client.workflow_run

The workflow run is an instance of a workflow.

Attributes

logger

Classes

`WorkflowRunFactory`	A utility class responsible for instantiating workflow run objects.
`WorkflowRun`	WorkflowRun enables tracking for multiple runs of a single Workflow.

Module Contents

client.workflow_run.logger

class client.workflow_run.WorkflowRunFactory(workflow_id: int, requester: jobmon.core.requester.Requester | None = None)

A utility class responsible for instantiating workflow run objects.

This class sends the appropriate resume signals so that the parent workflow object is in a state where the newly created workflowrun is ready to run, either on resume or not.

Initialization of client WorkflowRun.

workflow_id

requester = None

workflow_is_resumable = False

set_workflow_resume(reset_running_jobs: bool = True, resume_timeout: int = 300, force_cleanup: bool = False) → None

Set statuses of the given workflow ID’s workflow to a resumable state.

Move active workflow runs to hot/cold resume states, depending on reset_running_jobs.

Parameters:

reset_running_jobs – If True (cold resume), also terminate running tasks. If False (hot resume), let running tasks finish.
resume_timeout – Maximum time to wait for workflow to become resumable.
force_cleanup – If True and timeout is reached with pending KILL_SELF task instances, force cleanup and proceed. Use this when jobs have been externally terminated (scancel, node failure).

wait_for_workflow_resume(resume_timeout: int = 300, force_cleanup: bool = False) → None

Wait for workflow to become resumable.

Parameters:

resume_timeout – Maximum time to wait in seconds
force_cleanup – If True and KILL_SELF task instances are detected, clean them up immediately instead of waiting for workers

force_cleanup_kill_self() → int

Force cleanup of stuck KILL_SELF task instances.

Use this when jobs have been externally terminated (e.g., scancel, node failure) and the workflow is stuck waiting for cleanup that will never happen.

Returns:: Number of task instances cleaned up

reset_task_statuses(reset_if_running: bool = True, force_cleanup: bool = False) → None

Sets the tasks associated with a workflow to the appropriate states.

Parameters:

reset_if_running – If True, also reset running tasks.
force_cleanup – If True and workflow not resumable due to stuck KILL_SELF task instances, force cleanup and proceed.

create_workflow_run(workflow_run_heartbeat_interval: int | None = None, heartbeat_report_by_buffer: float | None = None) → WorkflowRun: Workflow should at least have signalled for a resume at this point.

class client.workflow_run.WorkflowRun(workflow_id: int, requester: jobmon.core.requester.Requester | None = None, workflow_run_heartbeat_interval: int | None = None, heartbeat_report_by_buffer: float | None = None)

Bases: object

WorkflowRun enables tracking for multiple runs of a single Workflow.

A Workflow may be started/paused/ and resumed multiple times. Each start or resume represents a new WorkflowRun.

In order for a Workflow can be deemed to be DONE (successfully), it must have 1 or more WorkflowRuns. In the current implementation, a Workflow Job may belong to one or more WorkflowRuns, but once the Job reaches a DONE state, it will no longer be added to a subsequent WorkflowRun. However, this is not enforced via any database constraints.

Initialize client WorkflowRun.

workflow_id

user

requester = None

heartbeat_interval

heartbeat_report_by_buffer

property workflow_run_id: int

property status: str | None

bind() → None: Link this workflow run with the workflow and add all tasks.