployment_dir + '/lib'
list_dir_workspace = cluster.fs.listdir(deployment_dir)
list_dir_deployement = cluster.fs.listdir(external_deployment_dir)
# All destinations there
assert_true(cluster.fs.exists(deployment_dir + '/udf1.jar'), list_dir_workspace)
assert_true(cluster.fs.exists(deployment_dir + '/udf2.jar'), list_dir_workspace)
assert_true(cluster.fs.exists(deployment_dir + '/udf3.jar'), list_dir_workspace)
assert_true(cluster.fs.exists(deployment_dir + '/udf4.jar'), list_dir_workspace)
assert_true(cluster.fs.exists(external_deployment_dir + '/udf1.jar'), list_dir_deployement)
assert_true(cluster.fs.exists(external_deployment_dir + '/udf2.jar'), list_dir_deployement)
assert_true(cluster.fs.exists(external_deployment_dir + '/udf3.jar'), list_dir_deployement)
assert_true(cluster.fs.exists(external_deployment_dir + '/udf4.jar'), list_dir_deployement)
stats_udf1 = cluster.fs.stats(deployment_dir + '/udf1.jar')
stats_udf2 = cluster.fs.stats(deployment_dir + '/udf2.jar')
stats_udf3 = cluster.fs.stats(deployment_dir + '/udf3.jar')
stats_udf4 = cluster.fs.stats(deployment_dir + '/udf4.jar')
submission._copy_files('%s/workspace' % prefix, "<xml>My XML</xml>", {'prop1': 'val1'})
assert_not_equal(stats_udf1['fileId'], cluster.fs.stats(deployment_dir + '/udf1.jar')['fileId'])
assert_not_equal(stats_udf2['fileId'], cluster.fs.stats(deployment_dir + '/udf2.jar')['fileId'])
assert_not_equal(stats_udf3['fileId'], cluster.fs.stats(deployment_dir + '/udf3.jar')['fileId'])
assert_equal(stats_udf4['fileId'], cluster.fs.stats(deployment_dir + '/udf4.jar')['fileId'])
finally:
try:
cluster.fs.rmtree(prefix)
except:
pass