def test_dagster_telemetry_enabled(caplog):
with instance_for_test(overrides={"telemetry": {"enabled": True}}):
runner = CliRunner()
with pushd(path_to_file("")):
pipeline_attribute = "foo_pipeline"
pipeline_name = "foo"
result = runner.invoke(
pipeline_execute_command,
[
"-f",
path_to_file("test_cli_commands.py"),
"-a",
pipeline_attribute,
],
)
for record in caplog.records:
message = json.loads(record.getMessage())
if message.get("action") == UPDATE_REPO_STATS:
metadata = message.get("metadata")
assert metadata.get("pipeline_name_hash") == hash_name(pipeline_name)
assert metadata.get("num_pipelines_in_repo") == str(1)
assert metadata.get("repo_hash") == hash_name(
get_ephemeral_repository_name(pipeline_name)
)
assert set(message.keys()) == EXPECTED_KEYS
assert len(caplog.records) == 5
assert result.exit_code == 0
def test_repo_stats(caplog):
with tempfile.TemporaryDirectory() as temp_dir:
with instance_for_test(temp_dir=temp_dir, overrides={"telemetry": {"enabled": True}}):
runner = CliRunner(env={"DAGSTER_HOME": temp_dir})
with pushd(path_to_file("")):
pipeline_name = "multi_mode_with_resources"
result = runner.invoke(
pipeline_execute_command,
[
"-f",
file_relative_path(__file__, "../../general_tests/test_repository.py"),
"-a",
"dagster_test_repository",
"-p",
pipeline_name,
"--preset",
"add",
"--tags",
'{ "foo": "bar" }',
],
)
assert result.exit_code == 0, result.stdout
for record in caplog.records:
message = json.loads(record.getMessage())
if message.get("action") == UPDATE_REPO_STATS:
metadata = message.get("metadata")
assert metadata.get("pipeline_name_hash") == hash_name(pipeline_name)
assert metadata.get("num_pipelines_in_repo") == str(4)
assert metadata.get("repo_hash") == hash_name("dagster_test_repository")
assert set(message.keys()) == EXPECTED_KEYS
assert len(caplog.records) == 5
assert result.exit_code == 0
def test_dagster_telemetry_enabled(caplog):
with seven.TemporaryDirectory() as temp_dir:
with environ({'DAGSTER_HOME': temp_dir}):
with open(os.path.join(temp_dir, 'dagster.yaml'), 'w') as fd:
yaml.dump({'telemetry': {
'enabled': True
}},
fd,
default_flow_style=False)
DagsterInstance.local_temp(temp_dir)
runner = CliRunner(env={'DAGSTER_HOME': temp_dir})
with pushd(path_to_file('')):
pipeline_name = 'foo_pipeline'
result = runner.invoke(
pipeline_execute_command,
[
'-f',
path_to_file('test_cli_commands.py'),
'-a',
pipeline_name,
],
)
for record in caplog.records:
message = json.loads(record.getMessage())
if message.get('action') == UPDATE_REPO_STATS:
assert message.get('pipeline_name_hash') == hash_name(
'foo')
assert message.get('num_pipelines_in_repo') == str(1)
assert message.get('repo_hash') == hash_name(
EPHEMERAL_NAME)
assert set(message.keys()) == EXPECTED_KEYS
assert len(caplog.records) == 5
assert result.exit_code == 0
def test_dagster_telemetry_unset(caplog):
with tempfile.TemporaryDirectory() as temp_dir:
with instance_for_test_tempdir(temp_dir):
runner = CliRunner(env={"DAGSTER_HOME": temp_dir})
with pushd(path_to_file("")):
pipeline_attribute = "foo_pipeline"
pipeline_name = "foo"
result = runner.invoke(
pipeline_execute_command,
[
"-f",
path_to_file("test_cli_commands.py"), "-a",
pipeline_attribute
],
)
for record in caplog.records:
message = json.loads(record.getMessage())
if message.get("action") == UPDATE_REPO_STATS:
assert message.get("pipeline_name_hash") == hash_name(
pipeline_name)
assert message.get("num_pipelines_in_repo") == str(1)
assert message.get("repo_hash") == hash_name(
get_ephemeral_repository_name(pipeline_name))
assert set(message.keys()) == EXPECTED_KEYS
assert len(caplog.records) == 5
assert result.exit_code == 0
def _create_scheduler_run(
instance,
schedule_time,
repo_location,
external_schedule,
external_pipeline,
run_request,
):
from dagster.daemon.daemon import get_telemetry_daemon_session_id
run_config = run_request.run_config
schedule_tags = run_request.tags
external_execution_plan = repo_location.get_external_execution_plan(
external_pipeline,
run_config,
external_schedule.mode,
step_keys_to_execute=None,
known_state=None,
)
execution_plan_snapshot = external_execution_plan.execution_plan_snapshot
pipeline_tags = external_pipeline.tags or {}
check_tags(pipeline_tags, "pipeline_tags")
tags = merge_dicts(pipeline_tags, schedule_tags)
tags[SCHEDULED_EXECUTION_TIME_TAG] = to_timezone(schedule_time, "UTC").isoformat()
if run_request.run_key:
tags[RUN_KEY_TAG] = run_request.run_key
log_action(
instance,
SCHEDULED_RUN_CREATED,
metadata={
"DAEMON_SESSION_ID": get_telemetry_daemon_session_id(),
"SCHEDULE_NAME_HASH": hash_name(external_schedule.name),
"repo_hash": hash_name(repo_location.name),
"pipeline_name_hash": hash_name(external_pipeline.name),
},
)
return instance.create_run(
pipeline_name=external_schedule.pipeline_name,
run_id=None,
run_config=run_config,
mode=external_schedule.mode,
solids_to_execute=external_pipeline.solids_to_execute,
step_keys_to_execute=None,
solid_selection=external_pipeline.solid_selection,
status=PipelineRunStatus.NOT_STARTED,
root_run_id=None,
parent_run_id=None,
tags=tags,
pipeline_snapshot=external_pipeline.pipeline_snapshot,
execution_plan_snapshot=execution_plan_snapshot,
parent_pipeline_snapshot=external_pipeline.parent_pipeline_snapshot,
external_pipeline_origin=external_pipeline.get_external_origin(),
pipeline_code_origin=external_pipeline.get_python_origin(),
)
def test_dagit_logs(
server_mock,
caplog,
):
with tempfile.TemporaryDirectory() as temp_dir:
with instance_for_test(temp_dir=temp_dir, overrides={"telemetry": {"enabled": True}}):
runner = CliRunner(env={"DAGSTER_HOME": temp_dir})
workspace_path = file_relative_path(__file__, "telemetry_repository.yaml")
result = runner.invoke(
ui,
["-w", workspace_path],
)
assert result.exit_code == 0, str(result.exception)
expected_repo_stats = {
hash_name("test_repository"): 1,
hash_name("dagster_test_repository"): 4,
}
actions = set()
records = []
for record in caplog.records:
try:
message = json.loads(record.getMessage())
except seven.JSONDecodeError:
continue
records.append(record)
actions.add(message.get("action"))
if message.get("action") == UPDATE_REPO_STATS:
assert message.get("pipeline_name_hash") == ""
repo_hash = message.get("repo_hash")
assert repo_hash in expected_repo_stats
expected_num_pipelines_in_repo = expected_repo_stats.get(repo_hash)
assert message.get("num_pipelines_in_repo") == str(
expected_num_pipelines_in_repo
)
assert set(message.keys()) == set(
[
"action",
"client_time",
"elapsed_time",
"event_id",
"instance_id",
"pipeline_name_hash",
"num_pipelines_in_repo",
"repo_hash",
"python_version",
"metadata",
"version",
]
)
assert actions == set([START_DAGIT_WEBSERVER, UPDATE_REPO_STATS])
assert len(records) == 3
assert server_mock.call_args_list == [mock.call()]
def _create_sensor_run(
instance, repo_location, external_sensor, external_pipeline, run_request, target_data
):
from dagster.daemon.daemon import get_telemetry_daemon_session_id
external_execution_plan = repo_location.get_external_execution_plan(
external_pipeline,
run_request.run_config,
target_data.mode,
step_keys_to_execute=None,
known_state=None,
instance=instance,
)
execution_plan_snapshot = external_execution_plan.execution_plan_snapshot
pipeline_tags = external_pipeline.tags or {}
check_tags(pipeline_tags, "pipeline_tags")
tags = merge_dicts(
merge_dicts(pipeline_tags, run_request.tags),
PipelineRun.tags_for_sensor(external_sensor),
)
if run_request.run_key:
tags[RUN_KEY_TAG] = run_request.run_key
log_action(
instance,
SENSOR_RUN_CREATED,
metadata={
"DAEMON_SESSION_ID": get_telemetry_daemon_session_id(),
"SENSOR_NAME_HASH": hash_name(external_sensor.name),
"pipeline_name_hash": hash_name(external_pipeline.name),
"repo_hash": hash_name(repo_location.name),
},
)
return instance.create_run(
pipeline_name=target_data.pipeline_name,
run_id=None,
run_config=run_request.run_config,
mode=target_data.mode,
solids_to_execute=external_pipeline.solids_to_execute,
step_keys_to_execute=None,
status=PipelineRunStatus.NOT_STARTED,
solid_selection=target_data.solid_selection,
root_run_id=None,
parent_run_id=None,
tags=tags,
pipeline_snapshot=external_pipeline.pipeline_snapshot,
execution_plan_snapshot=execution_plan_snapshot,
parent_pipeline_snapshot=external_pipeline.parent_pipeline_snapshot,
external_pipeline_origin=external_pipeline.get_external_origin(),
pipeline_code_origin=external_pipeline.get_python_origin(),
)
def test_hash_name():
pipelines = ['pipeline_1', 'pipeline_2', 'pipeline_3']
hashes = [hash_name(p) for p in pipelines]
for h in hashes:
assert len(h) == 64
assert SequenceMatcher(None, hashes[0], hashes[1]).ratio() < 0.4
assert SequenceMatcher(None, hashes[0], hashes[2]).ratio() < 0.4
assert SequenceMatcher(None, hashes[1], hashes[2]).ratio() < 0.4
def test_repo_stats(caplog):
with seven.TemporaryDirectory() as temp_dir:
with environ({"DAGSTER_HOME": temp_dir}):
with open(os.path.join(temp_dir, "dagster.yaml"), "w") as fd:
yaml.dump({}, fd, default_flow_style=False)
DagsterInstance.local_temp(temp_dir)
runner = CliRunner(env={"DAGSTER_HOME": temp_dir})
with pushd(path_to_file("")):
pipeline_name = "multi_mode_with_resources"
result = runner.invoke(
pipeline_execute_command,
[
"-f",
file_relative_path(
__file__,
"../../general_tests/test_repository.py"),
"-a",
"dagster_test_repository",
"-p",
pipeline_name,
"--preset",
"add",
"--tags",
'{ "foo": "bar" }',
],
)
assert result.exit_code == 0, result.stdout
for record in caplog.records:
message = json.loads(record.getMessage())
if message.get("action") == UPDATE_REPO_STATS:
assert message.get("pipeline_name_hash") == hash_name(
pipeline_name)
assert message.get("num_pipelines_in_repo") == str(4)
assert message.get("repo_hash") == hash_name(
"dagster_test_repository")
assert set(message.keys()) == EXPECTED_KEYS
assert len(caplog.records) == 5
assert result.exit_code == 0
def test_repo_stats(caplog):
with seven.TemporaryDirectory() as temp_dir:
with environ({'DAGSTER_HOME': temp_dir}):
with open(os.path.join(temp_dir, 'dagster.yaml'), 'w') as fd:
yaml.dump({}, fd, default_flow_style=False)
DagsterInstance.local_temp(temp_dir)
runner = CliRunner(env={'DAGSTER_HOME': temp_dir})
with pushd(path_to_file('')):
pipeline_name = 'multi_mode_with_resources'
result = runner.invoke(
pipeline_execute_command,
[
'-w',
file_relative_path(__file__, '../workspace.yaml'),
'-p',
pipeline_name,
'--preset',
'add',
'--tags',
'{ "foo": "bar" }',
],
)
assert result.exit_code == 0, result.stdout
for record in caplog.records:
message = json.loads(record.getMessage())
if message.get('action') == UPDATE_REPO_STATS:
assert message.get('pipeline_name_hash') == hash_name(
pipeline_name)
assert message.get('num_pipelines_in_repo') == str(4)
assert message.get('repo_hash') == hash_name(
'dagster_test_repository')
assert set(message.keys()) == EXPECTED_KEYS
assert len(caplog.records) == 5
assert result.exit_code == 0
def test_dagit_logs(
server_mock,
caplog,
):
with seven.TemporaryDirectory() as temp_dir:
with environ({'DAGSTER_HOME': temp_dir}):
with open(os.path.join(temp_dir, 'dagster.yaml'), 'w') as fd:
yaml.dump({}, fd, default_flow_style=False)
DagsterInstance.local_temp(temp_dir)
runner = CliRunner(env={'DAGSTER_HOME': temp_dir})
with pushd(path_to_tutorial_file('')):
result = runner.invoke(
ui,
[
'-w',
file_relative_path(__file__,
'telemetry_repository.yaml'),
],
)
assert result.exit_code == 0, str(result.exception)
actions = set()
for record in caplog.records:
message = json.loads(record.getMessage())
actions.add(message.get('action'))
if message.get('action') == UPDATE_REPO_STATS:
assert message.get('pipeline_name_hash') == ''
assert message.get('num_pipelines_in_repo') == str(4)
assert message.get('repo_hash') == hash_name(
'dagster_test_repository')
assert set(message.keys()) == set([
'action',
'client_time',
'elapsed_time',
'event_id',
'instance_id',
'pipeline_name_hash',
'num_pipelines_in_repo',
'repo_hash',
'python_version',
'metadata',
'version',
])
assert actions == set(
[START_DAGIT_WEBSERVER, UPDATE_REPO_STATS])
assert len(caplog.records) == 2
assert server_mock.call_args_list == [mock.call()]
def test_dagster_telemetry_enabled(caplog):
with seven.TemporaryDirectory() as temp_dir:
with environ({"DAGSTER_HOME": temp_dir}):
with open(os.path.join(temp_dir, "dagster.yaml"), "w") as fd:
yaml.dump({"telemetry": {
"enabled": True
}},
fd,
default_flow_style=False)
DagsterInstance.local_temp(temp_dir)
runner = CliRunner(env={"DAGSTER_HOME": temp_dir})
with pushd(path_to_file("")):
pipeline_attribute = "foo_pipeline"
pipeline_name = "foo"
result = runner.invoke(
pipeline_execute_command,
[
"-f",
path_to_file("test_cli_commands.py"),
"-a",
pipeline_attribute,
],
)
for record in caplog.records:
message = json.loads(record.getMessage())
if message.get("action") == UPDATE_REPO_STATS:
assert message.get("pipeline_name_hash") == hash_name(
pipeline_name)
assert message.get("num_pipelines_in_repo") == str(1)
assert message.get("repo_hash") == hash_name(
get_ephemeral_repository_name(pipeline_name))
assert set(message.keys()) == EXPECTED_KEYS
assert len(caplog.records) == 5
assert result.exit_code == 0
def test_dagit_logs(
server_mock,
caplog,
):
with seven.TemporaryDirectory() as temp_dir:
with environ({"DAGSTER_HOME": temp_dir}):
with open(os.path.join(temp_dir, "dagster.yaml"), "w") as fd:
yaml.dump({}, fd, default_flow_style=False)
DagsterInstance.local_temp(temp_dir)
runner = CliRunner(env={"DAGSTER_HOME": temp_dir})
result = runner.invoke(
ui,
[
"-w",
file_relative_path(__file__, "telemetry_repository.yaml"),
],
)
assert result.exit_code == 0, str(result.exception)
actions = set()
for record in caplog.records:
message = json.loads(record.getMessage())
actions.add(message.get("action"))
if message.get("action") == UPDATE_REPO_STATS:
assert message.get("pipeline_name_hash") == ""
assert message.get("num_pipelines_in_repo") == str(4)
assert message.get("repo_hash") == hash_name(
"dagster_test_repository")
assert set(message.keys()) == set([
"action",
"client_time",
"elapsed_time",
"event_id",
"instance_id",
"pipeline_name_hash",
"num_pipelines_in_repo",
"repo_hash",
"python_version",
"metadata",
"version",
])
assert actions == set([START_DAGIT_WEBSERVER, UPDATE_REPO_STATS])
assert len(caplog.records) == 2
assert server_mock.call_args_list == [mock.call()]
def create_backfill_run(instance, repo_location, external_pipeline,
external_partition_set, backfill_job, partition_data):
from dagster.daemon.daemon import get_telemetry_daemon_session_id
check.inst_param(instance, "instance", DagsterInstance)
check.inst_param(repo_location, "repo_location", RepositoryLocation)
check.inst_param(external_pipeline, "external_pipeline", ExternalPipeline)
check.inst_param(external_partition_set, "external_partition_set",
ExternalPartitionSet)
check.inst_param(backfill_job, "backfill_job", PartitionBackfill)
check.inst_param(partition_data, "partition_data",
ExternalPartitionExecutionParamData)
log_action(
instance,
BACKFILL_RUN_CREATED,
metadata={
"DAEMON_SESSION_ID": get_telemetry_daemon_session_id(),
"repo_hash": hash_name(repo_location.name),
"pipeline_name_hash": hash_name(external_pipeline.name),
},
)
tags = merge_dicts(
external_pipeline.tags,
partition_data.tags,
PipelineRun.tags_for_backfill_id(backfill_job.backfill_id),
backfill_job.tags,
)
solids_to_execute = None
solid_selection = None
if not backfill_job.from_failure and not backfill_job.reexecution_steps:
step_keys_to_execute = None
parent_run_id = None
root_run_id = None
known_state = None
if external_partition_set.solid_selection:
solids_to_execute = frozenset(
external_partition_set.solid_selection)
solid_selection = external_partition_set.solid_selection
elif backfill_job.from_failure:
last_run = _fetch_last_run(instance, external_partition_set,
partition_data.name)
if not last_run or last_run.status != PipelineRunStatus.FAILURE:
return None
return instance.create_reexecuted_run_from_failure(
last_run,
repo_location,
external_pipeline,
tags=tags,
run_config=partition_data.run_config,
mode=external_partition_set.mode,
)
elif backfill_job.reexecution_steps:
last_run = _fetch_last_run(instance, external_partition_set,
partition_data.name)
parent_run_id = last_run.run_id if last_run else None
root_run_id = (last_run.root_run_id
or last_run.run_id) if last_run else None
if parent_run_id and root_run_id:
tags = merge_dicts(tags, {
PARENT_RUN_ID_TAG: parent_run_id,
ROOT_RUN_ID_TAG: root_run_id
})
step_keys_to_execute = backfill_job.reexecution_steps
if last_run and last_run.status == PipelineRunStatus.SUCCESS:
known_state = KnownExecutionState.for_reexecution(
instance.all_logs(parent_run_id),
step_keys_to_execute,
)
else:
known_state = None
if external_partition_set.solid_selection:
solids_to_execute = frozenset(
external_partition_set.solid_selection)
solid_selection = external_partition_set.solid_selection
external_execution_plan = repo_location.get_external_execution_plan(
external_pipeline,
partition_data.run_config,
external_partition_set.mode,
step_keys_to_execute=step_keys_to_execute,
known_state=known_state,
instance=instance,
)
return instance.create_run(
pipeline_snapshot=external_pipeline.pipeline_snapshot,
execution_plan_snapshot=external_execution_plan.
execution_plan_snapshot,
parent_pipeline_snapshot=external_pipeline.parent_pipeline_snapshot,
pipeline_name=external_pipeline.name,
run_id=make_new_run_id(),
solids_to_execute=solids_to_execute,
run_config=partition_data.run_config,
mode=external_partition_set.mode,
step_keys_to_execute=step_keys_to_execute,
tags=tags,
root_run_id=root_run_id,
parent_run_id=parent_run_id,
status=PipelineRunStatus.NOT_STARTED,
external_pipeline_origin=external_pipeline.get_external_origin(),
pipeline_code_origin=external_pipeline.get_python_origin(),
solid_selection=solid_selection,
)
