dFlow~IQ manualChapter 06 · Application guide

Run bounded manual pipelines

Define a same-cluster raw-copy pipeline, dry-run a finite snapshot and confirm the exact target before producing anything.

DOCUMENTED VERSIONdFlow~IQ Community 1.0

First-party product documentation. Instructions are verified against the dFlow~IQ desktop implementation. Labels appear exactly as they do in the application where practical.

06.1

Define the pipeline

Pipeline name

Unique within the workspace, case-insensitively.

Source topic

Topic read with read_committed isolation.

Target topic

Must differ from the source.

Source position

Earliest retained or Latest retained tail.

Maximum records

Between 1 and 10,000.

Keys, values, headers and original timestamps are copied as raw bytes. Saved definitions never run automatically.

06.2

Create a dry-run plan

  1. Save the definition.
  2. Select Dry run.
  3. Wait while dFlow~IQ reads a bounded snapshot.
  4. Review record count, partition count, snapshot bytes and plan ID.

The snapshot is limited to 32 MiB and a 30-second broker-read deadline. A plan with no records cannot run.

06.3

Confirm and execute

  1. Select Run pipeline after a successful dry run.
  2. Read the at-least-once warning.
  3. Type the exact target topic into the confirmation field.
  4. Select Run pipeline and monitor broker-acknowledged progress.

Every run requires a fresh dry run. Cancellation, failure or editing invalidates the plan. An interruption can leave an in-flight result unknown, and retrying may create duplicates.

Documentation reference

Current dFlow~IQ Community desktop implementation and bundled UI labels, reviewed 13 August 2026. Product behaviour takes precedence if a later release differs from this page.