def test_portal_worker(self):
self._prepare_test()
map_task = dp_pb.MapTask()
map_task.output_base_dir = self._partition_output_dir
map_task.output_partition_num = self._output_partition_num
map_task.partition_id = 0
map_task.task_name = 'map_part_{}'.format(map_task.partition_id)
map_task.part_field = 'example_id'
map_task.data_portal_type = dp_pb.DataPortalType.Streaming
for partition_id in range(self._input_partition_num):
map_task.fpaths.append(self._get_input_fpath(partition_id))
# partitioner
task = dp_pb.NewTaskResponse()
task.map_task.CopyFrom(map_task)
self._portal_worker._run_map_task(task.map_task)
self._check_partitioner(task.map_task)
# merge
total_cnt = 0
for partition_id in range(self._output_partition_num):
reduce_task = dp_pb.ReduceTask()
reduce_task.map_base_dir = self._partition_output_dir
reduce_task.reduce_base_dir = self._merge_output_dir
reduce_task.partition_id = partition_id
reduce_task.task_name = 'reduce_part_{}'.format(partition_id)
self._portal_worker._run_reduce_task(reduce_task)
total_cnt += self._check_merge(reduce_task)
self.assertEqual(total_cnt,
self._partition_item_num * self._input_partition_num)
self._clean_up()
def test_portal_worker(self):
self._prepare_test()
map_task = dp_pb.MapTask()
map_task.output_base_dir = self._partition_output_dir
map_task.output_partition_num = self._output_partition_num
map_task.partition_id = 0
for partition_id in range(self._input_partition_num):
map_task.fpaths.append(self._get_input_fpath(partition_id))
# partitioner
task = dp_pb.NewTaskResponse()
task.map_task.CopyFrom(map_task)
self._portal_worker._run_map_task(task.map_task)
self._check_partitioner(task.map_task)
# merge
total_cnt = 0
for partition_id in range(self._output_partition_num):
reduce_task = dp_pb.ReduceTask()
reduce_task.map_base_dir = self._partition_output_dir
reduce_task.reduce_base_dir = self._merge_output_dir
reduce_task.partition_id = partition_id
self._portal_worker._run_reduce_task(reduce_task)
total_cnt += self._check_merge(reduce_task)
self.assertEqual(total_cnt, self._partition_item_num * self._input_partition_num)
self._clean_up()
def _create_reduce_task(self, partition_id):
assert self._processing_job is not None
job = self._processing_job
job_id = job.job_id
return dp_pb.ReduceTask(
map_base_dir=self._map_output_dir(job_id),
reduce_base_dir=self._reduce_output_dir(job_id),
partition_id=partition_id)
def _create_reduce_task(self, rank_id, partition_id):
assert self._processing_job is not None
job = self._processing_job
job_id = job.job_id
task_name = '{}-dp_portal_job_{:08}-part-{:04}-reduce'.format(
self._portal_manifest.name, job_id, partition_id)
logging.info("Data portal worker-%d is allocated reduce task %s for "\
"partition %d of job %d. the reduce base dir %s"\
"-----------------\n", rank_id, task_name,
partition_id, job_id, self._reduce_output_dir(job_id))
return dp_pb.ReduceTask(
task_name=task_name,
map_base_dir=self._map_output_dir(job_id),
reduce_base_dir=self._reduce_output_dir(job_id),
partition_id=partition_id)
def test_portal_worker(self):
self._prepare_test()
task = self._run_map_task()
total_cnt = self._check_partitioner(task.map_task)
self.assertEqual(total_cnt,
self._partition_item_num * self._input_partition_num)
# merge
total_cnt = 0
for partition_id in range(self._output_partition_num):
reduce_task = dp_pb.ReduceTask()
reduce_task.map_base_dir = self._partition_output_dir
reduce_task.reduce_base_dir = self._merge_output_dir
reduce_task.partition_id = partition_id
reduce_task.task_name = 'reduce_part_{}'.format(partition_id)
self._portal_worker._run_reduce_task(reduce_task)
total_cnt += self._check_merge(reduce_task)
self.assertEqual(total_cnt,
self._partition_item_num * self._input_partition_num)
self._clean_up()
