Skip to content

Provide Replay Support for a Single Job #739

@arinda-arif

Description

@arinda-arif

Description
User should be capable to replay for a single job in a sequential or parallel way for a single or multiple runs over a period of time.

Acceptance Criteria

  • Validation Failure in case a job run is in progress/scheduled/running or another replay request for the same.
  • Sequential & Parallel Execution of Replay to be provided to the user.
  • Overriding of Configurations which will should be used in job run input.

Out of Scope

  1. Status & List
  2. Dry run
  3. Auth
  4. Replay downstream

Tech Details

Replay CLI

  1. The user should be able to give the reason for backfill for audit purposes.
  2. Will check if the existing dag runs are in the running state or not. If it is, Replay is not able to run.
  3. Will check if there is any currently running replay run for the same dag. If it is on different runs (dates), then Replay is able to run.
  4. Will clear dag runs through Airflow API. Clearing can be done sequentially (default) or parallel.
    • Sequential: Wait for the previous run to succeed.
    • Parallel: Clear at once.

Impacted API: Job run input
Optimus will try to check whether there is a Replay request for the exact same job and date. If there is, Optimus will try to add (or replace if exists) the additional configs from Replay request. For example, the EXECUTION_PROJECT from Replay will replace the one specified in the job specification.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions