Managing Source lifecycle
Knowledge Hub is an Early Access release. Refer to the Early Access disclaimer for details.
When a Data Integration flow deploys with Knowledge Hub as the target, Knowledge Hub creates a Source record automatically. You can start that flow from Data Integration directly, or select Create Source on the Knowledge Base's Sources tab to open Data Integration with the Repository and Knowledge Base already selected. Refer to Configuring Knowledge Hub as a target for both options. Once the Source exists, you manage it entirely from the Sources tab inside the Knowledge Base. No return to Data Integration is needed unless you want to change the connector or flow configuration.
Actions available on a Source
Select a Source on the Sources tab to open its detail page. The header shows the actions available for its current state, plus a Delete icon that is always available:
| Action | Available when | What it does |
|---|---|---|
| Enable | Draft, Disabled | Starts or resumes ingestion on the configured schedule. |
| Run now | Enabled | Triggers an immediate ingestion run outside the schedule. |
| Disable | Enabled | Pauses scheduled ingestion. Existing data is retained. |
| Delete | Any | Permanently deletes the Source and all its indexed data. |
Source settings (schema mapping, match key, query fields, chunking, and loading mode) are editable directly on the detail page regardless of state. Refer to Editing Source configuration below.
Source states
| State | Description |
|---|---|
| Draft | Source created after flow deploy. No ingestion has run. Enabling is required before the first run. |
| Enabled | Source is active. Ingestion runs on the configured schedule. |
| Disabled | Source is paused. Scheduled ingestion does not run. Existing chunks are retained. |
Archiving is not currently supported for Sources. To permanently remove a Source, use Remove.
Enabling a Source
A Source starts in Draft state after the flow deploys. You must enable it to start ingestion.
- Open Knowledge Hub, open the Knowledge Base, select the Sources tab, then select the Source.
- Select Enable in the top-right corner.
- Confirm the action in the dialog.
The Source state changes to Enabled. The first scheduled ingestion run starts at the next scheduled interval. To run ingestion immediately, use Triggering a manual ingestion run below.
Triggering a manual ingestion run
You can trigger an ingestion run outside the schedule at any time when the Source is in Enabled state.
- Open Knowledge Hub, open the Knowledge Base, select the Sources tab, then select the Source.
- Select Run now in the top-right corner.
The run starts immediately. Refer to Viewing ingestion activity to monitor its progress.
Editing Source configuration
Open the Source to review and edit its configuration. Changes apply on the next ingestion run; re-ingestion is triggered manually. In-progress runs use the configuration that was active at run start.
Basic Information
- Knowledge Source Title: editable name for the Source.
- Source Type: read-only, set by the connector used in the Data Integration flow (for example, S3).
- Data Provider: a link to the corresponding flow in Boomi Data Integration.
- Last Ingestion and Ingestion Error Summary: read-only run history.
Target Schema
Map source fields to Knowledge Base schema attributes. Each mapping has an Expression (the source field), a Knowledge Base Attribute, and a Data Type. All mapped attributes are automatically available for SELECT, filter, and BM25 full-text search. Select + Direct Mapping to add a mapping.
Match Key
Select one or more mapped attributes as a composite match key for upsert-merge ingestion behavior. Order matters: Knowledge Hub matches on the fields in the sequence listed. Match key configuration applies only when the Source uses upsert-merge loading.
Query Fields
Select which mapped attributes Knowledge Hub queries at retrieval time. Order determines the concatenation sequence used when building the query text.
Chunking Configuration
Chunking settings are optimized for the Knowledge Base's embedding model.
| Setting | Description |
|---|---|
| Strategy Type | The chunking strategy. For example, Structural preserves semantic boundaries such as paragraphs and sentences. |
| Maximum Size | Target character count for each chunk. Larger sizes capture more context but may reduce query precision. |
| Overlap | Percentage of the previous chunk repeated at the start of the next chunk, for context continuity. |
Not editable
- Source connector type
- Target Knowledge Base
- Embedding model (controlled at the Knowledge Base level)
To change the source connector or target Knowledge Base, you must create a new S2T flow in Data Integration and deploy it. Refer to Configuring Knowledge Hub as a target.
Select Save and Enable to save your changes and enable the Source in one step, or Save to save without changing its state.
Changes to schema mapping, match key, query fields, or chunking settings do not retroactively update existing chunks. Existing embeddings remain valid under the previous configuration. Re-ingestion is triggered manually in Early Access; automatic re-ingestion on configuration change is planned for a future release.
Disabling a Source
Disabling pauses scheduled ingestion without removing existing data.
- Open Knowledge Hub, open the Knowledge Base, select the Sources tab, then select the Source.
- Select Disable in the top-right corner.
- Confirm the action in the dialog.
To re-enable the Source, select Enable.
Removing a Source
Removing a Source deletes the Source record and all chunks indexed from that Source. This action cannot be undone.
Removing a Source permanently deletes the vector data it contributed to the Knowledge Base. Other Sources and their data are not affected.
- Open Knowledge Hub, open the Knowledge Base, select the Sources tab, then select the Source.
- Select the trash can icon in the top-right corner.
- Confirm the deletion in the dialog that appears.
