def test_resourceprofile(self):
rp_builder = ResourceProfileBuilder()
ereqs = ExecutorResourceRequests().cores(2).memory(
"6g").memoryOverhead("1g")
ereqs.pysparkMemory("2g").resource("gpu", 2, "testGpus", "nvidia.com")
treqs = TaskResourceRequests().cpus(2).resource("gpu", 2)
def assert_request_contents(exec_reqs, task_reqs):
self.assertEqual(len(exec_reqs), 5)
self.assertEqual(exec_reqs["cores"].amount, 2)
self.assertEqual(exec_reqs["memory"].amount, 6144)
self.assertEqual(exec_reqs["memoryOverhead"].amount, 1024)
self.assertEqual(exec_reqs["pyspark.memory"].amount, 2048)
self.assertEqual(exec_reqs["gpu"].amount, 2)
self.assertEqual(exec_reqs["gpu"].discoveryScript, "testGpus")
self.assertEqual(exec_reqs["gpu"].resourceName, "gpu")
self.assertEqual(exec_reqs["gpu"].vendor, "nvidia.com")
self.assertEqual(len(task_reqs), 2)
self.assertEqual(task_reqs["cpus"].amount, 2.0)
self.assertEqual(task_reqs["gpu"].amount, 2.0)
assert_request_contents(ereqs.requests, treqs.requests)
rp = rp_builder.require(ereqs).require(treqs).build
assert_request_contents(rp.executorResources, rp.taskResources)
rdd = self.sc.parallelize(range(10)).withResources(rp)
return_rp = rdd.getResourceProfile()
assert_request_contents(return_rp.executorResources,
return_rp.taskResources)
rddWithoutRp = self.sc.parallelize(range(10))
self.assertEqual(rddWithoutRp.getResourceProfile(), None)
def test_profile_before_sc(self):
rpb = ResourceProfileBuilder()
ereqs = ExecutorResourceRequests().cores(2).memory(
"6g").memoryOverhead("1g")
ereqs.pysparkMemory("2g").offheapMemory("3g").resource(
"gpu", 2, "testGpus", "nvidia.com")
treqs = TaskResourceRequests().cpus(2).resource("gpu", 2)
def assert_request_contents(exec_reqs, task_reqs):
self.assertEqual(len(exec_reqs), 6)
self.assertEqual(exec_reqs["cores"].amount, 2)
self.assertEqual(exec_reqs["memory"].amount, 6144)
self.assertEqual(exec_reqs["memoryOverhead"].amount, 1024)
self.assertEqual(exec_reqs["pyspark.memory"].amount, 2048)
self.assertEqual(exec_reqs["offHeap"].amount, 3072)
self.assertEqual(exec_reqs["gpu"].amount, 2)
self.assertEqual(exec_reqs["gpu"].discoveryScript, "testGpus")
self.assertEqual(exec_reqs["gpu"].resourceName, "gpu")
self.assertEqual(exec_reqs["gpu"].vendor, "nvidia.com")
self.assertEqual(len(task_reqs), 2)
self.assertEqual(task_reqs["cpus"].amount, 2.0)
self.assertEqual(task_reqs["gpu"].amount, 2.0)
assert_request_contents(ereqs.requests, treqs.requests)
rp = rpb.require(ereqs).require(treqs).build
assert_request_contents(rp.executorResources, rp.taskResources)
from pyspark import SparkContext, SparkConf
sc = SparkContext(conf=SparkConf())
rdd = sc.parallelize(range(10)).withResources(rp)
return_rp = rdd.getResourceProfile()
assert_request_contents(return_rp.executorResources,
return_rp.taskResources)
# intermix objects created before SparkContext init and after
rpb2 = ResourceProfileBuilder()
# use reqs created before SparkContext with Builder after
rpb2.require(ereqs)
rpb2.require(treqs)
rp2 = rpb2.build
self.assertTrue(rp2.id > 0)
rdd2 = sc.parallelize(range(10)).withResources(rp2)
return_rp2 = rdd2.getResourceProfile()
assert_request_contents(return_rp2.executorResources,
return_rp2.taskResources)
ereqs2 = ExecutorResourceRequests().cores(2).memory(
"6g").memoryOverhead("1g")
ereqs.pysparkMemory("2g").resource("gpu", 2, "testGpus", "nvidia.com")
treqs2 = TaskResourceRequests().cpus(2).resource("gpu", 2)
# use reqs created after SparkContext with Builder before
rpb.require(ereqs2)
rpb.require(treqs2)
rp3 = rpb.build
assert_request_contents(rp3.executorResources, rp3.taskResources)
sc.stop()