def profile_memory_per_line(
get_maximum,
per_element,
input_file,
output_file,
show_logs,
entity_ids,
config_file,
):
from klio_exec.commands import profile
config_path = config_file or "klio-job.yaml"
config_data = _get_config(config_path)
klio_config = config.KlioConfig(config_data)
_require_profile_input_data(input_file, entity_ids)
if not show_logs:
logging.disable(logging.CRITICAL)
klio_pipeline = profile.KlioPipeline(
klio_config=klio_config,
input_file=input_file,
output_file=output_file,
entity_ids=entity_ids,
)
klio_pipeline.profile(what="memory_per_line", get_maximum=get_maximum)
def wrapper(*args, **kwargs):
raw_overrides = kwargs.pop("override")
raw_templates = kwargs.pop("template")
job_dir = kwargs.pop("job_dir")
config_file = kwargs.pop("config_file")
job_dir, config_path = get_config_job_dir(job_dir, config_file)
warn_if_py2_job(job_dir)
raw_config_data = get_config_by_path(config_path)
processed_config_data = config.KlioConfigPreprocessor.process(
raw_config_data=raw_config_data,
raw_template_list=raw_templates,
raw_override_list=raw_overrides,
)
meta = KlioConfigMeta(
job_dir=job_dir,
config_file=config_file,
config_path=config_path,
)
conf = config.KlioConfig(processed_config_data)
kwargs["klio_config"] = conf
kwargs["config_meta"] = meta
func(*args, **kwargs)
def klio_config():
conf = {
"job_name": "test-job",
"version": 1,
"pipeline_options": {
"worker_harness_container_image":
("gcr.io/sigint/gke-baseline-random-music-gke"),
"region":
"some-region",
"project":
"test-project",
},
"job_config": {
"inputs": [{
"topic": "foo-topic",
"subscription": "foo-sub",
"data_location": "foo-input-location",
}],
"outputs": [{
"topic": "foo-topic-output",
"data_location": "foo-output-location",
}],
},
}
return config.KlioConfig(conf)
def run_pipeline(image_tag, direct_runner, update, config_file, blocking):
config_path = config_file or "klio-job.yaml"
config_data = _get_config(config_path)
# Prompt user to continue if runtime config file is not the same as
# the buildtime config file. Do this after _get_config since that
# will prompt the user if their config file doesn't even exist first.
if _compare_runtime_to_buildtime_config(config_path) is False:
msg = (
"The Klio config file '{}' at runtime differs from the config "
"file used when building this Docker image. If this is unexpected "
"behavior, please double check your runtime config, or rebuild "
"your Docker image with the correct config file.")
logging.warning(msg.format(config_path))
if direct_runner:
config_data["pipeline_options"]["runner"] = "direct"
job_name = config_data["job_name"]
conf_obj = config.KlioConfig(config_data)
if update is None: # if it's not explicitly set in CLI, look at config
update = conf_obj.pipeline_options.update
if blocking is None: # if it's not explicitly set in CLI, look at config
blocking = conf_obj.job_config.blocking
runtime_conf = RuntimeConfig(image_tag, direct_runner, update, blocking)
klio_pipeline = run.KlioPipeline(job_name, conf_obj, runtime_conf)
klio_pipeline.run()
def setup(self, config_data, config_file, config_override=None):
self.config_override = config_override
self.config_data = config_data
self.config_file = config_file
self.mock_warn_if_py2_job = self.mocker.Mock()
self.monkeypatch.setattr(
self.module.core_utils,
"warn_if_py2_job",
self.mock_warn_if_py2_job,
)
self.mock_get_config_job_dir.return_value = (
self.patch_os_getcwd,
config_override or config_file,
)
self.mock_get_config.return_value = config_data
self.meta = core_utils.KlioConfigMeta(
job_dir=self.patch_os_getcwd,
config_file=config_override,
config_path=config_override or config_file,
)
self.klio_config = kconfig.KlioConfig(config_data)
self.mock_klio_config = self.mocker.patch.object(
core_utils.config, "KlioConfig")
self.mock_klio_config.return_value = self.klio_config
return self.klio_config
def test_compare_runtime_to_buildtime_config(
mocker, monkeypatch, addl_runtime_data, buildtime_exists, exp_retval
):
monkeypatch.setattr(os.path, "exists", lambda x: buildtime_exists)
buildtime_data = {"job_name": "foo", "job_config": {}}
runtime_data = buildtime_data.copy()
if addl_runtime_data:
runtime_data["job_config"] = runtime_data["job_config"].copy()
runtime_data["job_config"]["foo"] = "bar"
# multiple `open` mocks: https://stackoverflow.com/a/26830397/1579977
open_name = "klio_exec.cli.open"
buildtime_data_str = yaml.dump(buildtime_data).encode("utf-8")
runtime_conf = kconfig.KlioConfig(runtime_data)
mock_open_buildtime = mocker.mock_open(read_data=buildtime_data_str)
mock_open = mocker.patch(open_name, mock_open_buildtime)
side_effect = (mock_open_buildtime.return_value,)
mock_open.side_effect = side_effect
act_retval = cli._compare_runtime_to_buildtime_config(runtime_conf)
assert exp_retval == act_retval
def klio_job_config():
conf = {
"job_name": "test-job",
"version": 1,
"pipeline_options": {
"project": "test-gcp-project"
},
"job_config": {
"events": {
"inputs": [{
"type": "pubsub",
"topic": "an-input-topic",
"subscription": "a-subscription",
}],
"outputs": [{
"type": "pubsub",
"topic": "foo-topic-output"
}],
},
"data": {
"inputs": [{
"type": "gcs",
"location": "gs://a-test-input/location"
}],
"outputs": [{
"type": "gcs",
"location": "foo-output-location"
}],
},
},
}
return config.KlioConfig(conf)
def profile_cpu(
interval,
input_file,
output_file,
plot_graph,
show_logs,
entity_ids,
config_file,
):
from klio_exec.commands import profile
config_path = config_file or "klio-job.yaml"
config_data = _get_config(config_path)
klio_config = config.KlioConfig(config_data)
_require_profile_input_data(input_file, entity_ids)
klio_pipeline = profile.KlioPipeline(
klio_config=klio_config,
input_file=input_file,
output_file=output_file,
entity_ids=entity_ids,
)
kwargs = {
"interval": interval,
"show_logs": show_logs,
"plot_graph": plot_graph,
}
output_png = klio_pipeline.profile(what="cpu", **kwargs)
if output_png:
click.echo("CPU plot graph generated at: {}".format(output_png))
def stop_job(config_file):
job_dir = os.path.abspath(os.getcwd())
config_file = config_file or "klio-job.yaml"
config_path = os.path.join(job_dir, config_file)
config_data = _get_config(config_path)
conf_obj = config.KlioConfig(config_data)
# TODO: make this a click option once draining is supported @lynn
strategy = "cancel"
stop.stop(conf_obj, strategy)
def audit_job(config_file):
# NOTE: we assume that audit_job is the only method called in this
# subprocess, so setting KLIO_TEST_MODE will not impact subsequent
# method calls
os.environ["KLIO_TEST_MODE"] = "true"
job_dir = os.path.abspath(os.getcwd())
config_data = _get_config(config_file or "klio-job.yaml")
conf_obj = config.KlioConfig(config_data)
audit.audit(job_dir, conf_obj)
def test_klio_config(config_dict, final_config_dict):
config_obj = config.KlioConfig(config_dict, config_skip_preprocessing=True)
assert "test-job" == config_obj.job_name
assert isinstance(config_obj.job_config, config.KlioJobConfig)
assert isinstance(config_obj.pipeline_options, config.KlioPipelineConfig)
assert final_config_dict == config_obj.as_dict()
repr_actual = repr(config_obj)
assert "KlioConfig(job_name='test-job')" == repr_actual
def test_no_gcp_klio_config(no_gcp_config_dict):
config_obj = config.KlioConfig(no_gcp_config_dict)
assert "test-job" == config_obj.job_name
assert isinstance(config_obj.job_config, config.KlioJobConfig)
assert isinstance(config_obj.pipeline_options, config.KlioPipelineConfig)
# Default variables are added to the pipeline config
assert config_dict != config_obj.as_dict()
repr_actual = repr(config_obj)
assert "KlioConfig(job_name='test-job')" == repr_actual
def get_config():
"""Load KlioConfig object."""
config_path = os.path.join(os.path.dirname(__file__), "..",
"klio-job.yaml")
try:
with open(config_path) as f:
cfg_dict = yaml.safe_load(f)
return config.KlioConfig(cfg_dict)
except IOError as e:
logging.error(e)
raise SystemExit(1)
def test_get_environment(monkeypatch, project, config_data, klio_pipeline):
config_data["pipeline_options"]["project"] = project
config = kconfig.KlioConfig(config_data)
monkeypatch.setattr(klio_pipeline, "klio_config", config)
gcreds = "/usr/gcloud/application_default_credentials.json"
exp_envs = {
"PYTHONPATH": "/usr/src/app",
"GOOGLE_APPLICATION_CREDENTIALS": gcreds,
"USER": "******",
}
if project:
exp_envs["GOOGLE_CLOUD_PROJECT"] = project
actual_env = klio_pipeline._get_environment()
assert exp_envs == actual_env
def _load_config_from_file(cls):
# [Klio v2] this may get expensive, to always be reading config
# from a file. Can this be replaced by something in memory
# that's also globally accessible?
klio_job_file = "/usr/src/config/.effective-klio-job.yaml"
# for backwards compatibility, and user is using setup.py and we
# have to find it somewhere...
if not os.path.exists(klio_job_file):
# use iterator so we don't waste time searching everywhere upfront
files = glob.iglob("/usr/**/klio-job.yaml", recursive=True)
for f in files:
klio_job_file = f
# only grab the first one
break
with open(klio_job_file, "r") as f:
all_config_data = yaml.safe_load(f)
return config.KlioConfig(all_config_data)
def test_build(conf_file, exp_image_tag, mocker, monkeypatch):
mock_docker = mocker.Mock()
mock_client = mocker.Mock()
mock_docker.from_env.return_value = mock_client
monkeypatch.setattr(build_image, "docker", mock_docker)
mock_docker_utils = mocker.Mock()
monkeypatch.setattr(build_image, "docker_utils", mock_docker_utils)
mock_config = {
"job_name": "test-job",
"version": 1,
"pipeline_options": {
"worker_harness_container_image": "gcr.register.io/squad/feature"
},
"job_config": {
"inputs": [{
"topic": "foo-topic",
"subscription": "foo-sub",
"data_location": "foo-input-location",
}],
"outputs": [{
"topic": "foo-topic-output",
"data_location": "foo-output-location",
}],
},
}
conf_obj = config.KlioConfig(mock_config)
job_dir = "jerbs"
image_tag = "v1"
build_image.build(job_dir, conf_obj, conf_file, image_tag)
mock_docker.from_env.assert_called_once_with()
mock_docker_utils.check_docker_connection.assert_called_once_with(
mock_client)
mock_docker_utils.check_dockerfile_present.assert_called_once_with(job_dir)
mock_docker_utils.build_docker_image.assert_called_once_with(
job_dir,
conf_obj.pipeline_options.worker_harness_container_image,
exp_image_tag,
conf_file,
)
def _run_pipeline(input_file, show_logs, entity_ids, config_file):
from klio_exec.commands import profile
config_path = config_file or "klio-job.yaml"
config_data = _get_config(config_path)
klio_config = config.KlioConfig(config_data)
# safety check, even though it should be invoked by another klioexec
# command
_require_profile_input_data(input_file, entity_ids)
if not show_logs:
logging.disable(logging.CRITICAL)
klio_pipeline = profile.KlioPipeline(klio_config=klio_config,
input_file=input_file,
entity_ids=entity_ids)
klio_pipeline.profile(what="run")
def test_config_pickling(config_dict, final_config_dict):
# This test attempts to verify that class-level attributes aren't used as
# instance attributes, since they are not pickled, which can result in
# missing/wrong values when config is unpickled on dataflow workers
def get_class_attributes(cls):
attrs = {}
for key in cls.__dict__:
value = getattr(cls, key)
is_fn = inspect.ismethod(value) or inspect.isfunction(value)
if not key.startswith("__") and not is_fn:
attrs[key] = copy.copy(value)
return attrs
classes = [
config.KlioConfig,
config.KlioJobConfig,
config.KlioPipelineConfig,
]
cls_attribs = {}
for cls in classes:
cls_attribs[cls] = get_class_attributes(cls)
klio_config = config.KlioConfig(
config_dict, config_skip_preprocessing=True
)
pickled = dill.dumps(klio_config)
# reset any class-level attributes back to whatever value they had before
# instantiating KlioConfig
for cls, keyvals in cls_attribs.items():
for key, value in keyvals.items():
setattr(cls, key, value)
unpickled = dill.loads(pickled)
actual = unpickled.as_dict()
assert final_config_dict == actual
def profile_wall_time(input_file, output_file, iterations, show_logs,
entity_ids, config_file):
from klio_exec.commands import profile
config_path = config_file or "klio-job.yaml"
config_data = _get_config(config_path)
klio_config = config.KlioConfig(config_data)
_require_profile_input_data(input_file, entity_ids)
if not show_logs:
logging.disable(logging.CRITICAL)
klio_pipeline = profile.KlioPipeline(
klio_config=klio_config,
input_file=input_file,
output_file=output_file,
entity_ids=entity_ids,
)
klio_pipeline.profile(what="timeit", iterations=iterations)
def klio_config():
config_dict = {
"job_name": "test-job",
"version": 1,
"pipeline_options": {
"worker_harness_container_image": "test-image"
},
"job_config": {
"inputs": [{
"topic": "foo-topic",
"subscription": "foo-sub",
"data_location": "foo-input-location",
}],
"outputs": [{
"topic": "foo-topic-output",
"data_location": "foo-output-locaiton",
}],
},
}
return config.KlioConfig(config_dict)
def develop_job(job_dir, config_file, **kwargs):
job_dir, config_path = core_utils.get_config_job_dir(job_dir, config_file)
config_data = core_utils.get_config_by_path(config_path)
conf = config.KlioConfig(config_data)
git_sha = cli_utils.get_git_sha(job_dir, kwargs.get("image_tag"))
image_tag = kwargs.get("image_tag") or git_sha
if config_file:
basename = os.path.basename(config_file)
image_tag = "{}-{}".format(image_tag, basename)
runtime_config = main_cli.DockerRuntimeConfig(
image_tag=image_tag,
force_build=kwargs.get("force_build"),
config_file_override=config_file,
)
klio_pipeline = develop.DevelopKlioContainer(job_dir, conf, runtime_config,
kwargs["klio_path"],
kwargs["exclude"])
klio_pipeline.run()
def test_job(pytest_args):
"""Thin wrapper around pytest. Any arguments after -- are passed through.
"""
import os
import pytest
config_path = "klio-job.yaml"
config_data = _get_config(config_path)
conf_obj = config.KlioConfig(config_data)
# RunConfig ensures config is pickled and sent to worker. Note this
# depends on save_main_session being True
klio_transforms_core.RunConfig.set(conf_obj)
# NOTE: we assume that test_job is the only method called in this
# subprocess, so setting KLIO_TEST_MODE will not impact subsequent
# method calls
os.environ["KLIO_TEST_MODE"] = "true"
exit_code = pytest.main(list(pytest_args))
if exit_code != 0:
raise SystemExit("Tests failed with exit code %s" % exit_code)
def _load_config_from_file(cls):
klio_job_file = None
if os.path.exists(config_core.WORKER_RUN_EFFECTIVE_CONFIG_PATH):
klio_job_file = config_core.WORKER_RUN_EFFECTIVE_CONFIG_PATH
else:
run_config_path = os.path.join(
"/usr/**", config_core.RUN_EFFECTIVE_CONFIG_FILE)
files = glob.iglob(run_config_path, recursive=True)
for f in files:
klio_job_file = f
# only grab the first one
break
if not klio_job_file:
klio_job_file = "/usr/src/config/.effective-klio-job.yaml"
logger = logging.getLogger("klio")
logger.debug(f"Loading config file from {klio_job_file}.")
with open(klio_job_file, "r") as f:
all_config_data = yaml.safe_load(f)
return config.KlioConfig(all_config_data)
def test_test_job(
runner,
mocker,
config_file,
patch_os_getcwd,
pytest_args,
conf_override,
image_tag,
mock_get_git_sha,
mock_warn_if_py2_job,
mock_get_config_job_dir,
):
mock_test_pipeline = mocker.patch.object(cli.job_commands.test,
"TestPipeline")
mock_test_pipeline.return_value.run.return_value = 0
mock_get_config_job_dir.return_value = (
patch_os_getcwd,
conf_override or config_file,
)
cli_inputs = ["job", "test"]
if image_tag:
cli_inputs.extend(["--image-tag", image_tag])
if conf_override:
cli_inputs.extend(["--config-file", conf_override])
cli_inputs.extend(pytest_args)
config_data = {
"job_name": "test-job",
"pipeline_options": {
"worker_harness_container_image": "gcr.register.io/squad/feature",
"project": "test-project",
"region": "boonies",
"staging_location": "gs://somewhere/over/the/rainbow",
"temp_location": "gs://somewhere/over/the/rainbow",
},
"job_config": {
"inputs": [{
"topic": "foo-topic",
"subscription": "foo-sub",
"data_location": "foo-input-location",
}],
"outputs": [{
"topic": "foo-topic-output",
"data_location": "foo-output-location",
}],
},
}
mock_get_config = mocker.patch.object(core_utils, "get_config_by_path")
# deepcopy since KlioConfig will pop keys
mock_get_config.return_value = config_data
conf = kconfig.KlioConfig(copy.deepcopy(config_data))
mock_klio_config = mocker.patch.object(core_utils.config, "KlioConfig")
mock_klio_config.return_value = conf
result = runner.invoke(cli.main, cli_inputs)
core_testing.assert_execution_success(result)
assert "" == result.output
exp_image_tag = image_tag or mock_get_git_sha.return_value
if conf_override:
exp_image_tag = "{}-{}".format(exp_image_tag, conf_override)
mock_get_config_job_dir.assert_called_once_with(None, conf_override)
mock_warn_if_py2_job.assert_called_once_with(patch_os_getcwd)
if not image_tag:
mock_get_git_sha.assert_called_once_with(patch_os_getcwd)
else:
mock_get_git_sha.assert_not_called()
mock_get_config.assert_called_once_with(conf_override or config_file)
mock_klio_config.assert_called_once_with(config_data)
exp_docker_runtime_config = cli.DockerRuntimeConfig(
image_tag=exp_image_tag,
force_build=False,
config_file_override=conf_override,
)
mock_test_pipeline.assert_called_once_with(patch_os_getcwd, conf,
exp_docker_runtime_config)
mock_test_pipeline.return_value.run.assert_called_once_with(
pytest_args=pytest_args)
def _get_effective_config(self):
conf = kconfig.KlioConfig(self.config_data)
effective_config = conf.as_dict()
return self._order_config_keys(effective_config)
def klio_config(config):
return kconfig.KlioConfig(config)
def _load_klio_config(cls):
config_file_path = os.path.join(os.path.dirname(__file__),
"klio-job.yaml")
with open(config_file_path) as f:
return config.KlioConfig(yaml.safe_load(f))
def config():
return _config()
@pytest.fixture
def klio_config(config):
return kconfig.KlioConfig(config)
# NOTE: Python decorators are evaluated on import, and so importing
# `klio_exec.commands.profile` (which imports `klio.transforms.helpers`, which
# imports `klio.transforms.decorators`) and `klio_exec.cli` triggers the code
# in those decorators to get evaluated. Therefore, we must patch this part in
# order to import it, otherwise it will try to load the non-existant
# `/usr/src/config/.effective-klio-job.yaml`
mock_config = kconfig.KlioConfig(_config())
patcher = mock.patch(
"klio.transforms.core.RunConfig.get",
lambda: mock_config,
)
patcher.start()
from klio_exec.commands import profile # noqa E402
from klio_exec import cli # noqa E402
@pytest.fixture
def patch_get_config(monkeypatch, config):
monkeypatch.setattr(cli, "_get_config", lambda x: config)
def _klio_config():
return config.KlioConfig(_config_dict())
def klio_config(config_dict):
return config.KlioConfig(config_dict)
