Execution¶

Materializing Assets¶

dagster.materialize(assets, run_config=None, instance=None, resources=None, partition_key=None, raise_on_error=True, tags=None, selection=None)[source]¶

Executes a single-threaded, in-process run which materializes provided assets.

By default, will materialize assets to the local filesystem.

Parameters:

assets (Sequence[Union[AssetsDefinition, SourceAsset]]) –
The assets to materialize.

Unless you’re using deps or non_argument_deps, you must also include all assets that are upstream of the assets that you want to materialize. This is because those upstream asset definitions have information that is needed to load their contents while materializing the downstream assets.

You can use the selection argument to distinguish between assets that you want to materialize and assets that are just present for loading.
resources (Optional[Mapping[str, object]]) – The resources needed for execution. Can provide resource instances directly, or resource definitions. Note that if provided resources conflict with resources directly on assets, an error will be thrown.
run_config (Optional[Any]) – The run config to use for the run that materializes the assets.
partition_key – (Optional[str]) The string partition key that specifies the run config to execute. Can only be used to select run config for assets with partitioned config.
tags (Optional[Mapping[str, str]]) – Tags for the run.
selection (Optional[Union[str, Sequence[str], Sequence[AssetKey], Sequence[Union[AssetsDefinition, SourceAsset]], AssetSelection]]) –
A sub-selection of assets to materialize.

If not provided, then all assets will be materialized.

If providing a string or sequence of strings, https://docs.dagster.io/concepts/assets/asset-selection-syntax describes the accepted syntax.

Returns:

The result of the execution.

Return type:

ExecuteInProcessResult

Examples

@asset
def asset1():
    ...

@asset
def asset2(asset1):
    ...

# executes a run that materializes asset1 and then asset2
materialize([asset1, asset2])

# executes a run that materializes just asset2, loading its input from asset1
materialize([asset1, asset2], selection=[asset2])

dagster.materialize_to_memory(assets, run_config=None, instance=None, resources=None, partition_key=None, raise_on_error=True, tags=None, selection=None)[source]¶

Executes a single-threaded, in-process run which materializes provided assets in memory.

Will explicitly use mem_io_manager() for all required io manager keys. If any io managers are directly provided using the resources argument, a DagsterInvariantViolationError will be thrown.

Parameters:

assets (Sequence[Union[AssetsDefinition, SourceAsset]]) – The assets to materialize. Can also provide SourceAsset objects to fill dependencies for asset defs.
run_config (Optional[Any]) – The run config to use for the run that materializes the assets.
resources (Optional[Mapping[str, object]]) – The resources needed for execution. Can provide resource instances directly, or resource definitions. If provided resources conflict with resources directly on assets, an error will be thrown.
partition_key – (Optional[str]) The string partition key that specifies the run config to execute. Can only be used to select run config for assets with partitioned config.
tags (Optional[Mapping[str, str]]) – Tags for the run.
selection (Optional[Union[str, Sequence[str], Sequence[AssetKey], Sequence[Union[AssetsDefinition, SourceAsset]], AssetSelection]]) –
A sub-selection of assets to materialize.

If not provided, then all assets will be materialized.

If providing a string or sequence of strings, https://docs.dagster.io/concepts/assets/asset-selection-syntax describes the accepted syntax.

Returns:

The result of the execution.

Return type:

ExecuteInProcessResult

Examples

@asset
def asset1():
    ...

@asset
def asset2(asset1):
    ...

# executes a run that materializes asset1 and then asset2
materialize([asset1, asset2])

# executes a run that materializes just asset1
materialize([asset1, asset2], selection=[asset1])

Executing Jobs¶

class dagster.JobDefinition(*, graph_def, resource_defs=None, executor_def=None, logger_defs=None, name=None, config=None, description=None, partitions_def=None, tags=None, metadata=None, hook_defs=None, op_retry_policy=None, version_strategy=None, _subset_selection_data=None, asset_layer=None, input_values=None, _was_explicitly_provided_resources=None)[source]

Defines a Dagster job.

property config_mapping

The config mapping for the job, if it has one.

A config mapping defines a way to map a top-level config schema to run config for the job.

execute_in_process(run_config=None, instance=None, partition_key=None, raise_on_error=True, op_selection=None, asset_selection=None, run_id=None, input_values=None, tags=None, resources=None)[source]

Execute the Job in-process, gathering results in-memory.

The executor_def on the Job will be ignored, and replaced with the in-process executor. If using the default io_manager, it will switch from filesystem to in-memory.

Parameters:

(Optional[Mapping[str (run_config) – The configuration for the run
Any]] – The configuration for the run
instance (Optional[DagsterInstance]) – The instance to execute against, an ephemeral one will be used if none provided.
partition_key – (Optional[str]) The string partition key that specifies the run config to execute. Can only be used to select run config for jobs with partitioned config.
raise_on_error (Optional[bool]) – Whether or not to raise exceptions when they occur. Defaults to True.
op_selection (Optional[Sequence[str]]) – A list of op selection queries (including single op names) to execute. For example: * ['some_op']: selects some_op itself. * ['*some_op']: select some_op and all its ancestors (upstream dependencies). * ['*some_op+++']: select some_op, all its ancestors, and its descendants (downstream dependencies) within 3 levels down. * ['*some_op', 'other_op_a', 'other_op_b+']: select some_op and all its ancestors, other_op_a itself, and other_op_b and its direct child ops.
input_values (Optional[Mapping[str, Any]]) – A dictionary that maps python objects to the top-level inputs of the job. Input values provided here will override input values that have been provided to the job directly.
resources (Optional[Mapping[str, Any]]) – The resources needed if any are required. Can provide resource instances directly, or resource definitions.

Returns:

ExecuteInProcessResult

property executor_def

Returns the default ExecutorDefinition for the job.

If the user has not specified an executor definition, then this will default to the multi_or_in_process_executor(). If a default is specified on the Definitions object the job was provided to, then that will be used instead.

property has_specified_executor: Returns True if this job has explicitly specified an executor, and False if the executor was inherited through defaults or the Definitions object the job was provided to.

property has_specified_loggers: Returns true if the job explicitly set loggers, and False if loggers were inherited through defaults or the Definitions object the job was provided to.

property loggers

Returns the set of LoggerDefinition objects specified on the job.

If the user has not specified a mapping of LoggerDefinition objects, then this will default to the colored_console_logger() under the key console. If a default is specified on the Definitions object the job was provided to, then that will be used instead.

property partitioned_config

The partitioned config for the job, if it has one.

A partitioned config defines a way to map partition keys to run config for the job.

property partitions_def

Returns the PartitionsDefinition for the job, if it has one.

A partitions definition defines the set of partition keys the job operates on.

property resource_defs

Returns the set of ResourceDefinition objects specified on the job.

This may not be the complete set of resources required by the job, since those can also be provided on the Definitions object the job may be provided to.

run_request_for_partition(partition_key, run_key=None, tags=None, asset_selection=None, run_config=None, current_time=None, dynamic_partitions_store=None)[source]

Deprecated

Creates a RunRequest object for a run that processes the given partition.

Parameters:

partition_key – The key of the partition to request a run for.
run_key (Optional[str]) – A string key to identify this launched run. For sensors, ensures that only one run is created per run key across all sensor evaluations. For schedules, ensures that one run is created per tick, across failure recoveries. Passing in a None value means that a run will always be launched per evaluation.
tags (Optional[Dict[str, str]]) – A dictionary of tags (string key-value pairs) to attach to the launched run.
(Optional[Mapping[str (run_config) – Configuration for the run. If the job has a PartitionedConfig, this value will override replace the config provided by it.
Any]] – Configuration for the run. If the job has a PartitionedConfig, this value will override replace the config provided by it.
current_time (Optional[datetime]) – Used to determine which time-partitions exist. Defaults to now.
dynamic_partitions_store (Optional[DynamicPartitionsStore]) – The DynamicPartitionsStore object that is responsible for fetching dynamic partitions. Required when the partitions definition is a DynamicPartitionsDefinition with a name defined. Users can pass the DagsterInstance fetched via context.instance to this argument.

Returns:

an object that requests a run to process the given partition.

Return type:

RunRequest

with_hooks(hook_defs)[source]: Apply a set of hooks to all op instances within the job.

with_top_level_resources(resource_defs)[source]: Apply a set of resources to all op instances within the job.

dagster.execute_job(job, instance, run_config=None, tags=None, raise_on_error=False, op_selection=None, reexecution_options=None, asset_selection=None)[source]¶

Execute a job synchronously.

This API represents dagster’s python entrypoint for out-of-process execution. For most testing purposes, execute_in_process() will be more suitable, but when wanting to run execution using an out-of-process executor (such as dagster. multiprocess_executor), then execute_job is suitable.

execute_job expects a persistent DagsterInstance for execution, meaning the $DAGSTER_HOME environment variable must be set. It also expects a reconstructable pointer to a JobDefinition so that it can be reconstructed in separate processes. This can be done by wrapping the JobDefinition in a call to dagster. reconstructable().

from dagster import DagsterInstance, execute_job, job, reconstructable

@job
def the_job():
    ...

instance = DagsterInstance.get()
result = execute_job(reconstructable(the_job), instance=instance)
assert result.success

If using the to_job() method to construct the JobDefinition, then the invocation must be wrapped in a module-scope function, which can be passed to reconstructable.

from dagster import graph, reconstructable

@graph
def the_graph():
    ...

def define_job():
    return the_graph.to_job(...)

result = execute_job(reconstructable(define_job), ...)

Since execute_job is potentially executing outside of the current process, output objects need to be retrieved by use of the provided job’s io managers. Output objects can be retrieved by opening the result of execute_job as a context manager.

from dagster import execute_job

with execute_job(...) as result:
    output_obj = result.output_for_node("some_op")

execute_job can also be used to reexecute a run, by providing a ReexecutionOptions object.

from dagster import ReexecutionOptions, execute_job

instance = DagsterInstance.get()

options = ReexecutionOptions.from_failure(run_id=failed_run_id, instance)
execute_job(reconstructable(job), instance, reexecution_options=options)

Parameters:

job (ReconstructableJob) – A reconstructable pointer to a JobDefinition.
instance (DagsterInstance) – The instance to execute against.
run_config (Optional[dict]) – The configuration that parametrizes this run, as a dict.
tags (Optional[Dict[str, Any]]) – Arbitrary key-value pairs that will be added to run logs.
raise_on_error (Optional[bool]) – Whether or not to raise exceptions when they occur. Defaults to False.
op_selection (Optional[List[str]]) –
A list of op selection queries (including single op names) to execute. For example:
- ['some_op']: selects some_op itself.
- ['*some_op']: select some_op and all its ancestors (upstream dependencies).
- ['*some_op+++']: select some_op, all its ancestors, and its descendants (downstream dependencies) within 3 levels down.
- ['*some_op', 'other_op_a', 'other_op_b+']: select some_op and all its ancestors, other_op_a itself, and other_op_b and its direct child ops.
reexecution_options (Optional[ReexecutionOptions]) – Reexecution options to provide to the run, if this run is intended to be a reexecution of a previous run. Cannot be used in tandem with the op_selection argument.

Returns:

The result of job execution.

Return type:

JobExecutionResult

class dagster.ReexecutionOptions(parent_run_id, step_selection=[])[source]¶

Reexecution options for python-based execution in Dagster.

Parameters:

parent_run_id (str) – The run_id of the run to reexecute.
step_selection (Sequence[str]) –
The list of step selections to reexecute. Must be a subset or match of the set of steps executed in the original run. For example:
- ['some_op']: selects some_op itself.
- ['*some_op']: select some_op and all its ancestors (upstream dependencies).
- ['*some_op+++']: select some_op, all its ancestors, and its descendants (downstream dependencies) within 3 levels down.
- ['*some_op', 'other_op_a', 'other_op_b+']: select some_op and all its ancestors, other_op_a itself, and other_op_b and its direct child ops.

dagster.instance_for_test(overrides=None, set_dagster_home=True, temp_dir=None)[source]¶

Creates a persistent DagsterInstance available within a context manager.

When a context manager is opened, if no temp_dir parameter is set, a new temporary directory will be created for the duration of the context manager’s opening. If the set_dagster_home parameter is set to True (True by default), the $DAGSTER_HOME environment variable will be overridden to be this directory (or the directory passed in by temp_dir) for the duration of the context manager being open.

Parameters:

overrides (Optional[Mapping[str, Any]]) – Config to provide to instance (config format follows that typically found in an instance.yaml file).
set_dagster_home (Optional[bool]) – If set to True, the $DAGSTER_HOME environment variable will be overridden to be the directory used by this instance for the duration that the context manager is open. Upon the context manager closing, the $DAGSTER_HOME variable will be re-set to the original value. (Defaults to True).
temp_dir (Optional[str]) – The directory to use for storing local artifacts produced by the instance. If not set, a temporary directory will be created for the duration of the context manager being open, and all artifacts will be torn down afterward.

Executing Graphs¶

class dagster.GraphDefinition(name, *, description=None, node_defs=None, dependencies=None, input_mappings=None, output_mappings=None, config=None, tags=None, node_input_source_assets=None, **kwargs)[source]

Defines a Dagster op graph.

An op graph is made up of

Nodes, which can either be an op (the functional unit of computation), or another graph.
Dependencies, which determine how the values produced by nodes as outputs flow from one node to another. This tells Dagster how to arrange nodes into a directed, acyclic graph (DAG) of compute.

End users should prefer the @graph decorator. GraphDefinition is generally intended to be used by framework authors or for programatically generated graphs.

Parameters:

name (str) – The name of the graph. Must be unique within any GraphDefinition or JobDefinition containing the graph.
description (Optional[str]) – A human-readable description of the job.
node_defs (Optional[Sequence[NodeDefinition]]) – The set of ops / graphs used in this graph.
dependencies (Optional[Dict[Union[str, NodeInvocation], Dict[str, DependencyDefinition]]]) – A structure that declares the dependencies of each op’s inputs on the outputs of other ops in the graph. Keys of the top level dict are either the string names of ops in the graph or, in the case of aliased ops, NodeInvocations. Values of the top level dict are themselves dicts, which map input names belonging to the op or aliased op to DependencyDefinitions.
input_mappings (Optional[Sequence[InputMapping]]) – Defines the inputs to the nested graph, and how they map to the inputs of its constituent ops.
output_mappings (Optional[Sequence[OutputMapping]]) – Defines the outputs of the nested graph, and how they map from the outputs of its constituent ops.
config (Optional[ConfigMapping]) – Defines the config of the graph, and how its schema maps to the config of its constituent ops.
tags (Optional[Dict[str, Any]]) – Arbitrary metadata for any execution of the graph. Values that are not strings will be json encoded and must meet the criteria that json.loads(json.dumps(value)) == value. These tag values may be overwritten by tag values provided at invocation time.

Examples

@op
def return_one():
    return 1

@op
def add_one(num):
    return num + 1

graph_def = GraphDefinition(
    name='basic',
    node_defs=[return_one, add_one],
    dependencies={'add_one': {'num': DependencyDefinition('return_one')}},
)

alias(name)[source]

Aliases the graph with a new name.

Can only be used in the context of a @graph, @job, or @asset_graph decorated function.

Examples:

@job
def do_it_all():
    my_graph.alias("my_graph_alias")

property config_mapping

The config mapping for the graph, if present.

By specifying a config mapping function, you can override the configuration for the child nodes contained within a graph.

execute_in_process(run_config=None, instance=None, resources=None, raise_on_error=True, op_selection=None, run_id=None, input_values=None)[source]

Execute this graph in-process, collecting results in-memory.

Parameters:

run_config (Optional[Mapping[str, Any]]) – Run config to provide to execution. The configuration for the underlying graph should exist under the “ops” key.
instance (Optional[DagsterInstance]) – The instance to execute against, an ephemeral one will be used if none provided.
resources (Optional[Mapping[str, Any]]) – The resources needed if any are required. Can provide resource instances directly, or resource definitions.
raise_on_error (Optional[bool]) – Whether or not to raise exceptions when they occur. Defaults to True.
op_selection (Optional[List[str]]) – A list of op selection queries (including single op names) to execute. For example: * ['some_op']: selects some_op itself. * ['*some_op']: select some_op and all its ancestors (upstream dependencies). * ['*some_op+++']: select some_op, all its ancestors, and its descendants (downstream dependencies) within 3 levels down. * ['*some_op', 'other_op_a', 'other_op_b+']: select some_op and all its ancestors, other_op_a itself, and other_op_b and its direct child ops.
input_values (Optional[Mapping[str, Any]]) – A dictionary that maps python objects to the top-level inputs of the graph.

Returns:

ExecuteInProcessResult

property input_mappings

Input mappings for the graph.

An input mapping is a mapping from an input of the graph to an input of a child node.

property name: The name of the graph.

property output_mappings

Output mappings for the graph.

An output mapping is a mapping from an output of the graph to an output of a child node.

tag(tags)[source]

Attaches the provided tags to the graph immutably.

Can only be used in the context of a @graph, @job, or @asset_graph decorated function.

Examples:

@job
def do_it_all():
    my_graph.tag({"my_tag": "my_value"})

property tags: The tags associated with the graph.

to_job(name=None, description=None, resource_defs=None, config=None, tags=None, metadata=None, logger_defs=None, executor_def=None, hooks=None, op_retry_policy=None, version_strategy=None, op_selection=None, partitions_def=None, asset_layer=None, input_values=None, _asset_selection_data=None)[source]

Make this graph in to an executable Job by providing remaining components required for execution.

Parameters:

name (Optional[str]) – The name for the Job. Defaults to the name of the this graph.
resource_defs (Optional[Mapping [str, object]]) – Resources that are required by this graph for execution. If not defined, io_manager will default to filesystem.
config –
Describes how the job is parameterized at runtime.

If no value is provided, then the schema for the job’s run config is a standard format based on its ops and resources.

If a dictionary is provided, then it must conform to the standard config schema, and it will be used as the job’s run config for the job whenever the job is executed. The values provided will be viewable and editable in the Dagster UI, so be careful with secrets.

If a ConfigMapping object is provided, then the schema for the job’s run config is determined by the config mapping, and the ConfigMapping, which should return configuration in the standard format to configure the job.

If a PartitionedConfig object is provided, then it defines a discrete set of config values that can parameterize the job, as well as a function for mapping those values to the base config. The values provided will be viewable and editable in the Dagster UI, so be careful with secrets.
tags (Optional[Mapping[str, Any]]) – Arbitrary information that will be attached to the execution of the Job. Values that are not strings will be json encoded and must meet the criteria that json.loads(json.dumps(value)) == value. These tag values may be overwritten by tag values provided at invocation time.
metadata (Optional[Mapping[str, RawMetadataValue]]) – Arbitrary information that will be attached to the JobDefinition and be viewable in the Dagster UI. Keys must be strings, and values must be python primitive types or one of the provided MetadataValue types
logger_defs (Optional[Mapping[str, LoggerDefinition]]) – A dictionary of string logger identifiers to their implementations.
executor_def (Optional[ExecutorDefinition]) – How this Job will be executed. Defaults to multi_or_in_process_executor, which can be switched between multi-process and in-process modes of execution. The default mode of execution is multi-process.
op_retry_policy (Optional[RetryPolicy]) – The default retry policy for all ops in this job. Only used if retry policy is not defined on the op definition or op invocation.
version_strategy (Optional[VersionStrategy]) – Defines how each op (and optionally, resource) in the job can be versioned. If provided, memoizaton will be enabled for this job.
partitions_def (Optional[PartitionsDefinition]) – Defines a discrete set of partition keys that can parameterize the job. If this argument is supplied, the config argument can’t also be supplied.
asset_layer (Optional[AssetLayer]) – Top level information about the assets this job will produce. Generally should not be set manually.
input_values (Optional[Mapping[str, Any]]) – A dictionary that maps python objects to the top-level inputs of a job.

Returns:

JobDefinition

with_hooks(hook_defs)[source]

Attaches the provided hooks to the graph immutably.

Can only be used in the context of a @graph, @job, or @asset_graph decorated function.

Examples:

@job
def do_it_all():
    my_graph.with_hooks({my_hook})

with_retry_policy(retry_policy)[source]

Attaches the provided retry policy to the graph immutably.

Can only be used in the context of a @graph, @job, or @asset_graph decorated function.

Examples:

@job
def do_it_all():
    my_graph.with_retry_policy(RetryPolicy(max_retries=5))

Execution results¶

class dagster.ExecuteInProcessResult(event_list, dagster_run, output_capture, job_def)[source]¶

Result object returned by in-process testing APIs.

Users should not instantiate this object directly. Used for retrieving run success, events, and outputs from execution methods that return this object.

This object is returned by: - dagster.GraphDefinition.execute_in_process() - dagster.JobDefinition.execute_in_process() - dagster.materialize_to_memory() - dagster.materialize()

property all_events¶

All dagster events emitted during execution.

Type:: List[DagsterEvent]

asset_value(asset_key)[source]¶

Retrieves the value of an asset that was materialized during the execution of the job.

Parameters:: asset_key (CoercibleToAssetKey) – The key of the asset to retrieve.
Returns:: The value of the retrieved asset.
Return type:: Any

property dagster_run¶

The Dagster run that was executed.

Type:: DagsterRun

property job_def¶

The job definition that was executed.

Type:: JobDefinition

output_for_node(node_str, output_name='result')[source]¶

Retrieves output value with a particular name from the in-process run of the job.

Parameters:

node_str (str) – Name of the op/graph whose output should be retrieved. If the intended graph/op is nested within another graph, the syntax is outer_graph.inner_node.
output_name (Optional[str]) – Name of the output on the op/graph to retrieve. Defaults to result, the default output name in dagster.

Returns:

The value of the retrieved output.

Return type:

Any

output_value(output_name='result')[source]¶

Retrieves output of top-level job, if an output is returned.

Parameters:: output_name (Optional[str]) – The name of the output to retrieve. Defaults to result, the default output name in dagster.
Returns:: The value of the retrieved output.
Return type:: Any

property run_id¶

The run ID of the executed DagsterRun.

Type:: str

class dagster.JobExecutionResult(job_def, reconstruct_context, event_list, dagster_run)[source]¶

Result object returned by dagster.execute_job().

Used for retrieving run success, events, and outputs from execute_job. Users should not directly instantiate this class.

Events and run information can be retrieved off of the object directly. In order to access outputs, the ExecuteJobResult object needs to be opened as a context manager, which will re-initialize the resources from execution.

property all_events¶

List of all events yielded by the job execution.

Type:: Sequence[DagsterEvent]

property dagster_run¶

The Dagster run that was executed.

Type:: DagsterRun

property job_def¶

The job definition that was executed.

Type:: JobDefinition

output_for_node(node_str, output_name='result')[source]¶

Retrieves output value with a particular name from the run of the job.

Parameters:

node_str (str) – Name of the op/graph whose output should be retrieved. If the intended graph/op is nested within another graph, the syntax is outer_graph.inner_node.
output_name (Optional[str]) – Name of the output on the op/graph to retrieve. Defaults to result, the default output name in dagster.

Returns:

The value of the retrieved output.

Return type:

Any

output_value(output_name='result')[source]¶

Retrieves output of top-level job, if an output is returned.

In order to use this method, the ExecuteJobResult object must be opened as a context manager. If this method is used without opening the context manager, it will result in a DagsterInvariantViolationError. If the top-level job has no output, calling this method will also result in a DagsterInvariantViolationError.

Parameters:: output_name (Optional[str]) – The name of the output to retrieve. Defaults to result, the default output name in dagster.
Returns:: The value of the retrieved output.
Return type:: Any

property run_id¶

The id of the Dagster run that was executed.

Type:: str

class dagster.DagsterEvent(event_type_value, job_name, step_handle=None, node_handle=None, step_kind_value=None, logging_tags=None, event_specific_data=None, message=None, pid=None, step_key=None)[source]¶

Events yielded by op and job execution.

Users should not instantiate this class.

event_type_value¶

Value for a DagsterEventType.

Type:: str

job_name¶

Type:: str

node_handle¶

Type:: NodeHandle

step_kind_value¶

Value for a StepKind.

Type:: str

logging_tags¶

Type:: Dict[str, str]

event_specific_data¶

Type must correspond to event_type_value.

Type:: Any

message¶

Type:: str

pid¶

Type:: int

step_key¶

DEPRECATED

Type:: Optional[str]

property asset_key¶

For events that correspond to a specific asset_key / partition (ASSET_MATERIALIZTION, ASSET_OBSERVATION, ASSET_MATERIALIZATION_PLANNED), returns that asset key. Otherwise, returns None.

Type:: Optional[AssetKey]

property event_type¶

The type of this event.

Type:: DagsterEventType

property is_asset_materialization_planned¶

If this event is of type ASSET_MATERIALIZATION_PLANNED.

Type:: bool

property is_asset_observation¶

If this event is of type ASSET_OBSERVATION.

Type:: bool

property is_engine_event¶

If this event is of type ENGINE_EVENT.

Type:: bool

property is_expectation_result¶

If this event is of type STEP_EXPECTATION_RESULT.

Type:: bool

property is_failure¶

If this event represents the failure of a run or step.

Type:: bool

property is_handled_output¶

If this event is of type HANDLED_OUTPUT.

Type:: bool

property is_hook_event¶

If this event relates to the execution of a hook.

Type:: bool

property is_loaded_input¶

If this event is of type LOADED_INPUT.

Type:: bool

property is_resource_init_failure¶

If this event is of type RESOURCE_INIT_FAILURE.

Type:: bool

property is_step_event¶

If this event relates to a specific step.

Type:: bool

property is_step_failure¶

If this event is of type STEP_FAILURE.

Type:: bool

property is_step_materialization¶

If this event is of type ASSET_MATERIALIZATION.

Type:: bool

property is_step_restarted¶

If this event is of type STEP_RESTARTED.

Type:: bool

property is_step_skipped¶

If this event is of type STEP_SKIPPED.

Type:: bool

property is_step_start¶

If this event is of type STEP_START.

Type:: bool

property is_step_success¶

If this event is of type STEP_SUCCESS.

Type:: bool

property is_step_up_for_retry¶

If this event is of type STEP_UP_FOR_RETRY.

Type:: bool

property is_successful_output¶

If this event is of type STEP_OUTPUT.

Type:: bool

property partition¶

For events that correspond to a specific asset_key / partition (ASSET_MATERIALIZTION, ASSET_OBSERVATION, ASSET_MATERIALIZATION_PLANNED), returns that partition. Otherwise, returns None.

Type:: Optional[AssetKey]

class dagster.DagsterEventType(value)[source]¶: The types of events that may be yielded by op and job execution.

Reconstructable jobs¶

class dagster.reconstructable(target)[source]¶

Create a ReconstructableJob from a function that returns a JobDefinition/JobDefinition, or a function decorated with @job.

When your job must cross process boundaries, e.g., for execution on multiple nodes or in different systems (like dagstermill), Dagster must know how to reconstruct the job on the other side of the process boundary.

Passing a job created with ~dagster.GraphDefinition.to_job to reconstructable(), requires you to wrap that job’s definition in a module-scoped function, and pass that function instead:

from dagster import graph, reconstructable

@graph
def my_graph():
    ...

def define_my_job():
    return my_graph.to_job()

reconstructable(define_my_job)

This function implements a very conservative strategy for reconstruction, so that its behavior is easy to predict, but as a consequence it is not able to reconstruct certain kinds of jobs or jobs, such as those defined by lambdas, in nested scopes (e.g., dynamically within a method call), or in interactive environments such as the Python REPL or Jupyter notebooks.

If you need to reconstruct objects constructed in these ways, you should use build_reconstructable_job() instead, which allows you to specify your own reconstruction strategy.

Examples

from dagster import job, reconstructable

@job
def foo_job():
    ...

reconstructable_foo_job = reconstructable(foo_job)


@graph
def foo():
    ...

def make_bar_job():
    return foo.to_job()

reconstructable_bar_job = reconstructable(make_bar_job)

Executors¶

dagster.multi_or_in_process_executor ExecutorDefinition[source]¶

The default executor for a job.

This is the executor available by default on a JobDefinition that does not provide custom executors. This executor has a multiprocessing-enabled mode, and a single-process mode. By default, multiprocessing mode is enabled. Switching between multiprocess mode and in-process mode can be achieved via config.

execution:
  config:
    multiprocess:


execution:
  config:
    in_process:

When using the multiprocess mode, max_concurrent and retries can also be configured.

execution:
  config:
    multiprocess:
      max_concurrent: 4
      retries:
        enabled:

The max_concurrent arg is optional and tells the execution engine how many processes may run concurrently. By default, or if you set max_concurrent to be 0, this is the return value of python:multiprocessing.cpu_count().

When using the in_process mode, then only retries can be configured.

Execution priority can be configured using the dagster/priority tag via op metadata, where the higher the number the higher the priority. 0 is the default and both positive and negative numbers can be used.

dagster.in_process_executor ExecutorDefinition[source]¶

The in-process executor executes all steps in a single process.

To select it, include the following top-level fragment in config:

execution:
  in_process:

dagster.multiprocess_executor ExecutorDefinition[source]¶

The multiprocess executor executes each step in an individual process.

Any job that does not specify custom executors will use the multiprocess_executor by default. To configure the multiprocess executor, include a fragment such as the following in your run config:

execution:
  config:
    multiprocess:
      max_concurrent: 4

The max_concurrent arg is optional and tells the execution engine how many processes may run concurrently. By default, or if you set max_concurrent to be None or 0, this is the return value of python:multiprocessing.cpu_count().

Contexts¶

class dagster.OpExecutionContext(step_execution_context)[source]¶

The context object that can be made available as the first argument to the function used for computing an op or asset.

This context object provides system information such as resources, config, and logging.

To construct an execution context for testing purposes, use dagster.build_op_context().

Example

from dagster import op, OpExecutionContext

@op
def hello_world(context: OpExecutionContext):
    context.log.info("Hello, world!")

asset_key_for_input(input_name)[source]¶: Return the AssetKey for the corresponding input.

asset_key_for_output(output_name='result')[source]¶: Return the AssetKey for the corresponding output.

asset_partition_key_for_input(input_name)[source]¶: Returns the partition key of the upstream asset corresponding to the given input.

asset_partition_key_for_output(output_name='result')[source]¶: Returns the asset partition key for the given output. Defaults to “result”, which is the name of the default output.

property asset_partition_key_range¶

Deprecated

The range of partition keys for the current run.

If run is for a single partition key, return a PartitionKeyRange with the same start and end. Raises an error if the current run is not a partitioned run.

asset_partition_key_range_for_input(input_name)[source]¶: Return the PartitionKeyRange for the corresponding input. Errors if there is more or less than one.

asset_partition_key_range_for_output(output_name='result')[source]¶: Return the PartitionKeyRange for the corresponding output. Errors if not present.

asset_partition_keys_for_input(input_name)[source]¶: Returns a list of the partition keys of the upstream asset corresponding to the given input.

asset_partition_keys_for_output(output_name='result')[source]¶: Returns a list of the partition keys for the given output.

asset_partitions_def_for_input(input_name)[source]¶: The PartitionsDefinition on the upstream asset corresponding to this input.

asset_partitions_def_for_output(output_name='result')[source]¶: The PartitionsDefinition on the upstream asset corresponding to this input.

asset_partitions_time_window_for_input(input_name='result')[source]¶

The time window for the partitions of the input asset.

Raises an error if either of the following are true: - The input asset has no partitioning. - The input asset is not partitioned with a TimeWindowPartitionsDefinition or a MultiPartitionsDefinition with one time-partitioned dimension.

asset_partitions_time_window_for_output(output_name='result')[source]¶

The time window for the partitions of the output asset.

Raises an error if either of the following are true: - The output asset has no partitioning. - The output asset is not partitioned with a TimeWindowPartitionsDefinition or a MultiPartitionsDefinition with one time-partitioned dimension.

property assets_def¶: The backing AssetsDefinition for what is currently executing, errors if not available.

get_asset_provenance(asset_key)[source]¶

Return the provenance information for the most recent materialization of an asset.

Parameters:

asset_key (AssetKey) – Key of the asset for which to retrieve provenance.

Returns:

Provenance information for the most recent: materialization of the asset. Returns None if the asset was never materialized or the materialization record is too old to contain provenance information.

Return type:

Optional[DataProvenance]

get_mapping_key()[source]¶: Which mapping_key this execution is for if downstream of a DynamicOutput, otherwise None.

get_tag(key)[source]¶

Get a logging tag.

Parameters:: key (tag) – The tag to get.
Returns:: The value of the tag, if present.
Return type:: Optional[str]

property has_assets_def¶: If there is a backing AssetsDefinition for what is currently executing.

property has_partition_key¶: Whether the current run is a partitioned run.

has_tag(key)[source]¶

Check if a logging tag is set.

Parameters:: key (str) – The tag to check.
Returns:: Whether the tag is set.
Return type:: bool

property instance¶

The current Dagster instance.

Type:: DagsterInstance

property job_def¶

The currently executing pipeline.

Type:: JobDefinition

property job_name¶

The name of the currently executing pipeline.

Type:: str

property log¶

The log manager available in the execution context.

Type:: DagsterLogManager

log_event(event)[source]¶

Log an AssetMaterialization, AssetObservation, or ExpectationResult from within the body of an op.

Events logged with this method will appear in the list of DagsterEvents, as well as the event log.

Parameters:: event (Union[AssetMaterialization, AssetObservation, ExpectationResult]) – The event to log.

Examples:

from dagster import op, AssetMaterialization

@op
def log_materialization(context):
    context.log_event(AssetMaterialization("foo"))

property op_config¶

The parsed config specific to this op.

Type:: Any

property op_def¶

The current op definition.

Type:: OpDefinition

output_for_asset_key(asset_key)[source]¶: Return the output name for the corresponding asset key.

property partition_key¶

The partition key for the current run.

Raises an error if the current run is not a partitioned run.

property partition_key_range¶

The range of partition keys for the current run.

If run is for a single partition key, return a PartitionKeyRange with the same start and end. Raises an error if the current run is not a partitioned run.

property partition_time_window¶

The partition time window for the current run.

Raises an error if the current run is not a partitioned run, or if the job’s partition definition is not a TimeWindowPartitionsDefinition.

property pdb¶

Gives access to pdb debugging from within the op.

Example

@op
def debug(context):
    context.pdb.set_trace()

Type:: dagster.utils.forked_pdb.ForkedPdb

property resources¶

The currently available resources.

Type:: Resources

property retry_number¶: Which retry attempt is currently executing i.e. 0 for initial attempt, 1 for first retry, etc.

property run_config¶

The run config for the current execution.

Type:: dict

property run_id¶

The id of the current execution’s run.

Type:: str

property selected_asset_keys¶: Get the set of AssetKeys this execution is expected to materialize.

property selected_output_names¶: Get the output names that correspond to the current selection of assets this execution is expected to materialize.

dagster.AssetExecutionContext¶: alias of OpExecutionContext

dagster.build_op_context(resources=None, op_config=None, resources_config=None, instance=None, config=None, partition_key=None, partition_key_range=None, mapping_key=None, _assets_def=None)[source]¶

Builds op execution context from provided parameters.

build_op_context can be used as either a function or context manager. If there is a provided resource that is a context manager, then build_op_context must be used as a context manager. This function can be used to provide the context argument when directly invoking a op.

Parameters:

resources (Optional[Dict[str, Any]]) – The resources to provide to the context. These can be either values or resource definitions.
op_config (Optional[Mapping[str, Any]]) – The config to provide to the op.
resources_config (Optional[Mapping[str, Any]]) – The config to provide to the resources.
instance (Optional[DagsterInstance]) – The dagster instance configured for the context. Defaults to DagsterInstance.ephemeral().
mapping_key (Optional[str]) – A key representing the mapping key from an upstream dynamic output. Can be accessed using context.get_mapping_key().
partition_key (Optional[str]) – String value representing partition key to execute with.
partition_key_range (Optional[PartitionKeyRange]) – Partition key range to execute with.
_assets_def (Optional[AssetsDefinition]) – Internal argument that populates the op’s assets definition, not meant to be populated by users.

Examples

context = build_op_context()
op_to_invoke(context)

with build_op_context(resources={"foo": context_manager_resource}) as context:
    op_to_invoke(context)

dagster.build_asset_context(resources=None, resources_config=None, asset_config=None, instance=None, partition_key=None, partition_key_range=None)[source]¶

Builds asset execution context from provided parameters.

build_asset_context can be used as either a function or context manager. If there is a provided resource that is a context manager, then build_asset_context must be used as a context manager. This function can be used to provide the context argument when directly invoking an asset.

Parameters:

resources (Optional[Dict[str, Any]]) – The resources to provide to the context. These can be either values or resource definitions.
resources_config (Optional[Mapping[str, Any]]) – The config to provide to the resources.
asset_config (Optional[Mapping[str, Any]]) – The config to provide to the asset.
instance (Optional[DagsterInstance]) – The dagster instance configured for the context. Defaults to DagsterInstance.ephemeral().
partition_key (Optional[str]) – String value representing partition key to execute with.
partition_key_range (Optional[PartitionKeyRange]) – Partition key range to execute with.

Examples

context = build_asset_context()
asset_to_invoke(context)

with build_asset_context(resources={"foo": context_manager_resource}) as context:
    asset_to_invoke(context)

class dagster.TypeCheckContext(run_id, log_manager, scoped_resources_builder, dagster_type)[source]¶

The context object available to a type check function on a DagsterType.

property log¶: Centralized log dispatch from user code.

property resources¶: An object whose attributes contain the resources available to this op.

property run_id¶: The id of this job run.

Job configuration¶

dagster.validate_run_config(job_def, run_config=None)[source]¶

Function to validate a provided run config blob against a given job.

If validation is successful, this function will return a dictionary representation of the validated config actually used during execution.

Parameters:

job_def (JobDefinition) – The job definition to validate run config against
run_config (Optional[Dict[str, Any]]) – The run config to validate

Returns:

A dictionary representation of the validated config.

Return type:

Dict[str, Any]

Run Config Schema¶

The run_config used for jobs has the following schema:

{
  # configuration for execution, required if executors require config
  execution: {
    # the name of one, and only one available executor, typically 'in_process' or 'multiprocess'
    __executor_name__: {
      # executor-specific config, if required or permitted
      config: {
        ...
      }
    }
  },

  # configuration for loggers, required if loggers require config
  loggers: {
    # the name of an available logger
    __logger_name__: {
      # logger-specific config, if required or permitted
      config: {
        ...
      }
    },
    ...
  },

  # configuration for resources, required if resources require config
  resources: {
    # the name of a resource
    __resource_name__: {
      # resource-specific config, if required or permitted
      config: {
        ...
      }
    },
    ...
  },

  # configuration for underlying ops, required if ops require config
  ops: {

    # these keys align with the names of the ops, or their alias in this job
    __op_name__: {

      # pass any data that was defined via config_field
      config: ...,

      # configurably specify input values, keyed by input name
      inputs: {
        __input_name__: {
          # if an dagster_type_loader is specified, that schema must be satisfied here;
          # scalar, built-in types will generally allow their values to be specified directly:
          value: ...
        }
      },

    }
  },

}

You are viewing an outdated version of the documentation.

Execution¶

Materializing Assets¶

Executing Jobs¶

Executing Graphs¶

Execution results¶

Reconstructable jobs¶

Executors¶

Contexts¶

Job configuration¶

Run Config Schema¶