Skip to main content

​
current

The current object provides access to runtime information about the currently executing flow, run, step, and task.

​
Usage

from metaflow import FlowSpec, step, current

class MyFlow(FlowSpec):
    @step
    def start(self):
        print(f"Flow name: {current.flow_name}")
        print(f"Run ID: {current.run_id}")
        print(f"Step name: {current.step_name}")
        print(f"Task ID: {current.task_id}")
        self.next(self.end)
    
    @step
    def end(self):
        pass

​
Properties

​
is_running_flow

current.is_running_flow -> bool
Returns True if called inside a running Flow, False otherwise. You can use this property e.g. inside a library to choose the desired behavior depending on the execution context. Returns:
  • bool - True if called inside a run, False otherwise.

​
flow_name

current.flow_name -> Optional[str]
The name of the currently executing flow. Returns:
  • str | None - Flow name.

​
run_id

current.run_id -> Optional[str]
The run ID of the currently executing run. Returns:
  • str | None - Run ID.

​
step_name

current.step_name -> Optional[str]
The name of the currently executing step. Returns:
  • str | None - Step name.

​
task_id

current.task_id -> Optional[str]
The task ID of the currently executing task. Returns:
  • str | None - Task ID.

​
retry_count

current.retry_count -> int
The index of the task execution attempt. This property returns 0 for the first attempt to execute the task. If the @retry decorator is used and the first attempt fails, this property returns the number of times the task was attempted prior to the current attempt. Returns:
  • int - The retry count.

​
origin_run_id

current.origin_run_id -> Optional[str]
The run ID of the original run this run was resumed from. This property returns None for ordinary runs. If the run was started by the resume command, the property returns the ID of the original run. You can use this property to detect if the run is resumed or not. Returns:
  • str | None - Run ID of the original run.

​
pathspec

current.pathspec -> Optional[str]
Pathspec of the current task, i.e. a unique identifier of the current task. The returned string follows this format: 
{flow_name}/{run_id}/{step_name}/{task_id}
This is a shorthand to current.task.pathspec. Returns:
  • str | None - Pathspec.

​
task

current.task -> Optional[Task]
Task object of the current task. Returns:
  • Task | None - Current task.

​
run

current.run -> Optional[Run]
Run object of the current run. Returns:
  • Run | None - Current run.

​
namespace

current.namespace -> str
The current namespace. Returns:
  • str - Namespace.

​
username

current.username -> Optional[str]
The name of the user who started the run, if available. Returns:
  • str | None - User name.

​
tempdir

current.tempdir -> Optional[str]
Currently configured temporary directory. Returns:
  • str | None - Temporary directory.

​
graph

current.graph
Graph information for the current flow. This property provides access to the flow’s graph structure including steps and their relationships.

​
Examples

​
Accessing runtime information

from metaflow import FlowSpec, step, current

class MyFlow(FlowSpec):
    @step
    def start(self):
        print(f"Running {current.flow_name}/{current.run_id}")
        print(f"Current step: {current.step_name}")
        print(f"Task ID: {current.task_id}")
        self.next(self.end)
    
    @step
    def end(self):
        pass

​
Detecting resumed runs

from metaflow import FlowSpec, step, current

class MyFlow(FlowSpec):
    @step
    def start(self):
        if current.origin_run_id:
            print(f"This run was resumed from {current.origin_run_id}")
        else:
            print("This is a fresh run")
        self.next(self.end)
    
    @step
    def end(self):
        pass

​
Using in library code

from metaflow import current

def my_library_function():
    if current.is_running_flow:
        # Running inside Metaflow
        print(f"Logging to Metaflow task {current.pathspec}")
    else:
        # Running outside Metaflow
        print("Logging to stdout")

Build docs developers (and LLMs) love

Get started for freeTalk to us