metaflow-stubs 2.12.22__py2.py3-none-any.whl → 2.12.23__py2.py3-none-any.whl
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- metaflow-stubs/__init__.pyi +1029 -638
- metaflow-stubs/cards.pyi +211 -5
- metaflow-stubs/cli.pyi +23 -3
- metaflow-stubs/client/__init__.pyi +128 -3
- metaflow-stubs/client/core.pyi +227 -7
- metaflow-stubs/client/filecache.pyi +3 -3
- metaflow-stubs/clone_util.pyi +5 -2
- metaflow-stubs/events.pyi +20 -2
- metaflow-stubs/exception.pyi +2 -2
- metaflow-stubs/flowspec.pyi +71 -5
- metaflow-stubs/generated_for.txt +1 -1
- metaflow-stubs/includefile.pyi +158 -5
- metaflow-stubs/info_file.pyi +2 -2
- metaflow-stubs/metadata/metadata.pyi +9 -3
- metaflow-stubs/metadata/util.pyi +2 -2
- metaflow-stubs/metaflow_config.pyi +2 -2
- metaflow-stubs/metaflow_current.pyi +26 -23
- metaflow-stubs/mflog/mflog.pyi +2 -2
- metaflow-stubs/multicore_utils.pyi +2 -2
- metaflow-stubs/parameters.pyi +69 -3
- metaflow-stubs/plugins/__init__.pyi +14 -3
- metaflow-stubs/plugins/airflow/__init__.pyi +2 -2
- metaflow-stubs/plugins/airflow/airflow.pyi +15 -3
- metaflow-stubs/plugins/airflow/airflow_cli.pyi +66 -3
- metaflow-stubs/plugins/airflow/airflow_decorator.pyi +5 -2
- metaflow-stubs/plugins/airflow/airflow_utils.pyi +2 -2
- metaflow-stubs/plugins/airflow/exception.pyi +2 -2
- metaflow-stubs/plugins/airflow/sensors/__init__.pyi +80 -2
- metaflow-stubs/plugins/airflow/sensors/base_sensor.pyi +5 -2
- metaflow-stubs/plugins/airflow/sensors/external_task_sensor.pyi +44 -2
- metaflow-stubs/plugins/airflow/sensors/s3_sensor.pyi +44 -2
- metaflow-stubs/plugins/argo/__init__.pyi +2 -2
- metaflow-stubs/plugins/argo/argo_client.pyi +2 -2
- metaflow-stubs/plugins/argo/argo_events.pyi +16 -2
- metaflow-stubs/plugins/argo/argo_workflows.pyi +6 -4
- metaflow-stubs/plugins/argo/argo_workflows_cli.pyi +97 -6
- metaflow-stubs/plugins/argo/argo_workflows_decorator.pyi +34 -6
- metaflow-stubs/plugins/argo/argo_workflows_deployer.pyi +49 -4
- metaflow-stubs/plugins/aws/__init__.pyi +2 -2
- metaflow-stubs/plugins/aws/aws_client.pyi +2 -2
- metaflow-stubs/plugins/aws/aws_utils.pyi +2 -2
- metaflow-stubs/plugins/aws/batch/__init__.pyi +2 -2
- metaflow-stubs/plugins/aws/batch/batch.pyi +2 -2
- metaflow-stubs/plugins/aws/batch/batch_cli.pyi +2 -2
- metaflow-stubs/plugins/aws/batch/batch_client.pyi +2 -2
- metaflow-stubs/plugins/aws/batch/batch_decorator.pyi +103 -3
- metaflow-stubs/plugins/aws/secrets_manager/__init__.pyi +2 -2
- metaflow-stubs/plugins/aws/secrets_manager/aws_secrets_manager_secrets_provider.pyi +15 -3
- metaflow-stubs/plugins/aws/step_functions/__init__.pyi +2 -2
- metaflow-stubs/plugins/aws/step_functions/dynamo_db_client.pyi +2 -2
- metaflow-stubs/plugins/aws/step_functions/event_bridge_client.pyi +2 -2
- metaflow-stubs/plugins/aws/step_functions/production_token.pyi +2 -2
- metaflow-stubs/plugins/aws/step_functions/schedule_decorator.pyi +21 -2
- metaflow-stubs/plugins/aws/step_functions/step_functions.pyi +2 -2
- metaflow-stubs/plugins/aws/step_functions/step_functions_cli.pyi +69 -3
- metaflow-stubs/plugins/aws/step_functions/step_functions_client.pyi +2 -2
- metaflow-stubs/plugins/aws/step_functions/step_functions_decorator.pyi +5 -2
- metaflow-stubs/plugins/aws/step_functions/step_functions_deployer.pyi +49 -4
- metaflow-stubs/plugins/azure/__init__.pyi +2 -2
- metaflow-stubs/plugins/azure/azure_credential.pyi +2 -2
- metaflow-stubs/plugins/azure/azure_exceptions.pyi +2 -2
- metaflow-stubs/plugins/azure/azure_secret_manager_secrets_provider.pyi +21 -3
- metaflow-stubs/plugins/azure/azure_utils.pyi +2 -2
- metaflow-stubs/plugins/azure/blob_service_client_factory.pyi +2 -2
- metaflow-stubs/plugins/azure/includefile_support.pyi +2 -2
- metaflow-stubs/plugins/cards/__init__.pyi +2 -2
- metaflow-stubs/plugins/cards/card_cli.pyi +62 -4
- metaflow-stubs/plugins/cards/card_client.pyi +33 -2
- metaflow-stubs/plugins/cards/card_creator.pyi +5 -2
- metaflow-stubs/plugins/cards/card_datastore.pyi +8 -2
- metaflow-stubs/plugins/cards/card_decorator.pyi +52 -2
- metaflow-stubs/plugins/cards/card_modules/__init__.pyi +28 -2
- metaflow-stubs/plugins/cards/card_modules/basic.pyi +41 -2
- metaflow-stubs/plugins/cards/card_modules/card.pyi +28 -2
- metaflow-stubs/plugins/cards/card_modules/chevron/__init__.pyi +2 -2
- metaflow-stubs/plugins/cards/card_modules/chevron/main.pyi +2 -2
- metaflow-stubs/plugins/cards/card_modules/chevron/metadata.pyi +2 -2
- metaflow-stubs/plugins/cards/card_modules/chevron/renderer.pyi +2 -2
- metaflow-stubs/plugins/cards/card_modules/chevron/tokenizer.pyi +2 -2
- metaflow-stubs/plugins/cards/card_modules/components.pyi +184 -4
- metaflow-stubs/plugins/cards/card_modules/convert_to_native_type.pyi +5 -2
- metaflow-stubs/plugins/cards/card_modules/renderer_tools.pyi +2 -2
- metaflow-stubs/plugins/cards/card_modules/test_cards.pyi +35 -2
- metaflow-stubs/plugins/cards/card_resolver.pyi +2 -2
- metaflow-stubs/plugins/cards/component_serializer.pyi +57 -4
- metaflow-stubs/plugins/cards/exception.pyi +8 -2
- metaflow-stubs/plugins/catch_decorator.pyi +20 -3
- metaflow-stubs/plugins/datatools/__init__.pyi +62 -2
- metaflow-stubs/plugins/datatools/local.pyi +16 -2
- metaflow-stubs/plugins/datatools/s3/__init__.pyi +72 -3
- metaflow-stubs/plugins/datatools/s3/s3.pyi +82 -5
- metaflow-stubs/plugins/datatools/s3/s3tail.pyi +2 -2
- metaflow-stubs/plugins/datatools/s3/s3util.pyi +2 -2
- metaflow-stubs/plugins/debug_logger.pyi +2 -2
- metaflow-stubs/plugins/debug_monitor.pyi +2 -2
- metaflow-stubs/plugins/environment_decorator.pyi +10 -2
- metaflow-stubs/plugins/events_decorator.pyi +106 -2
- metaflow-stubs/plugins/frameworks/__init__.pyi +2 -2
- metaflow-stubs/plugins/frameworks/pytorch.pyi +24 -3
- metaflow-stubs/plugins/gcp/__init__.pyi +2 -2
- metaflow-stubs/plugins/gcp/gcp_secret_manager_secrets_provider.pyi +15 -3
- metaflow-stubs/plugins/gcp/gs_exceptions.pyi +2 -2
- metaflow-stubs/plugins/gcp/gs_storage_client_factory.pyi +2 -2
- metaflow-stubs/plugins/gcp/gs_utils.pyi +2 -2
- metaflow-stubs/plugins/gcp/includefile_support.pyi +2 -2
- metaflow-stubs/plugins/kubernetes/__init__.pyi +2 -2
- metaflow-stubs/plugins/kubernetes/kube_utils.pyi +2 -2
- metaflow-stubs/plugins/kubernetes/kubernetes.pyi +3 -3
- metaflow-stubs/plugins/kubernetes/kubernetes_cli.pyi +65 -2
- metaflow-stubs/plugins/kubernetes/kubernetes_client.pyi +2 -2
- metaflow-stubs/plugins/kubernetes/kubernetes_decorator.pyi +100 -3
- metaflow-stubs/plugins/kubernetes/kubernetes_job.pyi +2 -2
- metaflow-stubs/plugins/kubernetes/kubernetes_jobsets.pyi +5 -2
- metaflow-stubs/plugins/logs_cli.pyi +2 -2
- metaflow-stubs/plugins/package_cli.pyi +2 -2
- metaflow-stubs/plugins/parallel_decorator.pyi +29 -2
- metaflow-stubs/plugins/project_decorator.pyi +59 -2
- metaflow-stubs/plugins/pypi/__init__.pyi +2 -2
- metaflow-stubs/plugins/pypi/conda_decorator.pyi +45 -2
- metaflow-stubs/plugins/pypi/conda_environment.pyi +4 -4
- metaflow-stubs/plugins/pypi/pypi_decorator.pyi +33 -2
- metaflow-stubs/plugins/pypi/pypi_environment.pyi +3 -3
- metaflow-stubs/plugins/pypi/utils.pyi +2 -2
- metaflow-stubs/plugins/resources_decorator.pyi +33 -2
- metaflow-stubs/plugins/retry_decorator.pyi +21 -2
- metaflow-stubs/plugins/secrets/__init__.pyi +2 -2
- metaflow-stubs/plugins/secrets/inline_secrets_provider.pyi +2 -2
- metaflow-stubs/plugins/secrets/secrets_decorator.pyi +11 -2
- metaflow-stubs/plugins/storage_executor.pyi +6 -2
- metaflow-stubs/plugins/tag_cli.pyi +35 -4
- metaflow-stubs/plugins/test_unbounded_foreach_decorator.pyi +10 -3
- metaflow-stubs/plugins/timeout_decorator.pyi +23 -2
- metaflow-stubs/procpoll.pyi +2 -2
- metaflow-stubs/pylint_wrapper.pyi +2 -2
- metaflow-stubs/runner/__init__.pyi +2 -2
- metaflow-stubs/runner/deployer.pyi +70 -131
- metaflow-stubs/runner/metaflow_runner.pyi +117 -9
- metaflow-stubs/runner/nbdeploy.pyi +66 -2
- metaflow-stubs/runner/nbrun.pyi +79 -2
- metaflow-stubs/runner/subprocess_manager.pyi +16 -4
- metaflow-stubs/runner/utils.pyi +32 -2
- metaflow-stubs/system/__init__.pyi +4 -4
- metaflow-stubs/system/system_logger.pyi +3 -3
- metaflow-stubs/system/system_monitor.pyi +3 -3
- metaflow-stubs/tagging_util.pyi +2 -2
- metaflow-stubs/tuple_util.pyi +2 -2
- metaflow-stubs/version.pyi +2 -2
- {metaflow_stubs-2.12.22.dist-info → metaflow_stubs-2.12.23.dist-info}/METADATA +2 -2
- metaflow_stubs-2.12.23.dist-info/RECORD +152 -0
- metaflow_stubs-2.12.22.dist-info/RECORD +0 -152
- {metaflow_stubs-2.12.22.dist-info → metaflow_stubs-2.12.23.dist-info}/WHEEL +0 -0
- {metaflow_stubs-2.12.22.dist-info → metaflow_stubs-2.12.23.dist-info}/top_level.txt +0 -0
metaflow-stubs/__init__.pyi
CHANGED
@@ -1,25 +1,25 @@
|
|
1
1
|
##################################################################################
|
2
2
|
# Auto-generated Metaflow stub file #
|
3
|
-
# MF version: 2.12.
|
4
|
-
# Generated on 2024-
|
3
|
+
# MF version: 2.12.23 #
|
4
|
+
# Generated on 2024-10-01T14:32:39.945734 #
|
5
5
|
##################################################################################
|
6
6
|
|
7
7
|
from __future__ import annotations
|
8
8
|
|
9
9
|
import typing
|
10
10
|
if typing.TYPE_CHECKING:
|
11
|
-
import metaflow.flowspec
|
12
|
-
import metaflow.runner.metaflow_runner
|
13
|
-
import metaflow.events
|
14
|
-
import metaflow.client.core
|
15
11
|
import metaflow.plugins.datatools.s3.s3
|
16
|
-
import metaflow.
|
12
|
+
import metaflow.events
|
13
|
+
import metaflow._vendor.click.types
|
14
|
+
import io
|
17
15
|
import datetime
|
16
|
+
import metaflow.parameters
|
17
|
+
import metaflow.runner.metaflow_runner
|
18
18
|
import metaflow.datastore.inputs
|
19
|
+
import metaflow.metaflow_current
|
20
|
+
import metaflow.client.core
|
21
|
+
import metaflow.flowspec
|
19
22
|
import typing
|
20
|
-
import io
|
21
|
-
import metaflow._vendor.click.types
|
22
|
-
import metaflow.parameters
|
23
23
|
FlowSpecDerived = typing.TypeVar("FlowSpecDerived", bound="FlowSpec", contravariant=False, covariant=False)
|
24
24
|
StepFlag = typing.NewType("StepFlag", bool)
|
25
25
|
|
@@ -103,6 +103,14 @@ def metadata(ms: str) -> str:
|
|
103
103
|
...
|
104
104
|
|
105
105
|
class FlowSpec(object, metaclass=metaflow.flowspec._FlowSpecMeta):
|
106
|
+
"""
|
107
|
+
Main class from which all Flows should inherit.
|
108
|
+
|
109
|
+
Attributes
|
110
|
+
----------
|
111
|
+
index
|
112
|
+
input
|
113
|
+
"""
|
106
114
|
def __init__(self, use_cli = True):
|
107
115
|
"""
|
108
116
|
Construct a FlowSpec
|
@@ -324,6 +332,49 @@ class FlowSpec(object, metaclass=metaflow.flowspec._FlowSpecMeta):
|
|
324
332
|
...
|
325
333
|
|
326
334
|
class Parameter(object, metaclass=type):
|
335
|
+
"""
|
336
|
+
Defines a parameter for a flow.
|
337
|
+
|
338
|
+
Parameters must be instantiated as class variables in flow classes, e.g.
|
339
|
+
```
|
340
|
+
class MyFlow(FlowSpec):
|
341
|
+
param = Parameter('myparam')
|
342
|
+
```
|
343
|
+
in this case, the parameter is specified on the command line as
|
344
|
+
```
|
345
|
+
python myflow.py run --myparam=5
|
346
|
+
```
|
347
|
+
and its value is accessible through a read-only artifact like this:
|
348
|
+
```
|
349
|
+
print(self.param == 5)
|
350
|
+
```
|
351
|
+
Note that the user-visible parameter name, `myparam` above, can be
|
352
|
+
different from the artifact name, `param` above.
|
353
|
+
|
354
|
+
The parameter value is converted to a Python type based on the `type`
|
355
|
+
argument or to match the type of `default`, if it is set.
|
356
|
+
|
357
|
+
Parameters
|
358
|
+
----------
|
359
|
+
name : str
|
360
|
+
User-visible parameter name.
|
361
|
+
default : str or float or int or bool or `JSONType` or a function.
|
362
|
+
Default value for the parameter. Use a special `JSONType` class to
|
363
|
+
indicate that the value must be a valid JSON object. A function
|
364
|
+
implies that the parameter corresponds to a *deploy-time parameter*.
|
365
|
+
The type of the default value is used as the parameter `type`.
|
366
|
+
type : Type, default None
|
367
|
+
If `default` is not specified, define the parameter type. Specify
|
368
|
+
one of `str`, `float`, `int`, `bool`, or `JSONType`. If None, defaults
|
369
|
+
to the type of `default` or `str` if none specified.
|
370
|
+
help : str, optional
|
371
|
+
Help text to show in `run --help`.
|
372
|
+
required : bool, default False
|
373
|
+
Require that the user specified a value for the parameter.
|
374
|
+
`required=True` implies that the `default` is not used.
|
375
|
+
show_default : bool, default True
|
376
|
+
If True, show the default value in the help text.
|
377
|
+
"""
|
327
378
|
def __init__(self, name: str, default: typing.Union[str, float, int, bool, typing.Dict[str, typing.Any], typing.Callable[[], typing.Union[str, float, int, bool, typing.Dict[str, typing.Any]]], None] = None, type: typing.Union[typing.Type[str], typing.Type[float], typing.Type[int], typing.Type[bool], metaflow.parameters.JSONTypeClass, None] = None, help: typing.Optional[str] = None, required: bool = False, show_default: bool = True, **kwargs: typing.Dict[str, typing.Any]):
|
328
379
|
...
|
329
380
|
def __repr__(self):
|
@@ -353,6 +404,59 @@ class JSONTypeClass(metaflow._vendor.click.types.ParamType, metaclass=type):
|
|
353
404
|
JSONType: metaflow.parameters.JSONTypeClass
|
354
405
|
|
355
406
|
class S3(object, metaclass=type):
|
407
|
+
"""
|
408
|
+
The Metaflow S3 client.
|
409
|
+
|
410
|
+
This object manages the connection to S3 and a temporary diretory that is used
|
411
|
+
to download objects. Note that in most cases when the data fits in memory, no local
|
412
|
+
disk IO is needed as operations are cached by the operating system, which makes
|
413
|
+
operations fast as long as there is enough memory available.
|
414
|
+
|
415
|
+
The easiest way is to use this object as a context manager:
|
416
|
+
```
|
417
|
+
with S3() as s3:
|
418
|
+
data = [obj.blob for obj in s3.get_many(urls)]
|
419
|
+
print(data)
|
420
|
+
```
|
421
|
+
The context manager takes care of creating and deleting a temporary directory
|
422
|
+
automatically. Without a context manager, you must call `.close()` to delete
|
423
|
+
the directory explicitly:
|
424
|
+
```
|
425
|
+
s3 = S3()
|
426
|
+
data = [obj.blob for obj in s3.get_many(urls)]
|
427
|
+
s3.close()
|
428
|
+
```
|
429
|
+
You can customize the location of the temporary directory with `tmproot`. It
|
430
|
+
defaults to the current working directory.
|
431
|
+
|
432
|
+
To make it easier to deal with object locations, the client can be initialized
|
433
|
+
with an S3 path prefix. There are three ways to handle locations:
|
434
|
+
|
435
|
+
1. Use a `metaflow.Run` object or `self`, e.g. `S3(run=self)` which
|
436
|
+
initializes the prefix with the global `DATATOOLS_S3ROOT` path, combined
|
437
|
+
with the current run ID. This mode makes it easy to version data based
|
438
|
+
on the run ID consistently. You can use the `bucket` and `prefix` to
|
439
|
+
override parts of `DATATOOLS_S3ROOT`.
|
440
|
+
|
441
|
+
2. Specify an S3 prefix explicitly with `s3root`,
|
442
|
+
e.g. `S3(s3root='s3://mybucket/some/path')`.
|
443
|
+
|
444
|
+
3. Specify nothing, i.e. `S3()`, in which case all operations require
|
445
|
+
a full S3 url prefixed with `s3://`.
|
446
|
+
|
447
|
+
Parameters
|
448
|
+
----------
|
449
|
+
tmproot : str, default: '.'
|
450
|
+
Where to store the temporary directory.
|
451
|
+
bucket : str, optional
|
452
|
+
Override the bucket from `DATATOOLS_S3ROOT` when `run` is specified.
|
453
|
+
prefix : str, optional
|
454
|
+
Override the path from `DATATOOLS_S3ROOT` when `run` is specified.
|
455
|
+
run : FlowSpec or Run, optional
|
456
|
+
Derive path prefix from the current or a past run ID, e.g. S3(run=self).
|
457
|
+
s3root : str, optional
|
458
|
+
If `run` is not specified, use this as the S3 prefix.
|
459
|
+
"""
|
356
460
|
@classmethod
|
357
461
|
def get_root_from_config(cls, echo, create_on_absent = True):
|
358
462
|
...
|
@@ -644,6 +748,33 @@ class S3(object, metaclass=type):
|
|
644
748
|
...
|
645
749
|
|
646
750
|
class IncludeFile(metaflow.parameters.Parameter, metaclass=type):
|
751
|
+
"""
|
752
|
+
Includes a local file as a parameter for the flow.
|
753
|
+
|
754
|
+
`IncludeFile` behaves like `Parameter` except that it reads its value from a file instead of
|
755
|
+
the command line. The user provides a path to a file on the command line. The file contents
|
756
|
+
are saved as a read-only artifact which is available in all steps of the flow.
|
757
|
+
|
758
|
+
Parameters
|
759
|
+
----------
|
760
|
+
name : str
|
761
|
+
User-visible parameter name.
|
762
|
+
default : Union[str, Callable[ParameterContext, str]]
|
763
|
+
Default path to a local file. A function
|
764
|
+
implies that the parameter corresponds to a *deploy-time parameter*.
|
765
|
+
is_text : bool, default True
|
766
|
+
Convert the file contents to a string using the provided `encoding`.
|
767
|
+
If False, the artifact is stored in `bytes`.
|
768
|
+
encoding : str, optional, default 'utf-8'
|
769
|
+
Use this encoding to decode the file contexts if `is_text=True`.
|
770
|
+
required : bool, default False
|
771
|
+
Require that the user specified a value for the parameter.
|
772
|
+
`required=True` implies that the `default` is not used.
|
773
|
+
help : str, optional
|
774
|
+
Help text to show in `run --help`.
|
775
|
+
show_default : bool, default True
|
776
|
+
If True, show the default value in the help text.
|
777
|
+
"""
|
647
778
|
def __init__(self, name: str, required: bool = False, is_text: bool = True, encoding: str = "utf-8", help: typing.Optional[str] = None, **kwargs: typing.Dict[str, str]):
|
648
779
|
...
|
649
780
|
def load_parameter(self, v):
|
@@ -724,263 +855,429 @@ def step(f: typing.Union[typing.Callable[[FlowSpecDerived], None], typing.Callab
|
|
724
855
|
...
|
725
856
|
|
726
857
|
@typing.overload
|
727
|
-
def
|
858
|
+
def batch(*, cpu: int = 1, gpu: int = 0, memory: int = 4096, image: typing.Optional[str] = None, queue: str = "METAFLOW_BATCH_JOB_QUEUE", iam_role: str = "METAFLOW_ECS_S3_ACCESS_IAM_ROLE", execution_role: str = "METAFLOW_ECS_FARGATE_EXECUTION_ROLE", shared_memory: typing.Optional[int] = None, max_swap: typing.Optional[int] = None, swappiness: typing.Optional[int] = None, use_tmpfs: bool = False, tmpfs_tempdir: bool = True, tmpfs_size: typing.Optional[int] = None, tmpfs_path: typing.Optional[str] = None, inferentia: int = 0, trainium: int = None, efa: int = 0, ephemeral_storage: int = None, log_driver: typing.Optional[str] = None, log_options: typing.Optional[typing.List[str]] = None) -> typing.Callable[[typing.Union[typing.Callable[[FlowSpecDerived, StepFlag], None], typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None]]], typing.Union[typing.Callable[[FlowSpecDerived, StepFlag], None], typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None]]]:
|
728
859
|
"""
|
729
|
-
Specifies
|
730
|
-
|
731
|
-
Use `@resources` to specify the resource requirements
|
732
|
-
independently of the specific compute layer (`@batch`, `@kubernetes`).
|
733
|
-
|
734
|
-
You can choose the compute layer on the command line by executing e.g.
|
735
|
-
```
|
736
|
-
python myflow.py run --with batch
|
737
|
-
```
|
738
|
-
or
|
739
|
-
```
|
740
|
-
python myflow.py run --with kubernetes
|
741
|
-
```
|
742
|
-
which executes the flow on the desired system using the
|
743
|
-
requirements specified in `@resources`.
|
860
|
+
Specifies that this step should execute on [AWS Batch](https://aws.amazon.com/batch/).
|
744
861
|
|
745
862
|
Parameters
|
746
863
|
----------
|
747
864
|
cpu : int, default 1
|
748
|
-
Number of CPUs required for this step.
|
865
|
+
Number of CPUs required for this step. If `@resources` is
|
866
|
+
also present, the maximum value from all decorators is used.
|
749
867
|
gpu : int, default 0
|
750
|
-
Number of GPUs required for this step.
|
751
|
-
|
752
|
-
Disk size (in MB) required for this step. Only applies on Kubernetes.
|
868
|
+
Number of GPUs required for this step. If `@resources` is
|
869
|
+
also present, the maximum value from all decorators is used.
|
753
870
|
memory : int, default 4096
|
754
|
-
Memory size (in MB) required for this step.
|
871
|
+
Memory size (in MB) required for this step. If
|
872
|
+
`@resources` is also present, the maximum value from all decorators is
|
873
|
+
used.
|
874
|
+
image : str, optional, default None
|
875
|
+
Docker image to use when launching on AWS Batch. If not specified, and
|
876
|
+
METAFLOW_BATCH_CONTAINER_IMAGE is specified, that image is used. If
|
877
|
+
not, a default Docker image mapping to the current version of Python is used.
|
878
|
+
queue : str, default METAFLOW_BATCH_JOB_QUEUE
|
879
|
+
AWS Batch Job Queue to submit the job to.
|
880
|
+
iam_role : str, default METAFLOW_ECS_S3_ACCESS_IAM_ROLE
|
881
|
+
AWS IAM role that AWS Batch container uses to access AWS cloud resources.
|
882
|
+
execution_role : str, default METAFLOW_ECS_FARGATE_EXECUTION_ROLE
|
883
|
+
AWS IAM role that AWS Batch can use [to trigger AWS Fargate tasks]
|
884
|
+
(https://docs.aws.amazon.com/batch/latest/userguide/execution-IAM-role.html).
|
755
885
|
shared_memory : int, optional, default None
|
756
886
|
The value for the size (in MiB) of the /dev/shm volume for this step.
|
757
887
|
This parameter maps to the `--shm-size` option in Docker.
|
888
|
+
max_swap : int, optional, default None
|
889
|
+
The total amount of swap memory (in MiB) a container can use for this
|
890
|
+
step. This parameter is translated to the `--memory-swap` option in
|
891
|
+
Docker where the value is the sum of the container memory plus the
|
892
|
+
`max_swap` value.
|
893
|
+
swappiness : int, optional, default None
|
894
|
+
This allows you to tune memory swappiness behavior for this step.
|
895
|
+
A swappiness value of 0 causes swapping not to happen unless absolutely
|
896
|
+
necessary. A swappiness value of 100 causes pages to be swapped very
|
897
|
+
aggressively. Accepted values are whole numbers between 0 and 100.
|
898
|
+
use_tmpfs : bool, default False
|
899
|
+
This enables an explicit tmpfs mount for this step. Note that tmpfs is
|
900
|
+
not available on Fargate compute environments
|
901
|
+
tmpfs_tempdir : bool, default True
|
902
|
+
sets METAFLOW_TEMPDIR to tmpfs_path if set for this step.
|
903
|
+
tmpfs_size : int, optional, default None
|
904
|
+
The value for the size (in MiB) of the tmpfs mount for this step.
|
905
|
+
This parameter maps to the `--tmpfs` option in Docker. Defaults to 50% of the
|
906
|
+
memory allocated for this step.
|
907
|
+
tmpfs_path : str, optional, default None
|
908
|
+
Path to tmpfs mount for this step. Defaults to /metaflow_temp.
|
909
|
+
inferentia : int, default 0
|
910
|
+
Number of Inferentia chips required for this step.
|
911
|
+
trainium : int, default None
|
912
|
+
Alias for inferentia. Use only one of the two.
|
913
|
+
efa : int, default 0
|
914
|
+
Number of elastic fabric adapter network devices to attach to container
|
915
|
+
ephemeral_storage : int, default None
|
916
|
+
The total amount, in GiB, of ephemeral storage to set for the task, 21-200GiB.
|
917
|
+
This is only relevant for Fargate compute environments
|
918
|
+
log_driver: str, optional, default None
|
919
|
+
The log driver to use for the Amazon ECS container.
|
920
|
+
log_options: List[str], optional, default None
|
921
|
+
List of strings containing options for the chosen log driver. The configurable values
|
922
|
+
depend on the `log driver` chosen. Validation of these options is not supported yet.
|
923
|
+
Example: [`awslogs-group:aws/batch/job`]
|
758
924
|
"""
|
759
925
|
...
|
760
926
|
|
761
927
|
@typing.overload
|
762
|
-
def
|
928
|
+
def batch(f: typing.Callable[[FlowSpecDerived, StepFlag], None]) -> typing.Callable[[FlowSpecDerived, StepFlag], None]:
|
763
929
|
...
|
764
930
|
|
765
931
|
@typing.overload
|
766
|
-
def
|
932
|
+
def batch(f: typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None]) -> typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None]:
|
767
933
|
...
|
768
934
|
|
769
|
-
def
|
935
|
+
def batch(f: typing.Union[typing.Callable[[FlowSpecDerived, StepFlag], None], typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None], None] = None, *, cpu: int = 1, gpu: int = 0, memory: int = 4096, image: typing.Optional[str] = None, queue: str = "METAFLOW_BATCH_JOB_QUEUE", iam_role: str = "METAFLOW_ECS_S3_ACCESS_IAM_ROLE", execution_role: str = "METAFLOW_ECS_FARGATE_EXECUTION_ROLE", shared_memory: typing.Optional[int] = None, max_swap: typing.Optional[int] = None, swappiness: typing.Optional[int] = None, use_tmpfs: bool = False, tmpfs_tempdir: bool = True, tmpfs_size: typing.Optional[int] = None, tmpfs_path: typing.Optional[str] = None, inferentia: int = 0, trainium: int = None, efa: int = 0, ephemeral_storage: int = None, log_driver: typing.Optional[str] = None, log_options: typing.Optional[typing.List[str]] = None):
|
770
936
|
"""
|
771
|
-
Specifies
|
772
|
-
|
773
|
-
Use `@resources` to specify the resource requirements
|
774
|
-
independently of the specific compute layer (`@batch`, `@kubernetes`).
|
775
|
-
|
776
|
-
You can choose the compute layer on the command line by executing e.g.
|
777
|
-
```
|
778
|
-
python myflow.py run --with batch
|
779
|
-
```
|
780
|
-
or
|
781
|
-
```
|
782
|
-
python myflow.py run --with kubernetes
|
783
|
-
```
|
784
|
-
which executes the flow on the desired system using the
|
785
|
-
requirements specified in `@resources`.
|
937
|
+
Specifies that this step should execute on [AWS Batch](https://aws.amazon.com/batch/).
|
786
938
|
|
787
939
|
Parameters
|
788
940
|
----------
|
789
941
|
cpu : int, default 1
|
790
|
-
Number of CPUs required for this step.
|
942
|
+
Number of CPUs required for this step. If `@resources` is
|
943
|
+
also present, the maximum value from all decorators is used.
|
791
944
|
gpu : int, default 0
|
792
|
-
Number of GPUs required for this step.
|
793
|
-
|
794
|
-
Disk size (in MB) required for this step. Only applies on Kubernetes.
|
945
|
+
Number of GPUs required for this step. If `@resources` is
|
946
|
+
also present, the maximum value from all decorators is used.
|
795
947
|
memory : int, default 4096
|
796
|
-
Memory size (in MB) required for this step.
|
948
|
+
Memory size (in MB) required for this step. If
|
949
|
+
`@resources` is also present, the maximum value from all decorators is
|
950
|
+
used.
|
951
|
+
image : str, optional, default None
|
952
|
+
Docker image to use when launching on AWS Batch. If not specified, and
|
953
|
+
METAFLOW_BATCH_CONTAINER_IMAGE is specified, that image is used. If
|
954
|
+
not, a default Docker image mapping to the current version of Python is used.
|
955
|
+
queue : str, default METAFLOW_BATCH_JOB_QUEUE
|
956
|
+
AWS Batch Job Queue to submit the job to.
|
957
|
+
iam_role : str, default METAFLOW_ECS_S3_ACCESS_IAM_ROLE
|
958
|
+
AWS IAM role that AWS Batch container uses to access AWS cloud resources.
|
959
|
+
execution_role : str, default METAFLOW_ECS_FARGATE_EXECUTION_ROLE
|
960
|
+
AWS IAM role that AWS Batch can use [to trigger AWS Fargate tasks]
|
961
|
+
(https://docs.aws.amazon.com/batch/latest/userguide/execution-IAM-role.html).
|
797
962
|
shared_memory : int, optional, default None
|
798
963
|
The value for the size (in MiB) of the /dev/shm volume for this step.
|
799
964
|
This parameter maps to the `--shm-size` option in Docker.
|
965
|
+
max_swap : int, optional, default None
|
966
|
+
The total amount of swap memory (in MiB) a container can use for this
|
967
|
+
step. This parameter is translated to the `--memory-swap` option in
|
968
|
+
Docker where the value is the sum of the container memory plus the
|
969
|
+
`max_swap` value.
|
970
|
+
swappiness : int, optional, default None
|
971
|
+
This allows you to tune memory swappiness behavior for this step.
|
972
|
+
A swappiness value of 0 causes swapping not to happen unless absolutely
|
973
|
+
necessary. A swappiness value of 100 causes pages to be swapped very
|
974
|
+
aggressively. Accepted values are whole numbers between 0 and 100.
|
975
|
+
use_tmpfs : bool, default False
|
976
|
+
This enables an explicit tmpfs mount for this step. Note that tmpfs is
|
977
|
+
not available on Fargate compute environments
|
978
|
+
tmpfs_tempdir : bool, default True
|
979
|
+
sets METAFLOW_TEMPDIR to tmpfs_path if set for this step.
|
980
|
+
tmpfs_size : int, optional, default None
|
981
|
+
The value for the size (in MiB) of the tmpfs mount for this step.
|
982
|
+
This parameter maps to the `--tmpfs` option in Docker. Defaults to 50% of the
|
983
|
+
memory allocated for this step.
|
984
|
+
tmpfs_path : str, optional, default None
|
985
|
+
Path to tmpfs mount for this step. Defaults to /metaflow_temp.
|
986
|
+
inferentia : int, default 0
|
987
|
+
Number of Inferentia chips required for this step.
|
988
|
+
trainium : int, default None
|
989
|
+
Alias for inferentia. Use only one of the two.
|
990
|
+
efa : int, default 0
|
991
|
+
Number of elastic fabric adapter network devices to attach to container
|
992
|
+
ephemeral_storage : int, default None
|
993
|
+
The total amount, in GiB, of ephemeral storage to set for the task, 21-200GiB.
|
994
|
+
This is only relevant for Fargate compute environments
|
995
|
+
log_driver: str, optional, default None
|
996
|
+
The log driver to use for the Amazon ECS container.
|
997
|
+
log_options: List[str], optional, default None
|
998
|
+
List of strings containing options for the chosen log driver. The configurable values
|
999
|
+
depend on the `log driver` chosen. Validation of these options is not supported yet.
|
1000
|
+
Example: [`awslogs-group:aws/batch/job`]
|
800
1001
|
"""
|
801
1002
|
...
|
802
1003
|
|
803
1004
|
@typing.overload
|
804
|
-
def
|
1005
|
+
def timeout(*, seconds: int = 0, minutes: int = 0, hours: int = 0) -> typing.Callable[[typing.Union[typing.Callable[[FlowSpecDerived, StepFlag], None], typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None]]], typing.Union[typing.Callable[[FlowSpecDerived, StepFlag], None], typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None]]]:
|
805
1006
|
"""
|
806
|
-
Specifies
|
1007
|
+
Specifies a timeout for your step.
|
807
1008
|
|
808
|
-
|
809
|
-
|
810
|
-
|
811
|
-
|
1009
|
+
This decorator is useful if this step may hang indefinitely.
|
1010
|
+
|
1011
|
+
This can be used in conjunction with the `@retry` decorator as well as the `@catch` decorator.
|
1012
|
+
A timeout is considered to be an exception thrown by the step. It will cause the step to be
|
1013
|
+
retried if needed and the exception will be caught by the `@catch` decorator, if present.
|
1014
|
+
|
1015
|
+
Note that all the values specified in parameters are added together so if you specify
|
1016
|
+
60 seconds and 1 hour, the decorator will have an effective timeout of 1 hour and 1 minute.
|
812
1017
|
|
813
1018
|
Parameters
|
814
1019
|
----------
|
815
|
-
|
816
|
-
|
817
|
-
|
818
|
-
|
819
|
-
|
820
|
-
|
1020
|
+
seconds : int, default 0
|
1021
|
+
Number of seconds to wait prior to timing out.
|
1022
|
+
minutes : int, default 0
|
1023
|
+
Number of minutes to wait prior to timing out.
|
1024
|
+
hours : int, default 0
|
1025
|
+
Number of hours to wait prior to timing out.
|
821
1026
|
"""
|
822
1027
|
...
|
823
1028
|
|
824
1029
|
@typing.overload
|
825
|
-
def
|
1030
|
+
def timeout(f: typing.Callable[[FlowSpecDerived, StepFlag], None]) -> typing.Callable[[FlowSpecDerived, StepFlag], None]:
|
826
1031
|
...
|
827
1032
|
|
828
1033
|
@typing.overload
|
829
|
-
def
|
1034
|
+
def timeout(f: typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None]) -> typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None]:
|
830
1035
|
...
|
831
1036
|
|
832
|
-
def
|
1037
|
+
def timeout(f: typing.Union[typing.Callable[[FlowSpecDerived, StepFlag], None], typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None], None] = None, *, seconds: int = 0, minutes: int = 0, hours: int = 0):
|
833
1038
|
"""
|
834
|
-
Specifies
|
1039
|
+
Specifies a timeout for your step.
|
835
1040
|
|
836
|
-
|
837
|
-
contains the exception raised. You can use it to detect the presence
|
838
|
-
of errors, indicating that all happy-path artifacts produced by the step
|
839
|
-
are missing.
|
1041
|
+
This decorator is useful if this step may hang indefinitely.
|
840
1042
|
|
841
|
-
|
1043
|
+
This can be used in conjunction with the `@retry` decorator as well as the `@catch` decorator.
|
1044
|
+
A timeout is considered to be an exception thrown by the step. It will cause the step to be
|
1045
|
+
retried if needed and the exception will be caught by the `@catch` decorator, if present.
|
1046
|
+
|
1047
|
+
Note that all the values specified in parameters are added together so if you specify
|
1048
|
+
60 seconds and 1 hour, the decorator will have an effective timeout of 1 hour and 1 minute.
|
1049
|
+
|
1050
|
+
Parameters
|
842
1051
|
----------
|
843
|
-
|
844
|
-
|
845
|
-
|
846
|
-
|
847
|
-
|
848
|
-
|
1052
|
+
seconds : int, default 0
|
1053
|
+
Number of seconds to wait prior to timing out.
|
1054
|
+
minutes : int, default 0
|
1055
|
+
Number of minutes to wait prior to timing out.
|
1056
|
+
hours : int, default 0
|
1057
|
+
Number of hours to wait prior to timing out.
|
849
1058
|
"""
|
850
1059
|
...
|
851
1060
|
|
852
1061
|
@typing.overload
|
853
|
-
def
|
1062
|
+
def pypi(*, packages: typing.Dict[str, str] = {}, python: typing.Optional[str] = None) -> typing.Callable[[typing.Union[typing.Callable[[FlowSpecDerived, StepFlag], None], typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None]]], typing.Union[typing.Callable[[FlowSpecDerived, StepFlag], None], typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None]]]:
|
854
1063
|
"""
|
855
|
-
Specifies the
|
856
|
-
to a step needs to be retried.
|
857
|
-
|
858
|
-
This decorator is useful for handling transient errors, such as networking issues.
|
859
|
-
If your task contains operations that can't be retried safely, e.g. database updates,
|
860
|
-
it is advisable to annotate it with `@retry(times=0)`.
|
1064
|
+
Specifies the PyPI packages for the step.
|
861
1065
|
|
862
|
-
|
863
|
-
|
864
|
-
|
1066
|
+
Information in this decorator will augment any
|
1067
|
+
attributes set in the `@pyi_base` flow-level decorator. Hence,
|
1068
|
+
you can use `@pypi_base` to set packages required by all
|
1069
|
+
steps and use `@pypi` to specify step-specific overrides.
|
865
1070
|
|
866
1071
|
Parameters
|
867
1072
|
----------
|
868
|
-
|
869
|
-
|
870
|
-
|
871
|
-
|
1073
|
+
packages : Dict[str, str], default: {}
|
1074
|
+
Packages to use for this step. The key is the name of the package
|
1075
|
+
and the value is the version to use.
|
1076
|
+
python : str, optional, default: None
|
1077
|
+
Version of Python to use, e.g. '3.7.4'. A default value of None implies
|
1078
|
+
that the version used will correspond to the version of the Python interpreter used to start the run.
|
872
1079
|
"""
|
873
1080
|
...
|
874
1081
|
|
875
1082
|
@typing.overload
|
876
|
-
def
|
1083
|
+
def pypi(f: typing.Callable[[FlowSpecDerived, StepFlag], None]) -> typing.Callable[[FlowSpecDerived, StepFlag], None]:
|
877
1084
|
...
|
878
1085
|
|
879
1086
|
@typing.overload
|
880
|
-
def
|
1087
|
+
def pypi(f: typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None]) -> typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None]:
|
881
1088
|
...
|
882
1089
|
|
883
|
-
def
|
1090
|
+
def pypi(f: typing.Union[typing.Callable[[FlowSpecDerived, StepFlag], None], typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None], None] = None, *, packages: typing.Dict[str, str] = {}, python: typing.Optional[str] = None):
|
884
1091
|
"""
|
885
|
-
Specifies the
|
886
|
-
to a step needs to be retried.
|
887
|
-
|
888
|
-
This decorator is useful for handling transient errors, such as networking issues.
|
889
|
-
If your task contains operations that can't be retried safely, e.g. database updates,
|
890
|
-
it is advisable to annotate it with `@retry(times=0)`.
|
1092
|
+
Specifies the PyPI packages for the step.
|
891
1093
|
|
892
|
-
|
893
|
-
|
894
|
-
|
1094
|
+
Information in this decorator will augment any
|
1095
|
+
attributes set in the `@pyi_base` flow-level decorator. Hence,
|
1096
|
+
you can use `@pypi_base` to set packages required by all
|
1097
|
+
steps and use `@pypi` to specify step-specific overrides.
|
895
1098
|
|
896
1099
|
Parameters
|
897
1100
|
----------
|
898
|
-
|
899
|
-
|
900
|
-
|
901
|
-
|
1101
|
+
packages : Dict[str, str], default: {}
|
1102
|
+
Packages to use for this step. The key is the name of the package
|
1103
|
+
and the value is the version to use.
|
1104
|
+
python : str, optional, default: None
|
1105
|
+
Version of Python to use, e.g. '3.7.4'. A default value of None implies
|
1106
|
+
that the version used will correspond to the version of the Python interpreter used to start the run.
|
902
1107
|
"""
|
903
1108
|
...
|
904
1109
|
|
905
1110
|
@typing.overload
|
906
|
-
def
|
1111
|
+
def environment(*, vars: typing.Dict[str, str] = {}) -> typing.Callable[[typing.Union[typing.Callable[[FlowSpecDerived, StepFlag], None], typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None]]], typing.Union[typing.Callable[[FlowSpecDerived, StepFlag], None], typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None]]]:
|
907
1112
|
"""
|
908
|
-
Specifies
|
909
|
-
the execution of a step.
|
1113
|
+
Specifies environment variables to be set prior to the execution of a step.
|
910
1114
|
|
911
1115
|
Parameters
|
912
1116
|
----------
|
913
|
-
|
914
|
-
|
1117
|
+
vars : Dict[str, str], default {}
|
1118
|
+
Dictionary of environment variables to set.
|
915
1119
|
"""
|
916
1120
|
...
|
917
1121
|
|
918
1122
|
@typing.overload
|
919
|
-
def
|
1123
|
+
def environment(f: typing.Callable[[FlowSpecDerived, StepFlag], None]) -> typing.Callable[[FlowSpecDerived, StepFlag], None]:
|
920
1124
|
...
|
921
1125
|
|
922
1126
|
@typing.overload
|
923
|
-
def
|
1127
|
+
def environment(f: typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None]) -> typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None]:
|
924
1128
|
...
|
925
1129
|
|
926
|
-
def
|
1130
|
+
def environment(f: typing.Union[typing.Callable[[FlowSpecDerived, StepFlag], None], typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None], None] = None, *, vars: typing.Dict[str, str] = {}):
|
927
1131
|
"""
|
928
|
-
Specifies
|
929
|
-
the execution of a step.
|
1132
|
+
Specifies environment variables to be set prior to the execution of a step.
|
930
1133
|
|
931
1134
|
Parameters
|
932
1135
|
----------
|
933
|
-
|
934
|
-
|
1136
|
+
vars : Dict[str, str], default {}
|
1137
|
+
Dictionary of environment variables to set.
|
935
1138
|
"""
|
936
1139
|
...
|
937
1140
|
|
938
1141
|
@typing.overload
|
939
|
-
def
|
1142
|
+
def resources(*, cpu: int = 1, gpu: int = 0, disk: typing.Optional[int] = None, memory: int = 4096, shared_memory: typing.Optional[int] = None) -> typing.Callable[[typing.Union[typing.Callable[[FlowSpecDerived, StepFlag], None], typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None]]], typing.Union[typing.Callable[[FlowSpecDerived, StepFlag], None], typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None]]]:
|
940
1143
|
"""
|
941
|
-
Specifies the
|
1144
|
+
Specifies the resources needed when executing this step.
|
942
1145
|
|
943
|
-
|
944
|
-
|
945
|
-
|
946
|
-
|
1146
|
+
Use `@resources` to specify the resource requirements
|
1147
|
+
independently of the specific compute layer (`@batch`, `@kubernetes`).
|
1148
|
+
|
1149
|
+
You can choose the compute layer on the command line by executing e.g.
|
1150
|
+
```
|
1151
|
+
python myflow.py run --with batch
|
1152
|
+
```
|
1153
|
+
or
|
1154
|
+
```
|
1155
|
+
python myflow.py run --with kubernetes
|
1156
|
+
```
|
1157
|
+
which executes the flow on the desired system using the
|
1158
|
+
requirements specified in `@resources`.
|
947
1159
|
|
948
1160
|
Parameters
|
949
1161
|
----------
|
950
|
-
|
951
|
-
|
952
|
-
|
953
|
-
|
954
|
-
|
955
|
-
|
1162
|
+
cpu : int, default 1
|
1163
|
+
Number of CPUs required for this step.
|
1164
|
+
gpu : int, default 0
|
1165
|
+
Number of GPUs required for this step.
|
1166
|
+
disk : int, optional, default None
|
1167
|
+
Disk size (in MB) required for this step. Only applies on Kubernetes.
|
1168
|
+
memory : int, default 4096
|
1169
|
+
Memory size (in MB) required for this step.
|
1170
|
+
shared_memory : int, optional, default None
|
1171
|
+
The value for the size (in MiB) of the /dev/shm volume for this step.
|
1172
|
+
This parameter maps to the `--shm-size` option in Docker.
|
956
1173
|
"""
|
957
1174
|
...
|
958
1175
|
|
959
1176
|
@typing.overload
|
960
|
-
def
|
1177
|
+
def resources(f: typing.Callable[[FlowSpecDerived, StepFlag], None]) -> typing.Callable[[FlowSpecDerived, StepFlag], None]:
|
961
1178
|
...
|
962
1179
|
|
963
1180
|
@typing.overload
|
964
|
-
def
|
1181
|
+
def resources(f: typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None]) -> typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None]:
|
965
1182
|
...
|
966
1183
|
|
967
|
-
def
|
1184
|
+
def resources(f: typing.Union[typing.Callable[[FlowSpecDerived, StepFlag], None], typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None], None] = None, *, cpu: int = 1, gpu: int = 0, disk: typing.Optional[int] = None, memory: int = 4096, shared_memory: typing.Optional[int] = None):
|
968
1185
|
"""
|
969
|
-
Specifies the
|
1186
|
+
Specifies the resources needed when executing this step.
|
970
1187
|
|
971
|
-
|
972
|
-
|
973
|
-
|
974
|
-
|
1188
|
+
Use `@resources` to specify the resource requirements
|
1189
|
+
independently of the specific compute layer (`@batch`, `@kubernetes`).
|
1190
|
+
|
1191
|
+
You can choose the compute layer on the command line by executing e.g.
|
1192
|
+
```
|
1193
|
+
python myflow.py run --with batch
|
1194
|
+
```
|
1195
|
+
or
|
1196
|
+
```
|
1197
|
+
python myflow.py run --with kubernetes
|
1198
|
+
```
|
1199
|
+
which executes the flow on the desired system using the
|
1200
|
+
requirements specified in `@resources`.
|
975
1201
|
|
976
1202
|
Parameters
|
977
1203
|
----------
|
978
|
-
|
979
|
-
|
980
|
-
|
981
|
-
|
982
|
-
|
983
|
-
|
1204
|
+
cpu : int, default 1
|
1205
|
+
Number of CPUs required for this step.
|
1206
|
+
gpu : int, default 0
|
1207
|
+
Number of GPUs required for this step.
|
1208
|
+
disk : int, optional, default None
|
1209
|
+
Disk size (in MB) required for this step. Only applies on Kubernetes.
|
1210
|
+
memory : int, default 4096
|
1211
|
+
Memory size (in MB) required for this step.
|
1212
|
+
shared_memory : int, optional, default None
|
1213
|
+
The value for the size (in MiB) of the /dev/shm volume for this step.
|
1214
|
+
This parameter maps to the `--shm-size` option in Docker.
|
1215
|
+
"""
|
1216
|
+
...
|
1217
|
+
|
1218
|
+
def kubernetes(*, cpu: int = 1, memory: int = 4096, disk: int = 10240, image: typing.Optional[str] = None, image_pull_policy: str = "KUBERNETES_IMAGE_PULL_POLICY", service_account: str = "METAFLOW_KUBERNETES_SERVICE_ACCOUNT", secrets: typing.Optional[typing.List[str]] = None, node_selector: typing.Union[typing.Dict[str, str], str, None] = None, namespace: str = "METAFLOW_KUBERNETES_NAMESPACE", gpu: typing.Optional[int] = None, gpu_vendor: str = "KUBERNETES_GPU_VENDOR", tolerations: typing.List[str] = [], use_tmpfs: bool = False, tmpfs_tempdir: bool = True, tmpfs_size: typing.Optional[int] = None, tmpfs_path: typing.Optional[str] = "/metaflow_temp", persistent_volume_claims: typing.Optional[typing.Dict[str, str]] = None, shared_memory: typing.Optional[int] = None, port: typing.Optional[int] = None, compute_pool: typing.Optional[str] = None) -> typing.Callable[[typing.Union[typing.Callable[[FlowSpecDerived, StepFlag], None], typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None]]], typing.Union[typing.Callable[[FlowSpecDerived, StepFlag], None], typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None]]]:
|
1219
|
+
"""
|
1220
|
+
Specifies that this step should execute on Kubernetes.
|
1221
|
+
|
1222
|
+
Parameters
|
1223
|
+
----------
|
1224
|
+
cpu : int, default 1
|
1225
|
+
Number of CPUs required for this step. If `@resources` is
|
1226
|
+
also present, the maximum value from all decorators is used.
|
1227
|
+
memory : int, default 4096
|
1228
|
+
Memory size (in MB) required for this step. If
|
1229
|
+
`@resources` is also present, the maximum value from all decorators is
|
1230
|
+
used.
|
1231
|
+
disk : int, default 10240
|
1232
|
+
Disk size (in MB) required for this step. If
|
1233
|
+
`@resources` is also present, the maximum value from all decorators is
|
1234
|
+
used.
|
1235
|
+
image : str, optional, default None
|
1236
|
+
Docker image to use when launching on Kubernetes. If not specified, and
|
1237
|
+
METAFLOW_KUBERNETES_CONTAINER_IMAGE is specified, that image is used. If
|
1238
|
+
not, a default Docker image mapping to the current version of Python is used.
|
1239
|
+
image_pull_policy: str, default KUBERNETES_IMAGE_PULL_POLICY
|
1240
|
+
If given, the imagePullPolicy to be applied to the Docker image of the step.
|
1241
|
+
service_account : str, default METAFLOW_KUBERNETES_SERVICE_ACCOUNT
|
1242
|
+
Kubernetes service account to use when launching pod in Kubernetes.
|
1243
|
+
secrets : List[str], optional, default None
|
1244
|
+
Kubernetes secrets to use when launching pod in Kubernetes. These
|
1245
|
+
secrets are in addition to the ones defined in `METAFLOW_KUBERNETES_SECRETS`
|
1246
|
+
in Metaflow configuration.
|
1247
|
+
node_selector: Union[Dict[str,str], str], optional, default None
|
1248
|
+
Kubernetes node selector(s) to apply to the pod running the task.
|
1249
|
+
Can be passed in as a comma separated string of values e.g. "kubernetes.io/os=linux,kubernetes.io/arch=amd64"
|
1250
|
+
or as a dictionary {"kubernetes.io/os": "linux", "kubernetes.io/arch": "amd64"}
|
1251
|
+
namespace : str, default METAFLOW_KUBERNETES_NAMESPACE
|
1252
|
+
Kubernetes namespace to use when launching pod in Kubernetes.
|
1253
|
+
gpu : int, optional, default None
|
1254
|
+
Number of GPUs required for this step. A value of zero implies that
|
1255
|
+
the scheduled node should not have GPUs.
|
1256
|
+
gpu_vendor : str, default KUBERNETES_GPU_VENDOR
|
1257
|
+
The vendor of the GPUs to be used for this step.
|
1258
|
+
tolerations : List[str], default []
|
1259
|
+
The default is extracted from METAFLOW_KUBERNETES_TOLERATIONS.
|
1260
|
+
Kubernetes tolerations to use when launching pod in Kubernetes.
|
1261
|
+
use_tmpfs : bool, default False
|
1262
|
+
This enables an explicit tmpfs mount for this step.
|
1263
|
+
tmpfs_tempdir : bool, default True
|
1264
|
+
sets METAFLOW_TEMPDIR to tmpfs_path if set for this step.
|
1265
|
+
tmpfs_size : int, optional, default: None
|
1266
|
+
The value for the size (in MiB) of the tmpfs mount for this step.
|
1267
|
+
This parameter maps to the `--tmpfs` option in Docker. Defaults to 50% of the
|
1268
|
+
memory allocated for this step.
|
1269
|
+
tmpfs_path : str, optional, default /metaflow_temp
|
1270
|
+
Path to tmpfs mount for this step.
|
1271
|
+
persistent_volume_claims : Dict[str, str], optional, default None
|
1272
|
+
A map (dictionary) of persistent volumes to be mounted to the pod for this step. The map is from persistent
|
1273
|
+
volumes to the path to which the volume is to be mounted, e.g., `{'pvc-name': '/path/to/mount/on'}`.
|
1274
|
+
shared_memory: int, optional
|
1275
|
+
Shared memory size (in MiB) required for this step
|
1276
|
+
port: int, optional
|
1277
|
+
Port number to specify in the Kubernetes job object
|
1278
|
+
compute_pool : str, optional, default None
|
1279
|
+
Compute pool to be used for for this step.
|
1280
|
+
If not specified, any accessible compute pool within the perimeter is used.
|
984
1281
|
"""
|
985
1282
|
...
|
986
1283
|
|
@@ -1042,237 +1339,123 @@ def conda(f: typing.Union[typing.Callable[[FlowSpecDerived, StepFlag], None], ty
|
|
1042
1339
|
...
|
1043
1340
|
|
1044
1341
|
@typing.overload
|
1045
|
-
def
|
1342
|
+
def parallel(f: typing.Callable[[FlowSpecDerived, StepFlag], None]) -> typing.Callable[[FlowSpecDerived, StepFlag], None]:
|
1046
1343
|
"""
|
1047
|
-
|
1048
|
-
|
1049
|
-
Parameters
|
1050
|
-
----------
|
1051
|
-
cpu : int, default 1
|
1052
|
-
Number of CPUs required for this step. If `@resources` is
|
1053
|
-
also present, the maximum value from all decorators is used.
|
1054
|
-
gpu : int, default 0
|
1055
|
-
Number of GPUs required for this step. If `@resources` is
|
1056
|
-
also present, the maximum value from all decorators is used.
|
1057
|
-
memory : int, default 4096
|
1058
|
-
Memory size (in MB) required for this step. If
|
1059
|
-
`@resources` is also present, the maximum value from all decorators is
|
1060
|
-
used.
|
1061
|
-
image : str, optional, default None
|
1062
|
-
Docker image to use when launching on AWS Batch. If not specified, and
|
1063
|
-
METAFLOW_BATCH_CONTAINER_IMAGE is specified, that image is used. If
|
1064
|
-
not, a default Docker image mapping to the current version of Python is used.
|
1065
|
-
queue : str, default METAFLOW_BATCH_JOB_QUEUE
|
1066
|
-
AWS Batch Job Queue to submit the job to.
|
1067
|
-
iam_role : str, default METAFLOW_ECS_S3_ACCESS_IAM_ROLE
|
1068
|
-
AWS IAM role that AWS Batch container uses to access AWS cloud resources.
|
1069
|
-
execution_role : str, default METAFLOW_ECS_FARGATE_EXECUTION_ROLE
|
1070
|
-
AWS IAM role that AWS Batch can use [to trigger AWS Fargate tasks]
|
1071
|
-
(https://docs.aws.amazon.com/batch/latest/userguide/execution-IAM-role.html).
|
1072
|
-
shared_memory : int, optional, default None
|
1073
|
-
The value for the size (in MiB) of the /dev/shm volume for this step.
|
1074
|
-
This parameter maps to the `--shm-size` option in Docker.
|
1075
|
-
max_swap : int, optional, default None
|
1076
|
-
The total amount of swap memory (in MiB) a container can use for this
|
1077
|
-
step. This parameter is translated to the `--memory-swap` option in
|
1078
|
-
Docker where the value is the sum of the container memory plus the
|
1079
|
-
`max_swap` value.
|
1080
|
-
swappiness : int, optional, default None
|
1081
|
-
This allows you to tune memory swappiness behavior for this step.
|
1082
|
-
A swappiness value of 0 causes swapping not to happen unless absolutely
|
1083
|
-
necessary. A swappiness value of 100 causes pages to be swapped very
|
1084
|
-
aggressively. Accepted values are whole numbers between 0 and 100.
|
1085
|
-
use_tmpfs : bool, default False
|
1086
|
-
This enables an explicit tmpfs mount for this step. Note that tmpfs is
|
1087
|
-
not available on Fargate compute environments
|
1088
|
-
tmpfs_tempdir : bool, default True
|
1089
|
-
sets METAFLOW_TEMPDIR to tmpfs_path if set for this step.
|
1090
|
-
tmpfs_size : int, optional, default None
|
1091
|
-
The value for the size (in MiB) of the tmpfs mount for this step.
|
1092
|
-
This parameter maps to the `--tmpfs` option in Docker. Defaults to 50% of the
|
1093
|
-
memory allocated for this step.
|
1094
|
-
tmpfs_path : str, optional, default None
|
1095
|
-
Path to tmpfs mount for this step. Defaults to /metaflow_temp.
|
1096
|
-
inferentia : int, default 0
|
1097
|
-
Number of Inferentia chips required for this step.
|
1098
|
-
trainium : int, default None
|
1099
|
-
Alias for inferentia. Use only one of the two.
|
1100
|
-
efa : int, default 0
|
1101
|
-
Number of elastic fabric adapter network devices to attach to container
|
1102
|
-
ephemeral_storage : int, default None
|
1103
|
-
The total amount, in GiB, of ephemeral storage to set for the task, 21-200GiB.
|
1104
|
-
This is only relevant for Fargate compute environments
|
1105
|
-
log_driver: str, optional, default None
|
1106
|
-
The log driver to use for the Amazon ECS container.
|
1107
|
-
log_options: List[str], optional, default None
|
1108
|
-
List of strings containing options for the chosen log driver. The configurable values
|
1109
|
-
depend on the `log driver` chosen. Validation of these options is not supported yet.
|
1110
|
-
Example: [`awslogs-group:aws/batch/job`]
|
1344
|
+
Decorator prototype for all step decorators. This function gets specialized
|
1345
|
+
and imported for all decorators types by _import_plugin_decorators().
|
1111
1346
|
"""
|
1112
1347
|
...
|
1113
1348
|
|
1114
1349
|
@typing.overload
|
1115
|
-
def
|
1116
|
-
...
|
1117
|
-
|
1118
|
-
@typing.overload
|
1119
|
-
def batch(f: typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None]) -> typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None]:
|
1350
|
+
def parallel(f: typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None]) -> typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None]:
|
1120
1351
|
...
|
1121
1352
|
|
1122
|
-
def
|
1353
|
+
def parallel(f: typing.Union[typing.Callable[[FlowSpecDerived, StepFlag], None], typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None], None] = None):
|
1123
1354
|
"""
|
1124
|
-
|
1125
|
-
|
1126
|
-
Parameters
|
1127
|
-
----------
|
1128
|
-
cpu : int, default 1
|
1129
|
-
Number of CPUs required for this step. If `@resources` is
|
1130
|
-
also present, the maximum value from all decorators is used.
|
1131
|
-
gpu : int, default 0
|
1132
|
-
Number of GPUs required for this step. If `@resources` is
|
1133
|
-
also present, the maximum value from all decorators is used.
|
1134
|
-
memory : int, default 4096
|
1135
|
-
Memory size (in MB) required for this step. If
|
1136
|
-
`@resources` is also present, the maximum value from all decorators is
|
1137
|
-
used.
|
1138
|
-
image : str, optional, default None
|
1139
|
-
Docker image to use when launching on AWS Batch. If not specified, and
|
1140
|
-
METAFLOW_BATCH_CONTAINER_IMAGE is specified, that image is used. If
|
1141
|
-
not, a default Docker image mapping to the current version of Python is used.
|
1142
|
-
queue : str, default METAFLOW_BATCH_JOB_QUEUE
|
1143
|
-
AWS Batch Job Queue to submit the job to.
|
1144
|
-
iam_role : str, default METAFLOW_ECS_S3_ACCESS_IAM_ROLE
|
1145
|
-
AWS IAM role that AWS Batch container uses to access AWS cloud resources.
|
1146
|
-
execution_role : str, default METAFLOW_ECS_FARGATE_EXECUTION_ROLE
|
1147
|
-
AWS IAM role that AWS Batch can use [to trigger AWS Fargate tasks]
|
1148
|
-
(https://docs.aws.amazon.com/batch/latest/userguide/execution-IAM-role.html).
|
1149
|
-
shared_memory : int, optional, default None
|
1150
|
-
The value for the size (in MiB) of the /dev/shm volume for this step.
|
1151
|
-
This parameter maps to the `--shm-size` option in Docker.
|
1152
|
-
max_swap : int, optional, default None
|
1153
|
-
The total amount of swap memory (in MiB) a container can use for this
|
1154
|
-
step. This parameter is translated to the `--memory-swap` option in
|
1155
|
-
Docker where the value is the sum of the container memory plus the
|
1156
|
-
`max_swap` value.
|
1157
|
-
swappiness : int, optional, default None
|
1158
|
-
This allows you to tune memory swappiness behavior for this step.
|
1159
|
-
A swappiness value of 0 causes swapping not to happen unless absolutely
|
1160
|
-
necessary. A swappiness value of 100 causes pages to be swapped very
|
1161
|
-
aggressively. Accepted values are whole numbers between 0 and 100.
|
1162
|
-
use_tmpfs : bool, default False
|
1163
|
-
This enables an explicit tmpfs mount for this step. Note that tmpfs is
|
1164
|
-
not available on Fargate compute environments
|
1165
|
-
tmpfs_tempdir : bool, default True
|
1166
|
-
sets METAFLOW_TEMPDIR to tmpfs_path if set for this step.
|
1167
|
-
tmpfs_size : int, optional, default None
|
1168
|
-
The value for the size (in MiB) of the tmpfs mount for this step.
|
1169
|
-
This parameter maps to the `--tmpfs` option in Docker. Defaults to 50% of the
|
1170
|
-
memory allocated for this step.
|
1171
|
-
tmpfs_path : str, optional, default None
|
1172
|
-
Path to tmpfs mount for this step. Defaults to /metaflow_temp.
|
1173
|
-
inferentia : int, default 0
|
1174
|
-
Number of Inferentia chips required for this step.
|
1175
|
-
trainium : int, default None
|
1176
|
-
Alias for inferentia. Use only one of the two.
|
1177
|
-
efa : int, default 0
|
1178
|
-
Number of elastic fabric adapter network devices to attach to container
|
1179
|
-
ephemeral_storage : int, default None
|
1180
|
-
The total amount, in GiB, of ephemeral storage to set for the task, 21-200GiB.
|
1181
|
-
This is only relevant for Fargate compute environments
|
1182
|
-
log_driver: str, optional, default None
|
1183
|
-
The log driver to use for the Amazon ECS container.
|
1184
|
-
log_options: List[str], optional, default None
|
1185
|
-
List of strings containing options for the chosen log driver. The configurable values
|
1186
|
-
depend on the `log driver` chosen. Validation of these options is not supported yet.
|
1187
|
-
Example: [`awslogs-group:aws/batch/job`]
|
1355
|
+
Decorator prototype for all step decorators. This function gets specialized
|
1356
|
+
and imported for all decorators types by _import_plugin_decorators().
|
1188
1357
|
"""
|
1189
1358
|
...
|
1190
1359
|
|
1191
1360
|
@typing.overload
|
1192
|
-
def
|
1361
|
+
def retry(*, times: int = 3, minutes_between_retries: int = 2) -> typing.Callable[[typing.Union[typing.Callable[[FlowSpecDerived, StepFlag], None], typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None]]], typing.Union[typing.Callable[[FlowSpecDerived, StepFlag], None], typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None]]]:
|
1193
1362
|
"""
|
1194
|
-
Specifies
|
1363
|
+
Specifies the number of times the task corresponding
|
1364
|
+
to a step needs to be retried.
|
1365
|
+
|
1366
|
+
This decorator is useful for handling transient errors, such as networking issues.
|
1367
|
+
If your task contains operations that can't be retried safely, e.g. database updates,
|
1368
|
+
it is advisable to annotate it with `@retry(times=0)`.
|
1369
|
+
|
1370
|
+
This can be used in conjunction with the `@catch` decorator. The `@catch`
|
1371
|
+
decorator will execute a no-op task after all retries have been exhausted,
|
1372
|
+
ensuring that the flow execution can continue.
|
1195
1373
|
|
1196
1374
|
Parameters
|
1197
1375
|
----------
|
1198
|
-
|
1199
|
-
|
1376
|
+
times : int, default 3
|
1377
|
+
Number of times to retry this task.
|
1378
|
+
minutes_between_retries : int, default 2
|
1379
|
+
Number of minutes between retries.
|
1200
1380
|
"""
|
1201
1381
|
...
|
1202
1382
|
|
1203
1383
|
@typing.overload
|
1204
|
-
def
|
1384
|
+
def retry(f: typing.Callable[[FlowSpecDerived, StepFlag], None]) -> typing.Callable[[FlowSpecDerived, StepFlag], None]:
|
1205
1385
|
...
|
1206
1386
|
|
1207
1387
|
@typing.overload
|
1208
|
-
def
|
1388
|
+
def retry(f: typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None]) -> typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None]:
|
1209
1389
|
...
|
1210
1390
|
|
1211
|
-
def
|
1391
|
+
def retry(f: typing.Union[typing.Callable[[FlowSpecDerived, StepFlag], None], typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None], None] = None, *, times: int = 3, minutes_between_retries: int = 2):
|
1212
1392
|
"""
|
1213
|
-
Specifies
|
1393
|
+
Specifies the number of times the task corresponding
|
1394
|
+
to a step needs to be retried.
|
1395
|
+
|
1396
|
+
This decorator is useful for handling transient errors, such as networking issues.
|
1397
|
+
If your task contains operations that can't be retried safely, e.g. database updates,
|
1398
|
+
it is advisable to annotate it with `@retry(times=0)`.
|
1399
|
+
|
1400
|
+
This can be used in conjunction with the `@catch` decorator. The `@catch`
|
1401
|
+
decorator will execute a no-op task after all retries have been exhausted,
|
1402
|
+
ensuring that the flow execution can continue.
|
1214
1403
|
|
1215
1404
|
Parameters
|
1216
1405
|
----------
|
1217
|
-
|
1218
|
-
|
1406
|
+
times : int, default 3
|
1407
|
+
Number of times to retry this task.
|
1408
|
+
minutes_between_retries : int, default 2
|
1409
|
+
Number of minutes between retries.
|
1219
1410
|
"""
|
1220
1411
|
...
|
1221
1412
|
|
1222
1413
|
@typing.overload
|
1223
|
-
def
|
1414
|
+
def catch(*, var: typing.Optional[str] = None, print_exception: bool = True) -> typing.Callable[[typing.Union[typing.Callable[[FlowSpecDerived, StepFlag], None], typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None]]], typing.Union[typing.Callable[[FlowSpecDerived, StepFlag], None], typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None]]]:
|
1224
1415
|
"""
|
1225
|
-
Specifies
|
1226
|
-
|
1227
|
-
This decorator is useful if this step may hang indefinitely.
|
1228
|
-
|
1229
|
-
This can be used in conjunction with the `@retry` decorator as well as the `@catch` decorator.
|
1230
|
-
A timeout is considered to be an exception thrown by the step. It will cause the step to be
|
1231
|
-
retried if needed and the exception will be caught by the `@catch` decorator, if present.
|
1416
|
+
Specifies that the step will success under all circumstances.
|
1232
1417
|
|
1233
|
-
|
1234
|
-
|
1418
|
+
The decorator will create an optional artifact, specified by `var`, which
|
1419
|
+
contains the exception raised. You can use it to detect the presence
|
1420
|
+
of errors, indicating that all happy-path artifacts produced by the step
|
1421
|
+
are missing.
|
1235
1422
|
|
1236
1423
|
Parameters
|
1237
1424
|
----------
|
1238
|
-
|
1239
|
-
|
1240
|
-
|
1241
|
-
|
1242
|
-
|
1243
|
-
|
1425
|
+
var : str, optional, default None
|
1426
|
+
Name of the artifact in which to store the caught exception.
|
1427
|
+
If not specified, the exception is not stored.
|
1428
|
+
print_exception : bool, default True
|
1429
|
+
Determines whether or not the exception is printed to
|
1430
|
+
stdout when caught.
|
1244
1431
|
"""
|
1245
1432
|
...
|
1246
1433
|
|
1247
1434
|
@typing.overload
|
1248
|
-
def
|
1435
|
+
def catch(f: typing.Callable[[FlowSpecDerived, StepFlag], None]) -> typing.Callable[[FlowSpecDerived, StepFlag], None]:
|
1249
1436
|
...
|
1250
1437
|
|
1251
1438
|
@typing.overload
|
1252
|
-
def
|
1439
|
+
def catch(f: typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None]) -> typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None]:
|
1253
1440
|
...
|
1254
1441
|
|
1255
|
-
def
|
1442
|
+
def catch(f: typing.Union[typing.Callable[[FlowSpecDerived, StepFlag], None], typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None], None] = None, *, var: typing.Optional[str] = None, print_exception: bool = True):
|
1256
1443
|
"""
|
1257
|
-
Specifies
|
1258
|
-
|
1259
|
-
This decorator is useful if this step may hang indefinitely.
|
1260
|
-
|
1261
|
-
This can be used in conjunction with the `@retry` decorator as well as the `@catch` decorator.
|
1262
|
-
A timeout is considered to be an exception thrown by the step. It will cause the step to be
|
1263
|
-
retried if needed and the exception will be caught by the `@catch` decorator, if present.
|
1444
|
+
Specifies that the step will success under all circumstances.
|
1264
1445
|
|
1265
|
-
|
1266
|
-
|
1446
|
+
The decorator will create an optional artifact, specified by `var`, which
|
1447
|
+
contains the exception raised. You can use it to detect the presence
|
1448
|
+
of errors, indicating that all happy-path artifacts produced by the step
|
1449
|
+
are missing.
|
1267
1450
|
|
1268
1451
|
Parameters
|
1269
1452
|
----------
|
1270
|
-
|
1271
|
-
|
1272
|
-
|
1273
|
-
|
1274
|
-
|
1275
|
-
|
1453
|
+
var : str, optional, default None
|
1454
|
+
Name of the artifact in which to store the caught exception.
|
1455
|
+
If not specified, the exception is not stored.
|
1456
|
+
print_exception : bool, default True
|
1457
|
+
Determines whether or not the exception is printed to
|
1458
|
+
stdout when caught.
|
1276
1459
|
"""
|
1277
1460
|
...
|
1278
1461
|
|
@@ -1327,134 +1510,117 @@ def card(f: typing.Union[typing.Callable[[FlowSpecDerived, StepFlag], None], typ
|
|
1327
1510
|
"""
|
1328
1511
|
...
|
1329
1512
|
|
1330
|
-
|
1513
|
+
@typing.overload
|
1514
|
+
def secrets(*, sources: typing.List[typing.Union[str, typing.Dict[str, typing.Any]]] = []) -> typing.Callable[[typing.Union[typing.Callable[[FlowSpecDerived, StepFlag], None], typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None]]], typing.Union[typing.Callable[[FlowSpecDerived, StepFlag], None], typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None]]]:
|
1331
1515
|
"""
|
1332
|
-
Specifies
|
1516
|
+
Specifies secrets to be retrieved and injected as environment variables prior to
|
1517
|
+
the execution of a step.
|
1333
1518
|
|
1334
1519
|
Parameters
|
1335
1520
|
----------
|
1336
|
-
|
1337
|
-
|
1338
|
-
also present, the maximum value from all decorators is used.
|
1339
|
-
memory : int, default 4096
|
1340
|
-
Memory size (in MB) required for this step. If
|
1341
|
-
`@resources` is also present, the maximum value from all decorators is
|
1342
|
-
used.
|
1343
|
-
disk : int, default 10240
|
1344
|
-
Disk size (in MB) required for this step. If
|
1345
|
-
`@resources` is also present, the maximum value from all decorators is
|
1346
|
-
used.
|
1347
|
-
image : str, optional, default None
|
1348
|
-
Docker image to use when launching on Kubernetes. If not specified, and
|
1349
|
-
METAFLOW_KUBERNETES_CONTAINER_IMAGE is specified, that image is used. If
|
1350
|
-
not, a default Docker image mapping to the current version of Python is used.
|
1351
|
-
image_pull_policy: str, default KUBERNETES_IMAGE_PULL_POLICY
|
1352
|
-
If given, the imagePullPolicy to be applied to the Docker image of the step.
|
1353
|
-
service_account : str, default METAFLOW_KUBERNETES_SERVICE_ACCOUNT
|
1354
|
-
Kubernetes service account to use when launching pod in Kubernetes.
|
1355
|
-
secrets : List[str], optional, default None
|
1356
|
-
Kubernetes secrets to use when launching pod in Kubernetes. These
|
1357
|
-
secrets are in addition to the ones defined in `METAFLOW_KUBERNETES_SECRETS`
|
1358
|
-
in Metaflow configuration.
|
1359
|
-
node_selector: Union[Dict[str,str], str], optional, default None
|
1360
|
-
Kubernetes node selector(s) to apply to the pod running the task.
|
1361
|
-
Can be passed in as a comma separated string of values e.g. "kubernetes.io/os=linux,kubernetes.io/arch=amd64"
|
1362
|
-
or as a dictionary {"kubernetes.io/os": "linux", "kubernetes.io/arch": "amd64"}
|
1363
|
-
namespace : str, default METAFLOW_KUBERNETES_NAMESPACE
|
1364
|
-
Kubernetes namespace to use when launching pod in Kubernetes.
|
1365
|
-
gpu : int, optional, default None
|
1366
|
-
Number of GPUs required for this step. A value of zero implies that
|
1367
|
-
the scheduled node should not have GPUs.
|
1368
|
-
gpu_vendor : str, default KUBERNETES_GPU_VENDOR
|
1369
|
-
The vendor of the GPUs to be used for this step.
|
1370
|
-
tolerations : List[str], default []
|
1371
|
-
The default is extracted from METAFLOW_KUBERNETES_TOLERATIONS.
|
1372
|
-
Kubernetes tolerations to use when launching pod in Kubernetes.
|
1373
|
-
use_tmpfs : bool, default False
|
1374
|
-
This enables an explicit tmpfs mount for this step.
|
1375
|
-
tmpfs_tempdir : bool, default True
|
1376
|
-
sets METAFLOW_TEMPDIR to tmpfs_path if set for this step.
|
1377
|
-
tmpfs_size : int, optional, default: None
|
1378
|
-
The value for the size (in MiB) of the tmpfs mount for this step.
|
1379
|
-
This parameter maps to the `--tmpfs` option in Docker. Defaults to 50% of the
|
1380
|
-
memory allocated for this step.
|
1381
|
-
tmpfs_path : str, optional, default /metaflow_temp
|
1382
|
-
Path to tmpfs mount for this step.
|
1383
|
-
persistent_volume_claims : Dict[str, str], optional, default None
|
1384
|
-
A map (dictionary) of persistent volumes to be mounted to the pod for this step. The map is from persistent
|
1385
|
-
volumes to the path to which the volume is to be mounted, e.g., `{'pvc-name': '/path/to/mount/on'}`.
|
1386
|
-
shared_memory: int, optional
|
1387
|
-
Shared memory size (in MiB) required for this step
|
1388
|
-
port: int, optional
|
1389
|
-
Port number to specify in the Kubernetes job object
|
1390
|
-
compute_pool : str, optional, default None
|
1391
|
-
Compute pool to be used for for this step.
|
1392
|
-
If not specified, any accessible compute pool within the perimeter is used.
|
1521
|
+
sources : List[Union[str, Dict[str, Any]]], default: []
|
1522
|
+
List of secret specs, defining how the secrets are to be retrieved
|
1393
1523
|
"""
|
1394
1524
|
...
|
1395
1525
|
|
1396
1526
|
@typing.overload
|
1397
|
-
def
|
1527
|
+
def secrets(f: typing.Callable[[FlowSpecDerived, StepFlag], None]) -> typing.Callable[[FlowSpecDerived, StepFlag], None]:
|
1528
|
+
...
|
1529
|
+
|
1530
|
+
@typing.overload
|
1531
|
+
def secrets(f: typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None]) -> typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None]:
|
1532
|
+
...
|
1533
|
+
|
1534
|
+
def secrets(f: typing.Union[typing.Callable[[FlowSpecDerived, StepFlag], None], typing.Callable[[FlowSpecDerived, typing.Any, StepFlag], None], None] = None, *, sources: typing.List[typing.Union[str, typing.Dict[str, typing.Any]]] = []):
|
1398
1535
|
"""
|
1399
|
-
|
1400
|
-
|
1536
|
+
Specifies secrets to be retrieved and injected as environment variables prior to
|
1537
|
+
the execution of a step.
|
1538
|
+
|
1539
|
+
Parameters
|
1540
|
+
----------
|
1541
|
+
sources : List[Union[str, Dict[str, Any]]], default: []
|
1542
|
+
List of secret specs, defining how the secrets are to be retrieved
|
1401
1543
|
"""
|
1402
1544
|
...
|
1403
1545
|
|
1404
1546
|
@typing.overload
|
1405
|
-
def
|
1547
|
+
def pypi_base(*, packages: typing.Dict[str, str] = {}, python: typing.Optional[str] = None) -> typing.Callable[[typing.Type[FlowSpecDerived]], typing.Type[FlowSpecDerived]]:
|
1548
|
+
"""
|
1549
|
+
Specifies the PyPI packages for all steps of the flow.
|
1550
|
+
|
1551
|
+
Use `@pypi_base` to set common packages required by all
|
1552
|
+
steps and use `@pypi` to specify step-specific overrides.
|
1553
|
+
Parameters
|
1554
|
+
----------
|
1555
|
+
packages : Dict[str, str], default: {}
|
1556
|
+
Packages to use for this flow. The key is the name of the package
|
1557
|
+
and the value is the version to use.
|
1558
|
+
python : str, optional, default: None
|
1559
|
+
Version of Python to use, e.g. '3.7.4'. A default value of None implies
|
1560
|
+
that the version used will correspond to the version of the Python interpreter used to start the run.
|
1561
|
+
"""
|
1406
1562
|
...
|
1407
1563
|
|
1408
|
-
|
1564
|
+
@typing.overload
|
1565
|
+
def pypi_base(f: typing.Type[FlowSpecDerived]) -> typing.Type[FlowSpecDerived]:
|
1566
|
+
...
|
1567
|
+
|
1568
|
+
def pypi_base(f: typing.Optional[typing.Type[FlowSpecDerived]] = None, *, packages: typing.Dict[str, str] = {}, python: typing.Optional[str] = None):
|
1409
1569
|
"""
|
1410
|
-
|
1411
|
-
|
1570
|
+
Specifies the PyPI packages for all steps of the flow.
|
1571
|
+
|
1572
|
+
Use `@pypi_base` to set common packages required by all
|
1573
|
+
steps and use `@pypi` to specify step-specific overrides.
|
1574
|
+
Parameters
|
1575
|
+
----------
|
1576
|
+
packages : Dict[str, str], default: {}
|
1577
|
+
Packages to use for this flow. The key is the name of the package
|
1578
|
+
and the value is the version to use.
|
1579
|
+
python : str, optional, default: None
|
1580
|
+
Version of Python to use, e.g. '3.7.4'. A default value of None implies
|
1581
|
+
that the version used will correspond to the version of the Python interpreter used to start the run.
|
1412
1582
|
"""
|
1413
1583
|
...
|
1414
1584
|
|
1415
1585
|
@typing.overload
|
1416
|
-
def
|
1586
|
+
def trigger(*, event: typing.Union[str, typing.Dict[str, typing.Any], None] = None, events: typing.List[typing.Union[str, typing.Dict[str, typing.Any]]] = [], options: typing.Dict[str, typing.Any] = {}) -> typing.Callable[[typing.Type[FlowSpecDerived]], typing.Type[FlowSpecDerived]]:
|
1417
1587
|
"""
|
1418
|
-
Specifies the
|
1588
|
+
Specifies the event(s) that this flow depends on.
|
1419
1589
|
|
1420
1590
|
```
|
1421
|
-
@
|
1591
|
+
@trigger(event='foo')
|
1422
1592
|
```
|
1423
1593
|
or
|
1424
1594
|
```
|
1425
|
-
@
|
1595
|
+
@trigger(events=['foo', 'bar'])
|
1426
1596
|
```
|
1427
|
-
This decorator respects the @project decorator and triggers the flow
|
1428
|
-
when upstream runs within the same namespace complete successfully
|
1429
1597
|
|
1430
|
-
Additionally, you can specify
|
1431
|
-
|
1598
|
+
Additionally, you can specify the parameter mappings
|
1599
|
+
to map event payload to Metaflow parameters for the flow.
|
1432
1600
|
```
|
1433
|
-
@
|
1601
|
+
@trigger(event={'name':'foo', 'parameters':{'flow_param': 'event_field'}})
|
1434
1602
|
```
|
1435
1603
|
or
|
1436
1604
|
```
|
1437
|
-
@
|
1605
|
+
@trigger(events=[{'name':'foo', 'parameters':{'flow_param_1': 'event_field_1'},
|
1606
|
+
{'name':'bar', 'parameters':{'flow_param_2': 'event_field_2'}])
|
1438
1607
|
```
|
1439
1608
|
|
1440
|
-
|
1441
|
-
inferred from the current project or project branch):
|
1609
|
+
'parameters' can also be a list of strings and tuples like so:
|
1442
1610
|
```
|
1443
|
-
@
|
1611
|
+
@trigger(event={'name':'foo', 'parameters':['common_name', ('flow_param', 'event_field')]})
|
1612
|
+
```
|
1613
|
+
This is equivalent to:
|
1614
|
+
```
|
1615
|
+
@trigger(event={'name':'foo', 'parameters':{'common_name': 'common_name', 'flow_param': 'event_field'}})
|
1444
1616
|
```
|
1445
|
-
|
1446
|
-
Note that `branch` is typically one of:
|
1447
|
-
- `prod`
|
1448
|
-
- `user.bob`
|
1449
|
-
- `test.my_experiment`
|
1450
|
-
- `prod.staging`
|
1451
1617
|
|
1452
1618
|
Parameters
|
1453
1619
|
----------
|
1454
|
-
|
1455
|
-
|
1456
|
-
|
1457
|
-
|
1620
|
+
event : Union[str, Dict[str, Any]], optional, default None
|
1621
|
+
Event dependency for this flow.
|
1622
|
+
events : List[Union[str, Dict[str, Any]]], default []
|
1623
|
+
Events dependency for this flow.
|
1458
1624
|
options : Dict[str, Any], default {}
|
1459
1625
|
Backend-specific configuration for tuning eventing behavior.
|
1460
1626
|
|
@@ -1463,51 +1629,47 @@ def trigger_on_finish(*, flow: typing.Union[typing.Dict[str, str], str, None] =
|
|
1463
1629
|
...
|
1464
1630
|
|
1465
1631
|
@typing.overload
|
1466
|
-
def
|
1632
|
+
def trigger(f: typing.Type[FlowSpecDerived]) -> typing.Type[FlowSpecDerived]:
|
1467
1633
|
...
|
1468
1634
|
|
1469
|
-
def
|
1635
|
+
def trigger(f: typing.Optional[typing.Type[FlowSpecDerived]] = None, *, event: typing.Union[str, typing.Dict[str, typing.Any], None] = None, events: typing.List[typing.Union[str, typing.Dict[str, typing.Any]]] = [], options: typing.Dict[str, typing.Any] = {}):
|
1470
1636
|
"""
|
1471
|
-
Specifies the
|
1637
|
+
Specifies the event(s) that this flow depends on.
|
1472
1638
|
|
1473
1639
|
```
|
1474
|
-
@
|
1640
|
+
@trigger(event='foo')
|
1475
1641
|
```
|
1476
1642
|
or
|
1477
1643
|
```
|
1478
|
-
@
|
1644
|
+
@trigger(events=['foo', 'bar'])
|
1479
1645
|
```
|
1480
|
-
This decorator respects the @project decorator and triggers the flow
|
1481
|
-
when upstream runs within the same namespace complete successfully
|
1482
1646
|
|
1483
|
-
Additionally, you can specify
|
1484
|
-
|
1647
|
+
Additionally, you can specify the parameter mappings
|
1648
|
+
to map event payload to Metaflow parameters for the flow.
|
1485
1649
|
```
|
1486
|
-
@
|
1650
|
+
@trigger(event={'name':'foo', 'parameters':{'flow_param': 'event_field'}})
|
1487
1651
|
```
|
1488
1652
|
or
|
1489
1653
|
```
|
1490
|
-
@
|
1654
|
+
@trigger(events=[{'name':'foo', 'parameters':{'flow_param_1': 'event_field_1'},
|
1655
|
+
{'name':'bar', 'parameters':{'flow_param_2': 'event_field_2'}])
|
1491
1656
|
```
|
1492
1657
|
|
1493
|
-
|
1494
|
-
inferred from the current project or project branch):
|
1658
|
+
'parameters' can also be a list of strings and tuples like so:
|
1495
1659
|
```
|
1496
|
-
@
|
1660
|
+
@trigger(event={'name':'foo', 'parameters':['common_name', ('flow_param', 'event_field')]})
|
1661
|
+
```
|
1662
|
+
This is equivalent to:
|
1663
|
+
```
|
1664
|
+
@trigger(event={'name':'foo', 'parameters':{'common_name': 'common_name', 'flow_param': 'event_field'}})
|
1497
1665
|
```
|
1498
|
-
|
1499
|
-
Note that `branch` is typically one of:
|
1500
|
-
- `prod`
|
1501
|
-
- `user.bob`
|
1502
|
-
- `test.my_experiment`
|
1503
|
-
- `prod.staging`
|
1504
1666
|
|
1505
1667
|
Parameters
|
1506
1668
|
----------
|
1507
|
-
|
1508
|
-
|
1509
|
-
|
1510
|
-
|
1669
|
+
event : Union[str, Dict[str, Any]], optional, default None
|
1670
|
+
Event dependency for this flow.
|
1671
|
+
events : List[Union[str, Dict[str, Any]]], default []
|
1672
|
+
Events dependency for this flow.
|
1511
1673
|
options : Dict[str, Any], default {}
|
1512
1674
|
Backend-specific configuration for tuning eventing behavior.
|
1513
1675
|
|
@@ -1515,49 +1677,13 @@ def trigger_on_finish(f: typing.Optional[typing.Type[FlowSpecDerived]] = None, *
|
|
1515
1677
|
"""
|
1516
1678
|
...
|
1517
1679
|
|
1518
|
-
|
1519
|
-
def pypi_base(*, packages: typing.Dict[str, str] = {}, python: typing.Optional[str] = None) -> typing.Callable[[typing.Type[FlowSpecDerived]], typing.Type[FlowSpecDerived]]:
|
1520
|
-
"""
|
1521
|
-
Specifies the PyPI packages for all steps of the flow.
|
1522
|
-
|
1523
|
-
Use `@pypi_base` to set common packages required by all
|
1524
|
-
steps and use `@pypi` to specify step-specific overrides.
|
1525
|
-
Parameters
|
1526
|
-
----------
|
1527
|
-
packages : Dict[str, str], default: {}
|
1528
|
-
Packages to use for this flow. The key is the name of the package
|
1529
|
-
and the value is the version to use.
|
1530
|
-
python : str, optional, default: None
|
1531
|
-
Version of Python to use, e.g. '3.7.4'. A default value of None implies
|
1532
|
-
that the version used will correspond to the version of the Python interpreter used to start the run.
|
1533
|
-
"""
|
1534
|
-
...
|
1535
|
-
|
1536
|
-
@typing.overload
|
1537
|
-
def pypi_base(f: typing.Type[FlowSpecDerived]) -> typing.Type[FlowSpecDerived]:
|
1538
|
-
...
|
1539
|
-
|
1540
|
-
def pypi_base(f: typing.Optional[typing.Type[FlowSpecDerived]] = None, *, packages: typing.Dict[str, str] = {}, python: typing.Optional[str] = None):
|
1541
|
-
"""
|
1542
|
-
Specifies the PyPI packages for all steps of the flow.
|
1543
|
-
|
1544
|
-
Use `@pypi_base` to set common packages required by all
|
1545
|
-
steps and use `@pypi` to specify step-specific overrides.
|
1546
|
-
Parameters
|
1547
|
-
----------
|
1548
|
-
packages : Dict[str, str], default: {}
|
1549
|
-
Packages to use for this flow. The key is the name of the package
|
1550
|
-
and the value is the version to use.
|
1551
|
-
python : str, optional, default: None
|
1552
|
-
Version of Python to use, e.g. '3.7.4'. A default value of None implies
|
1553
|
-
that the version used will correspond to the version of the Python interpreter used to start the run.
|
1554
|
-
"""
|
1555
|
-
...
|
1556
|
-
|
1557
|
-
def airflow_external_task_sensor(*, timeout: int, poke_interval: int, mode: str, exponential_backoff: bool, pool: str, soft_fail: bool, name: str, description: str, external_dag_id: str, external_task_ids: typing.List[str], allowed_states: typing.List[str], failed_states: typing.List[str], execution_delta: "datetime.timedelta", check_existence: bool) -> typing.Callable[[typing.Type[FlowSpecDerived]], typing.Type[FlowSpecDerived]]:
|
1680
|
+
def airflow_s3_key_sensor(*, timeout: int, poke_interval: int, mode: str, exponential_backoff: bool, pool: str, soft_fail: bool, name: str, description: str, bucket_key: typing.Union[str, typing.List[str]], bucket_name: str, wildcard_match: bool, aws_conn_id: str, verify: bool) -> typing.Callable[[typing.Type[FlowSpecDerived]], typing.Type[FlowSpecDerived]]:
|
1558
1681
|
"""
|
1559
|
-
The `@
|
1560
|
-
This decorator only works when a flow is scheduled on Airflow
|
1682
|
+
The `@airflow_s3_key_sensor` decorator attaches a Airflow [S3KeySensor](https://airflow.apache.org/docs/apache-airflow-providers-amazon/stable/_api/airflow/providers/amazon/aws/sensors/s3/index.html#airflow.providers.amazon.aws.sensors.s3.S3KeySensor)
|
1683
|
+
before the start step of the flow. This decorator only works when a flow is scheduled on Airflow
|
1684
|
+
and is compiled using `airflow create`. More than one `@airflow_s3_key_sensor` can be
|
1685
|
+
added as a flow decorators. Adding more than one decorator will ensure that `start` step
|
1686
|
+
starts only after all sensors finish.
|
1561
1687
|
|
1562
1688
|
Parameters
|
1563
1689
|
----------
|
@@ -1578,80 +1704,92 @@ def airflow_external_task_sensor(*, timeout: int, poke_interval: int, mode: str,
|
|
1578
1704
|
Name of the sensor on Airflow
|
1579
1705
|
description : str
|
1580
1706
|
Description of sensor in the Airflow UI
|
1581
|
-
|
1582
|
-
The
|
1583
|
-
|
1584
|
-
|
1585
|
-
|
1586
|
-
|
1587
|
-
|
1588
|
-
|
1589
|
-
|
1590
|
-
|
1591
|
-
|
1592
|
-
|
1593
|
-
check_existence: bool
|
1594
|
-
Set to True to check if the external task exists or check if
|
1595
|
-
the DAG to wait for exists. (Default: True)
|
1707
|
+
bucket_key : Union[str, List[str]]
|
1708
|
+
The key(s) being waited on. Supports full s3:// style url or relative path from root level.
|
1709
|
+
When it's specified as a full s3:// url, please leave `bucket_name` as None
|
1710
|
+
bucket_name : str
|
1711
|
+
Name of the S3 bucket. Only needed when bucket_key is not provided as a full s3:// url.
|
1712
|
+
When specified, all the keys passed to bucket_key refers to this bucket. (Default:None)
|
1713
|
+
wildcard_match : bool
|
1714
|
+
whether the bucket_key should be interpreted as a Unix wildcard pattern. (Default: False)
|
1715
|
+
aws_conn_id : str
|
1716
|
+
a reference to the s3 connection on Airflow. (Default: None)
|
1717
|
+
verify : bool
|
1718
|
+
Whether or not to verify SSL certificates for S3 connection. (Default: None)
|
1596
1719
|
"""
|
1597
1720
|
...
|
1598
1721
|
|
1599
|
-
|
1600
|
-
def schedule(*, hourly: bool = False, daily: bool = True, weekly: bool = False, cron: typing.Optional[str] = None, timezone: typing.Optional[str] = None) -> typing.Callable[[typing.Type[FlowSpecDerived]], typing.Type[FlowSpecDerived]]:
|
1722
|
+
def project(*, name: str) -> typing.Callable[[typing.Type[FlowSpecDerived]], typing.Type[FlowSpecDerived]]:
|
1601
1723
|
"""
|
1602
|
-
Specifies
|
1603
|
-
|
1724
|
+
Specifies what flows belong to the same project.
|
1725
|
+
|
1726
|
+
A project-specific namespace is created for all flows that
|
1727
|
+
use the same `@project(name)`.
|
1604
1728
|
|
1605
1729
|
Parameters
|
1606
1730
|
----------
|
1607
|
-
|
1608
|
-
|
1609
|
-
|
1610
|
-
|
1611
|
-
|
1612
|
-
|
1613
|
-
cron : str, optional, default None
|
1614
|
-
Run the workflow at [a custom Cron schedule](https://docs.aws.amazon.com/eventbridge/latest/userguide/scheduled-events.html#cron-expressions)
|
1615
|
-
specified by this expression.
|
1616
|
-
timezone : str, optional, default None
|
1617
|
-
Timezone on which the schedule runs (default: None). Currently supported only for Argo workflows,
|
1618
|
-
which accepts timezones in [IANA format](https://nodatime.org/TimeZones).
|
1731
|
+
name : str
|
1732
|
+
Project name. Make sure that the name is unique amongst all
|
1733
|
+
projects that use the same production scheduler. The name may
|
1734
|
+
contain only lowercase alphanumeric characters and underscores.
|
1735
|
+
|
1736
|
+
|
1619
1737
|
"""
|
1620
1738
|
...
|
1621
1739
|
|
1622
1740
|
@typing.overload
|
1623
|
-
def
|
1624
|
-
...
|
1625
|
-
|
1626
|
-
def schedule(f: typing.Optional[typing.Type[FlowSpecDerived]] = None, *, hourly: bool = False, daily: bool = True, weekly: bool = False, cron: typing.Optional[str] = None, timezone: typing.Optional[str] = None):
|
1741
|
+
def conda_base(*, packages: typing.Dict[str, str] = {}, libraries: typing.Dict[str, str] = {}, python: typing.Optional[str] = None, disabled: bool = False) -> typing.Callable[[typing.Type[FlowSpecDerived]], typing.Type[FlowSpecDerived]]:
|
1627
1742
|
"""
|
1628
|
-
Specifies the
|
1629
|
-
|
1743
|
+
Specifies the Conda environment for all steps of the flow.
|
1744
|
+
|
1745
|
+
Use `@conda_base` to set common libraries required by all
|
1746
|
+
steps and use `@conda` to specify step-specific additions.
|
1630
1747
|
|
1631
1748
|
Parameters
|
1632
1749
|
----------
|
1633
|
-
|
1634
|
-
|
1635
|
-
|
1636
|
-
|
1637
|
-
|
1638
|
-
|
1639
|
-
|
1640
|
-
|
1641
|
-
|
1642
|
-
|
1643
|
-
Timezone on which the schedule runs (default: None). Currently supported only for Argo workflows,
|
1644
|
-
which accepts timezones in [IANA format](https://nodatime.org/TimeZones).
|
1750
|
+
packages : Dict[str, str], default {}
|
1751
|
+
Packages to use for this flow. The key is the name of the package
|
1752
|
+
and the value is the version to use.
|
1753
|
+
libraries : Dict[str, str], default {}
|
1754
|
+
Supported for backward compatibility. When used with packages, packages will take precedence.
|
1755
|
+
python : str, optional, default None
|
1756
|
+
Version of Python to use, e.g. '3.7.4'. A default value of None implies
|
1757
|
+
that the version used will correspond to the version of the Python interpreter used to start the run.
|
1758
|
+
disabled : bool, default False
|
1759
|
+
If set to True, disables Conda.
|
1645
1760
|
"""
|
1646
1761
|
...
|
1647
1762
|
|
1648
|
-
|
1763
|
+
@typing.overload
|
1764
|
+
def conda_base(f: typing.Type[FlowSpecDerived]) -> typing.Type[FlowSpecDerived]:
|
1765
|
+
...
|
1766
|
+
|
1767
|
+
def conda_base(f: typing.Optional[typing.Type[FlowSpecDerived]] = None, *, packages: typing.Dict[str, str] = {}, libraries: typing.Dict[str, str] = {}, python: typing.Optional[str] = None, disabled: bool = False):
|
1649
1768
|
"""
|
1650
|
-
|
1651
|
-
|
1652
|
-
|
1653
|
-
|
1654
|
-
|
1769
|
+
Specifies the Conda environment for all steps of the flow.
|
1770
|
+
|
1771
|
+
Use `@conda_base` to set common libraries required by all
|
1772
|
+
steps and use `@conda` to specify step-specific additions.
|
1773
|
+
|
1774
|
+
Parameters
|
1775
|
+
----------
|
1776
|
+
packages : Dict[str, str], default {}
|
1777
|
+
Packages to use for this flow. The key is the name of the package
|
1778
|
+
and the value is the version to use.
|
1779
|
+
libraries : Dict[str, str], default {}
|
1780
|
+
Supported for backward compatibility. When used with packages, packages will take precedence.
|
1781
|
+
python : str, optional, default None
|
1782
|
+
Version of Python to use, e.g. '3.7.4'. A default value of None implies
|
1783
|
+
that the version used will correspond to the version of the Python interpreter used to start the run.
|
1784
|
+
disabled : bool, default False
|
1785
|
+
If set to True, disables Conda.
|
1786
|
+
"""
|
1787
|
+
...
|
1788
|
+
|
1789
|
+
def airflow_external_task_sensor(*, timeout: int, poke_interval: int, mode: str, exponential_backoff: bool, pool: str, soft_fail: bool, name: str, description: str, external_dag_id: str, external_task_ids: typing.List[str], allowed_states: typing.List[str], failed_states: typing.List[str], execution_delta: "datetime.timedelta", check_existence: bool) -> typing.Callable[[typing.Type[FlowSpecDerived]], typing.Type[FlowSpecDerived]]:
|
1790
|
+
"""
|
1791
|
+
The `@airflow_external_task_sensor` decorator attaches a Airflow [ExternalTaskSensor](https://airflow.apache.org/docs/apache-airflow/stable/_api/airflow/sensors/external_task/index.html#airflow.sensors.external_task.ExternalTaskSensor) before the start step of the flow.
|
1792
|
+
This decorator only works when a flow is scheduled on Airflow and is compiled using `airflow create`. More than one `@airflow_external_task_sensor` can be added as a flow decorators. Adding more than one decorator will ensure that `start` step starts only after all sensors finish.
|
1655
1793
|
|
1656
1794
|
Parameters
|
1657
1795
|
----------
|
@@ -1672,60 +1810,67 @@ def airflow_s3_key_sensor(*, timeout: int, poke_interval: int, mode: str, expone
|
|
1672
1810
|
Name of the sensor on Airflow
|
1673
1811
|
description : str
|
1674
1812
|
Description of sensor in the Airflow UI
|
1675
|
-
|
1676
|
-
The
|
1677
|
-
|
1678
|
-
|
1679
|
-
|
1680
|
-
|
1681
|
-
|
1682
|
-
|
1683
|
-
|
1684
|
-
|
1685
|
-
|
1686
|
-
|
1813
|
+
external_dag_id : str
|
1814
|
+
The dag_id that contains the task you want to wait for.
|
1815
|
+
external_task_ids : List[str]
|
1816
|
+
The list of task_ids that you want to wait for.
|
1817
|
+
If None (default value) the sensor waits for the DAG. (Default: None)
|
1818
|
+
allowed_states : List[str]
|
1819
|
+
Iterable of allowed states, (Default: ['success'])
|
1820
|
+
failed_states : List[str]
|
1821
|
+
Iterable of failed or dis-allowed states. (Default: None)
|
1822
|
+
execution_delta : datetime.timedelta
|
1823
|
+
time difference with the previous execution to look at,
|
1824
|
+
the default is the same logical date as the current task or DAG. (Default: None)
|
1825
|
+
check_existence: bool
|
1826
|
+
Set to True to check if the external task exists or check if
|
1827
|
+
the DAG to wait for exists. (Default: True)
|
1687
1828
|
"""
|
1688
1829
|
...
|
1689
1830
|
|
1690
1831
|
@typing.overload
|
1691
|
-
def
|
1832
|
+
def trigger_on_finish(*, flow: typing.Union[typing.Dict[str, str], str, None] = None, flows: typing.List[typing.Union[str, typing.Dict[str, str]]] = [], options: typing.Dict[str, typing.Any] = {}) -> typing.Callable[[typing.Type[FlowSpecDerived]], typing.Type[FlowSpecDerived]]:
|
1692
1833
|
"""
|
1693
|
-
Specifies the
|
1834
|
+
Specifies the flow(s) that this flow depends on.
|
1694
1835
|
|
1695
1836
|
```
|
1696
|
-
@
|
1837
|
+
@trigger_on_finish(flow='FooFlow')
|
1697
1838
|
```
|
1698
1839
|
or
|
1699
1840
|
```
|
1700
|
-
@
|
1841
|
+
@trigger_on_finish(flows=['FooFlow', 'BarFlow'])
|
1701
1842
|
```
|
1843
|
+
This decorator respects the @project decorator and triggers the flow
|
1844
|
+
when upstream runs within the same namespace complete successfully
|
1702
1845
|
|
1703
|
-
Additionally, you can specify
|
1704
|
-
|
1846
|
+
Additionally, you can specify project aware upstream flow dependencies
|
1847
|
+
by specifying the fully qualified project_flow_name.
|
1705
1848
|
```
|
1706
|
-
@
|
1849
|
+
@trigger_on_finish(flow='my_project.branch.my_branch.FooFlow')
|
1707
1850
|
```
|
1708
1851
|
or
|
1709
1852
|
```
|
1710
|
-
@
|
1711
|
-
{'name':'bar', 'parameters':{'flow_param_2': 'event_field_2'}])
|
1853
|
+
@trigger_on_finish(flows=['my_project.branch.my_branch.FooFlow', 'BarFlow'])
|
1712
1854
|
```
|
1713
1855
|
|
1714
|
-
|
1715
|
-
|
1716
|
-
@trigger(event={'name':'foo', 'parameters':['common_name', ('flow_param', 'event_field')]})
|
1717
|
-
```
|
1718
|
-
This is equivalent to:
|
1856
|
+
You can also specify just the project or project branch (other values will be
|
1857
|
+
inferred from the current project or project branch):
|
1719
1858
|
```
|
1720
|
-
@
|
1859
|
+
@trigger_on_finish(flow={"name": "FooFlow", "project": "my_project", "project_branch": "branch"})
|
1721
1860
|
```
|
1722
1861
|
|
1862
|
+
Note that `branch` is typically one of:
|
1863
|
+
- `prod`
|
1864
|
+
- `user.bob`
|
1865
|
+
- `test.my_experiment`
|
1866
|
+
- `prod.staging`
|
1867
|
+
|
1723
1868
|
Parameters
|
1724
1869
|
----------
|
1725
|
-
|
1726
|
-
|
1727
|
-
|
1728
|
-
|
1870
|
+
flow : Union[str, Dict[str, str]], optional, default None
|
1871
|
+
Upstream flow dependency for this flow.
|
1872
|
+
flows : List[Union[str, Dict[str, str]]], default []
|
1873
|
+
Upstream flow dependencies for this flow.
|
1729
1874
|
options : Dict[str, Any], default {}
|
1730
1875
|
Backend-specific configuration for tuning eventing behavior.
|
1731
1876
|
|
@@ -1734,47 +1879,51 @@ def trigger(*, event: typing.Union[str, typing.Dict[str, typing.Any], None] = No
|
|
1734
1879
|
...
|
1735
1880
|
|
1736
1881
|
@typing.overload
|
1737
|
-
def
|
1882
|
+
def trigger_on_finish(f: typing.Type[FlowSpecDerived]) -> typing.Type[FlowSpecDerived]:
|
1738
1883
|
...
|
1739
1884
|
|
1740
|
-
def
|
1885
|
+
def trigger_on_finish(f: typing.Optional[typing.Type[FlowSpecDerived]] = None, *, flow: typing.Union[typing.Dict[str, str], str, None] = None, flows: typing.List[typing.Union[str, typing.Dict[str, str]]] = [], options: typing.Dict[str, typing.Any] = {}):
|
1741
1886
|
"""
|
1742
|
-
Specifies the
|
1887
|
+
Specifies the flow(s) that this flow depends on.
|
1743
1888
|
|
1744
1889
|
```
|
1745
|
-
@
|
1890
|
+
@trigger_on_finish(flow='FooFlow')
|
1746
1891
|
```
|
1747
1892
|
or
|
1748
1893
|
```
|
1749
|
-
@
|
1894
|
+
@trigger_on_finish(flows=['FooFlow', 'BarFlow'])
|
1750
1895
|
```
|
1896
|
+
This decorator respects the @project decorator and triggers the flow
|
1897
|
+
when upstream runs within the same namespace complete successfully
|
1751
1898
|
|
1752
|
-
Additionally, you can specify
|
1753
|
-
|
1899
|
+
Additionally, you can specify project aware upstream flow dependencies
|
1900
|
+
by specifying the fully qualified project_flow_name.
|
1754
1901
|
```
|
1755
|
-
@
|
1902
|
+
@trigger_on_finish(flow='my_project.branch.my_branch.FooFlow')
|
1756
1903
|
```
|
1757
1904
|
or
|
1758
1905
|
```
|
1759
|
-
@
|
1760
|
-
{'name':'bar', 'parameters':{'flow_param_2': 'event_field_2'}])
|
1906
|
+
@trigger_on_finish(flows=['my_project.branch.my_branch.FooFlow', 'BarFlow'])
|
1761
1907
|
```
|
1762
1908
|
|
1763
|
-
|
1764
|
-
|
1765
|
-
@trigger(event={'name':'foo', 'parameters':['common_name', ('flow_param', 'event_field')]})
|
1766
|
-
```
|
1767
|
-
This is equivalent to:
|
1909
|
+
You can also specify just the project or project branch (other values will be
|
1910
|
+
inferred from the current project or project branch):
|
1768
1911
|
```
|
1769
|
-
@
|
1912
|
+
@trigger_on_finish(flow={"name": "FooFlow", "project": "my_project", "project_branch": "branch"})
|
1770
1913
|
```
|
1771
1914
|
|
1915
|
+
Note that `branch` is typically one of:
|
1916
|
+
- `prod`
|
1917
|
+
- `user.bob`
|
1918
|
+
- `test.my_experiment`
|
1919
|
+
- `prod.staging`
|
1920
|
+
|
1772
1921
|
Parameters
|
1773
1922
|
----------
|
1774
|
-
|
1775
|
-
|
1776
|
-
|
1777
|
-
|
1923
|
+
flow : Union[str, Dict[str, str]], optional, default None
|
1924
|
+
Upstream flow dependency for this flow.
|
1925
|
+
flows : List[Union[str, Dict[str, str]]], default []
|
1926
|
+
Upstream flow dependencies for this flow.
|
1778
1927
|
options : Dict[str, Any], default {}
|
1779
1928
|
Backend-specific configuration for tuning eventing behavior.
|
1780
1929
|
|
@@ -1782,70 +1931,52 @@ def trigger(f: typing.Optional[typing.Type[FlowSpecDerived]] = None, *, event: t
|
|
1782
1931
|
"""
|
1783
1932
|
...
|
1784
1933
|
|
1785
|
-
def project(*, name: str) -> typing.Callable[[typing.Type[FlowSpecDerived]], typing.Type[FlowSpecDerived]]:
|
1786
|
-
"""
|
1787
|
-
Specifies what flows belong to the same project.
|
1788
|
-
|
1789
|
-
A project-specific namespace is created for all flows that
|
1790
|
-
use the same `@project(name)`.
|
1791
|
-
|
1792
|
-
Parameters
|
1793
|
-
----------
|
1794
|
-
name : str
|
1795
|
-
Project name. Make sure that the name is unique amongst all
|
1796
|
-
projects that use the same production scheduler. The name may
|
1797
|
-
contain only lowercase alphanumeric characters and underscores.
|
1798
|
-
|
1799
|
-
|
1800
|
-
"""
|
1801
|
-
...
|
1802
|
-
|
1803
1934
|
@typing.overload
|
1804
|
-
def
|
1935
|
+
def schedule(*, hourly: bool = False, daily: bool = True, weekly: bool = False, cron: typing.Optional[str] = None, timezone: typing.Optional[str] = None) -> typing.Callable[[typing.Type[FlowSpecDerived]], typing.Type[FlowSpecDerived]]:
|
1805
1936
|
"""
|
1806
|
-
Specifies the
|
1807
|
-
|
1808
|
-
Use `@conda_base` to set common libraries required by all
|
1809
|
-
steps and use `@conda` to specify step-specific additions.
|
1937
|
+
Specifies the times when the flow should be run when running on a
|
1938
|
+
production scheduler.
|
1810
1939
|
|
1811
1940
|
Parameters
|
1812
1941
|
----------
|
1813
|
-
|
1814
|
-
|
1815
|
-
|
1816
|
-
|
1817
|
-
|
1818
|
-
|
1819
|
-
|
1820
|
-
|
1821
|
-
|
1822
|
-
|
1942
|
+
hourly : bool, default False
|
1943
|
+
Run the workflow hourly.
|
1944
|
+
daily : bool, default True
|
1945
|
+
Run the workflow daily.
|
1946
|
+
weekly : bool, default False
|
1947
|
+
Run the workflow weekly.
|
1948
|
+
cron : str, optional, default None
|
1949
|
+
Run the workflow at [a custom Cron schedule](https://docs.aws.amazon.com/eventbridge/latest/userguide/scheduled-events.html#cron-expressions)
|
1950
|
+
specified by this expression.
|
1951
|
+
timezone : str, optional, default None
|
1952
|
+
Timezone on which the schedule runs (default: None). Currently supported only for Argo workflows,
|
1953
|
+
which accepts timezones in [IANA format](https://nodatime.org/TimeZones).
|
1823
1954
|
"""
|
1824
1955
|
...
|
1825
1956
|
|
1826
1957
|
@typing.overload
|
1827
|
-
def
|
1958
|
+
def schedule(f: typing.Type[FlowSpecDerived]) -> typing.Type[FlowSpecDerived]:
|
1828
1959
|
...
|
1829
1960
|
|
1830
|
-
def
|
1961
|
+
def schedule(f: typing.Optional[typing.Type[FlowSpecDerived]] = None, *, hourly: bool = False, daily: bool = True, weekly: bool = False, cron: typing.Optional[str] = None, timezone: typing.Optional[str] = None):
|
1831
1962
|
"""
|
1832
|
-
Specifies the
|
1833
|
-
|
1834
|
-
Use `@conda_base` to set common libraries required by all
|
1835
|
-
steps and use `@conda` to specify step-specific additions.
|
1963
|
+
Specifies the times when the flow should be run when running on a
|
1964
|
+
production scheduler.
|
1836
1965
|
|
1837
1966
|
Parameters
|
1838
1967
|
----------
|
1839
|
-
|
1840
|
-
|
1841
|
-
|
1842
|
-
|
1843
|
-
|
1844
|
-
|
1845
|
-
|
1846
|
-
|
1847
|
-
|
1848
|
-
|
1968
|
+
hourly : bool, default False
|
1969
|
+
Run the workflow hourly.
|
1970
|
+
daily : bool, default True
|
1971
|
+
Run the workflow daily.
|
1972
|
+
weekly : bool, default False
|
1973
|
+
Run the workflow weekly.
|
1974
|
+
cron : str, optional, default None
|
1975
|
+
Run the workflow at [a custom Cron schedule](https://docs.aws.amazon.com/eventbridge/latest/userguide/scheduled-events.html#cron-expressions)
|
1976
|
+
specified by this expression.
|
1977
|
+
timezone : str, optional, default None
|
1978
|
+
Timezone on which the schedule runs (default: None). Currently supported only for Argo workflows,
|
1979
|
+
which accepts timezones in [IANA format](https://nodatime.org/TimeZones).
|
1849
1980
|
"""
|
1850
1981
|
...
|
1851
1982
|
|
@@ -1927,6 +2058,19 @@ def default_metadata() -> str:
|
|
1927
2058
|
...
|
1928
2059
|
|
1929
2060
|
class Metaflow(object, metaclass=type):
|
2061
|
+
"""
|
2062
|
+
Entry point to all objects in the Metaflow universe.
|
2063
|
+
|
2064
|
+
This object can be used to list all the flows present either through the explicit property
|
2065
|
+
or by iterating over this object.
|
2066
|
+
|
2067
|
+
Attributes
|
2068
|
+
----------
|
2069
|
+
flows : List[Flow]
|
2070
|
+
Returns the list of all `Flow` objects known to this metadata provider. Note that only
|
2071
|
+
flows present in the current namespace will be returned. A `Flow` is present in a namespace
|
2072
|
+
if it has at least one run in the namespace.
|
2073
|
+
"""
|
1930
2074
|
def __init__(self):
|
1931
2075
|
...
|
1932
2076
|
@property
|
@@ -1978,6 +2122,17 @@ class Metaflow(object, metaclass=type):
|
|
1978
2122
|
...
|
1979
2123
|
|
1980
2124
|
class Flow(metaflow.client.core.MetaflowObject, metaclass=type):
|
2125
|
+
"""
|
2126
|
+
A Flow represents all existing flows with a certain name, in other words,
|
2127
|
+
classes derived from `FlowSpec`. A container of `Run` objects.
|
2128
|
+
|
2129
|
+
Attributes
|
2130
|
+
----------
|
2131
|
+
latest_run : Run
|
2132
|
+
Latest `Run` (in progress or completed, successfully or not) of this flow.
|
2133
|
+
latest_successful_run : Run
|
2134
|
+
Latest successfully completed `Run` of this flow.
|
2135
|
+
"""
|
1981
2136
|
def __init__(self, *args, **kwargs):
|
1982
2137
|
...
|
1983
2138
|
@property
|
@@ -2064,6 +2219,26 @@ class Flow(metaflow.client.core.MetaflowObject, metaclass=type):
|
|
2064
2219
|
...
|
2065
2220
|
|
2066
2221
|
class Run(metaflow.client.core.MetaflowObject, metaclass=type):
|
2222
|
+
"""
|
2223
|
+
A `Run` represents an execution of a `Flow`. It is a container of `Step`s.
|
2224
|
+
|
2225
|
+
Attributes
|
2226
|
+
----------
|
2227
|
+
data : MetaflowData
|
2228
|
+
a shortcut to run['end'].task.data, i.e. data produced by this run.
|
2229
|
+
successful : bool
|
2230
|
+
True if the run completed successfully.
|
2231
|
+
finished : bool
|
2232
|
+
True if the run completed.
|
2233
|
+
finished_at : datetime
|
2234
|
+
Time this run finished.
|
2235
|
+
code : MetaflowCode
|
2236
|
+
Code package for this run (if present). See `MetaflowCode`.
|
2237
|
+
trigger : MetaflowTrigger
|
2238
|
+
Information about event(s) that triggered this run (if present). See `MetaflowTrigger`.
|
2239
|
+
end_task : Task
|
2240
|
+
`Task` for the end step (if it is present already).
|
2241
|
+
"""
|
2067
2242
|
def steps(self, *tags: str) -> typing.Iterator[metaflow.client.core.Step]:
|
2068
2243
|
"""
|
2069
2244
|
[Legacy function - do not use]
|
@@ -2296,6 +2471,23 @@ class Run(metaflow.client.core.MetaflowObject, metaclass=type):
|
|
2296
2471
|
...
|
2297
2472
|
|
2298
2473
|
class Step(metaflow.client.core.MetaflowObject, metaclass=type):
|
2474
|
+
"""
|
2475
|
+
A `Step` represents a user-defined step, that is, a method annotated with the `@step` decorator.
|
2476
|
+
|
2477
|
+
It contains `Task` objects associated with the step, that is, all executions of the
|
2478
|
+
`Step`. The step may contain multiple `Task`s in the case of a foreach step.
|
2479
|
+
|
2480
|
+
Attributes
|
2481
|
+
----------
|
2482
|
+
task : Task
|
2483
|
+
The first `Task` object in this step. This is a shortcut for retrieving the only
|
2484
|
+
task contained in a non-foreach step.
|
2485
|
+
finished_at : datetime
|
2486
|
+
Time when the latest `Task` of this step finished. Note that in the case of foreaches,
|
2487
|
+
this time may change during execution of the step.
|
2488
|
+
environment_info : Dict[str, Any]
|
2489
|
+
Information about the execution environment.
|
2490
|
+
"""
|
2299
2491
|
@property
|
2300
2492
|
def task(self) -> typing.Optional[metaflow.client.core.Task]:
|
2301
2493
|
"""
|
@@ -2430,6 +2622,55 @@ class Step(metaflow.client.core.MetaflowObject, metaclass=type):
|
|
2430
2622
|
...
|
2431
2623
|
|
2432
2624
|
class Task(metaflow.client.core.MetaflowObject, metaclass=type):
|
2625
|
+
"""
|
2626
|
+
A `Task` represents an execution of a `Step`.
|
2627
|
+
|
2628
|
+
It contains all `DataArtifact` objects produced by the task as
|
2629
|
+
well as metadata related to execution.
|
2630
|
+
|
2631
|
+
Note that the `@retry` decorator may cause multiple attempts of
|
2632
|
+
the task to be present. Usually you want the latest attempt, which
|
2633
|
+
is what instantiating a `Task` object returns by default. If
|
2634
|
+
you need to e.g. retrieve logs from a failed attempt, you can
|
2635
|
+
explicitly get information about a specific attempt by using the
|
2636
|
+
following syntax when creating a task:
|
2637
|
+
|
2638
|
+
`Task('flow/run/step/task', attempt=<attempt>)`
|
2639
|
+
|
2640
|
+
where `attempt=0` corresponds to the first attempt etc.
|
2641
|
+
|
2642
|
+
Attributes
|
2643
|
+
----------
|
2644
|
+
metadata : List[Metadata]
|
2645
|
+
List of all metadata events associated with the task.
|
2646
|
+
metadata_dict : Dict[str, str]
|
2647
|
+
A condensed version of `metadata`: A dictionary where keys
|
2648
|
+
are names of metadata events and values the latest corresponding event.
|
2649
|
+
data : MetaflowData
|
2650
|
+
Container of all data artifacts produced by this task. Note that this
|
2651
|
+
call downloads all data locally, so it can be slower than accessing
|
2652
|
+
artifacts individually. See `MetaflowData` for more information.
|
2653
|
+
artifacts : MetaflowArtifacts
|
2654
|
+
Container of `DataArtifact` objects produced by this task.
|
2655
|
+
successful : bool
|
2656
|
+
True if the task completed successfully.
|
2657
|
+
finished : bool
|
2658
|
+
True if the task completed.
|
2659
|
+
exception : object
|
2660
|
+
Exception raised by this task if there was one.
|
2661
|
+
finished_at : datetime
|
2662
|
+
Time this task finished.
|
2663
|
+
runtime_name : str
|
2664
|
+
Runtime this task was executed on.
|
2665
|
+
stdout : str
|
2666
|
+
Standard output for the task execution.
|
2667
|
+
stderr : str
|
2668
|
+
Standard error output for the task execution.
|
2669
|
+
code : MetaflowCode
|
2670
|
+
Code package for this task (if present). See `MetaflowCode`.
|
2671
|
+
environment_info : Dict[str, str]
|
2672
|
+
Information about the execution environment.
|
2673
|
+
"""
|
2433
2674
|
def __init__(self, *args, **kwargs):
|
2434
2675
|
...
|
2435
2676
|
@property
|
@@ -2744,6 +2985,21 @@ class Task(metaflow.client.core.MetaflowObject, metaclass=type):
|
|
2744
2985
|
...
|
2745
2986
|
|
2746
2987
|
class DataArtifact(metaflow.client.core.MetaflowObject, metaclass=type):
|
2988
|
+
"""
|
2989
|
+
A single data artifact and associated metadata. Note that this object does
|
2990
|
+
not contain other objects as it is the leaf object in the hierarchy.
|
2991
|
+
|
2992
|
+
Attributes
|
2993
|
+
----------
|
2994
|
+
data : object
|
2995
|
+
The data contained in this artifact, that is, the object produced during
|
2996
|
+
execution of this run.
|
2997
|
+
sha : string
|
2998
|
+
A unique ID of this artifact.
|
2999
|
+
finished_at : datetime
|
3000
|
+
Corresponds roughly to the `Task.finished_at` time of the parent `Task`.
|
3001
|
+
An alias for `DataArtifact.created_at`.
|
3002
|
+
"""
|
2747
3003
|
@property
|
2748
3004
|
def data(self) -> typing.Any:
|
2749
3005
|
"""
|
@@ -2800,6 +3056,44 @@ class DataArtifact(metaflow.client.core.MetaflowObject, metaclass=type):
|
|
2800
3056
|
...
|
2801
3057
|
|
2802
3058
|
class Runner(object, metaclass=type):
|
3059
|
+
"""
|
3060
|
+
Metaflow's Runner API that presents a programmatic interface
|
3061
|
+
to run flows and perform other operations either synchronously or asynchronously.
|
3062
|
+
The class expects a path to the flow file along with optional arguments
|
3063
|
+
that match top-level options on the command-line.
|
3064
|
+
|
3065
|
+
This class works as a context manager, calling `cleanup()` to remove
|
3066
|
+
temporary files at exit.
|
3067
|
+
|
3068
|
+
Example:
|
3069
|
+
```python
|
3070
|
+
with Runner('slowflow.py', pylint=False) as runner:
|
3071
|
+
result = runner.run(alpha=5, tags=["abc", "def"], max_workers=5)
|
3072
|
+
print(result.run.finished)
|
3073
|
+
```
|
3074
|
+
|
3075
|
+
Parameters
|
3076
|
+
----------
|
3077
|
+
flow_file : str
|
3078
|
+
Path to the flow file to run
|
3079
|
+
show_output : bool, default True
|
3080
|
+
Show the 'stdout' and 'stderr' to the console by default,
|
3081
|
+
Only applicable for synchronous 'run' and 'resume' functions.
|
3082
|
+
profile : Optional[str], default None
|
3083
|
+
Metaflow profile to use to run this run. If not specified, the default
|
3084
|
+
profile is used (or the one already set using `METAFLOW_PROFILE`)
|
3085
|
+
env : Optional[Dict], default None
|
3086
|
+
Additional environment variables to set for the Run. This overrides the
|
3087
|
+
environment set for this process.
|
3088
|
+
cwd : Optional[str], default None
|
3089
|
+
The directory to run the subprocess in; if not specified, the current
|
3090
|
+
directory is used.
|
3091
|
+
file_read_timeout : int, default 3600
|
3092
|
+
The timeout until which we try to read the runner attribute file.
|
3093
|
+
**kwargs : Any
|
3094
|
+
Additional arguments that you would pass to `python myflow.py` before
|
3095
|
+
the `run` command.
|
3096
|
+
"""
|
2803
3097
|
def __init__(self, flow_file: str, show_output: bool = True, profile: typing.Optional[str] = None, env: typing.Optional[typing.Dict] = None, cwd: typing.Optional[str] = None, file_read_timeout: int = 3600, **kwargs):
|
2804
3098
|
...
|
2805
3099
|
def __enter__(self) -> metaflow.runner.metaflow_runner.Runner:
|
@@ -2892,6 +3186,42 @@ class Runner(object, metaclass=type):
|
|
2892
3186
|
...
|
2893
3187
|
|
2894
3188
|
class NBRunner(object, metaclass=type):
|
3189
|
+
"""
|
3190
|
+
A wrapper over `Runner` for executing flows defined in a Jupyter
|
3191
|
+
notebook cell.
|
3192
|
+
|
3193
|
+
Instantiate this class on the last line of a notebook cell where
|
3194
|
+
a `flow` is defined. In contrast to `Runner`, this class is not
|
3195
|
+
meant to be used in a context manager. Instead, use a blocking helper
|
3196
|
+
function like `nbrun` (which calls `cleanup()` internally) or call
|
3197
|
+
`cleanup()` explictly when using non-blocking APIs.
|
3198
|
+
|
3199
|
+
```python
|
3200
|
+
run = NBRunner(FlowName).nbrun()
|
3201
|
+
```
|
3202
|
+
|
3203
|
+
Parameters
|
3204
|
+
----------
|
3205
|
+
flow : FlowSpec
|
3206
|
+
Flow defined in the same cell
|
3207
|
+
show_output : bool, default True
|
3208
|
+
Show the 'stdout' and 'stderr' to the console by default,
|
3209
|
+
Only applicable for synchronous 'run' and 'resume' functions.
|
3210
|
+
profile : Optional[str], default None
|
3211
|
+
Metaflow profile to use to run this run. If not specified, the default
|
3212
|
+
profile is used (or the one already set using `METAFLOW_PROFILE`)
|
3213
|
+
env : Optional[Dict], default None
|
3214
|
+
Additional environment variables to set for the Run. This overrides the
|
3215
|
+
environment set for this process.
|
3216
|
+
base_dir : Optional[str], default None
|
3217
|
+
The directory to run the subprocess in; if not specified, a temporary
|
3218
|
+
directory is used.
|
3219
|
+
file_read_timeout : int, default 3600
|
3220
|
+
The timeout until which we try to read the runner attribute file.
|
3221
|
+
**kwargs : Any
|
3222
|
+
Additional arguments that you would pass to `python myflow.py` before
|
3223
|
+
the `run` command.
|
3224
|
+
"""
|
2895
3225
|
def __init__(self, flow, show_output: bool = True, profile: typing.Optional[str] = None, env: typing.Optional[typing.Dict] = None, base_dir: str = "/tmp", file_read_timeout: int = 3600, **kwargs):
|
2896
3226
|
...
|
2897
3227
|
def nbrun(self, **kwargs):
|
@@ -2998,6 +3328,30 @@ class NBRunner(object, metaclass=type):
|
|
2998
3328
|
...
|
2999
3329
|
|
3000
3330
|
class Deployer(object, metaclass=type):
|
3331
|
+
"""
|
3332
|
+
Use the `Deployer` class to configure and access one of the production
|
3333
|
+
orchestrators supported by Metaflow.
|
3334
|
+
|
3335
|
+
Parameters
|
3336
|
+
----------
|
3337
|
+
flow_file : str
|
3338
|
+
Path to the flow file to deploy.
|
3339
|
+
show_output : bool, default True
|
3340
|
+
Show the 'stdout' and 'stderr' to the console by default.
|
3341
|
+
profile : Optional[str], default None
|
3342
|
+
Metaflow profile to use for the deployment. If not specified, the default
|
3343
|
+
profile is used.
|
3344
|
+
env : Optional[Dict[str, str]], default None
|
3345
|
+
Additional environment variables to set for the deployment.
|
3346
|
+
cwd : Optional[str], default None
|
3347
|
+
The directory to run the subprocess in; if not specified, the current
|
3348
|
+
directory is used.
|
3349
|
+
file_read_timeout : int, default 3600
|
3350
|
+
The timeout until which we try to read the deployer attribute file.
|
3351
|
+
**kwargs : Any
|
3352
|
+
Additional arguments that you would pass to `python myflow.py` before
|
3353
|
+
the deployment command.
|
3354
|
+
"""
|
3001
3355
|
def __init__(self, flow_file: str, show_output: bool = True, profile: typing.Optional[str] = None, env: typing.Optional[typing.Dict] = None, cwd: typing.Optional[str] = None, file_read_timeout: int = 3600, **kwargs):
|
3002
3356
|
...
|
3003
3357
|
def _Deployer__make_function(self, deployer_class):
|
@@ -3018,6 +3372,43 @@ class Deployer(object, metaclass=type):
|
|
3018
3372
|
...
|
3019
3373
|
|
3020
3374
|
class NBDeployer(object, metaclass=type):
|
3375
|
+
"""
|
3376
|
+
A wrapper over `Deployer` for deploying flows defined in a Jupyter
|
3377
|
+
notebook cell.
|
3378
|
+
|
3379
|
+
Instantiate this class on the last line of a notebook cell where
|
3380
|
+
a `flow` is defined. In contrast to `Deployer`, this class is not
|
3381
|
+
meant to be used in a context manager.
|
3382
|
+
|
3383
|
+
```python
|
3384
|
+
deployer = NBDeployer(FlowName)
|
3385
|
+
ar = deployer.argo_workflows(name="madhur")
|
3386
|
+
ar_obj = ar.create()
|
3387
|
+
result = ar_obj.trigger(alpha=300)
|
3388
|
+
print(result.status)
|
3389
|
+
print(result.run)
|
3390
|
+
result.terminate()
|
3391
|
+
```
|
3392
|
+
|
3393
|
+
Parameters
|
3394
|
+
----------
|
3395
|
+
flow : FlowSpec
|
3396
|
+
Flow defined in the same cell
|
3397
|
+
show_output : bool, default True
|
3398
|
+
Show the 'stdout' and 'stderr' to the console by default,
|
3399
|
+
profile : Optional[str], default None
|
3400
|
+
Metaflow profile to use to deploy this run. If not specified, the default
|
3401
|
+
profile is used (or the one already set using `METAFLOW_PROFILE`)
|
3402
|
+
env : Optional[Dict[str, str]], default None
|
3403
|
+
Additional environment variables to set. This overrides the
|
3404
|
+
environment set for this process.
|
3405
|
+
base_dir : Optional[str], default None
|
3406
|
+
The directory to run the subprocess in; if not specified, a temporary
|
3407
|
+
directory is used.
|
3408
|
+
**kwargs : Any
|
3409
|
+
Additional arguments that you would pass to `python myflow.py` i.e. options
|
3410
|
+
listed in `python myflow.py --help`
|
3411
|
+
"""
|
3021
3412
|
def __init__(self, flow, show_output: bool = True, profile: typing.Optional[str] = None, env: typing.Optional[typing.Dict] = None, base_dir: str = "/tmp", file_read_timeout: int = 3600, **kwargs):
|
3022
3413
|
...
|
3023
3414
|
def cleanup(self):
|