Troubleshoot and Accept Scout

Start with the first incomplete step in the current task. Refresh the task once, inspect the latest operation and record the visible reason, time and request reference before retrying a write.

Diagnose by Symptom

Symptom	Next action	Owner
Scout is missing or access is denied	Check tenant, DFS Pro and Scout activation, then action permissions	Administrator
Business goals are unavailable	Use exploration for discovery, or publish the required business requirements	Application owner
Expected equipment is missing	Reconcile the existing MDM or asset identity and review queue	Data steward
Semantic models are unavailable	Create or import the intended model in model management	Model owner
A model has no linked points or equipment	Associate its nodes with existing equipment and attributes	Model owner
Matching candidates are absent	Check discovery completion, source selection, scope and result pages	Data operator
Unit, axis, temperature or battery meaning is unclear	Obtain a field dictionary and installation evidence	Domain specialist
Compatible Point classes are absent	Check the property unit, selected catalog and complete result list	Model administrator
AI analysis is unavailable	Test the configured model route with an actual Scout analysis	Model-service owner
Approval action is unavailable	Complete required evidence and verify independent release roles	Reviewer or administrator
Model or binding version changed	Refresh the task and validate a plan for the current version	Operator
CSV preview is ready without imported values	Assign the governed import to the data-integration owner	Data-integration owner
MQTT is connected without fields	Check topic, publisher, payload format and parser support	Source owner
Publication succeeded without readings	Check source access, observation time and ingestion path	Data operations
Readings exist without a business result	Check application inputs, processing and output	Application owner

Treat missing data, currently unavailable data and a valid measured zero as separate states. Record the source, observation time and unit whenever a real value is available.

Recover a Timed-Out or Partial Operation

1. Refresh the task and open the existing operation or release.
2. Inspect completed, failed and pending stages.
3. Preserve the request or release reference for support and audit.
4. Use the recovery action presented for that workflow.
5. Verify the resulting active version and downstream state.

For an API retry, reuse the original idempotency key only when the endpoint documents that retry behavior. Semantic-model recovery can restore a supported model or reference version; raw-data recovery follows the data workflow's own controls.

Customer Acceptance Checklist

• The intended tenant and delivered frontend/backend versions are in use.
• Viewer, operator, reviewer, approver and release-operator roles work as assigned.
• Authorized customer sources and real equipment identities are confirmed.
• Scope, units, physical meaning and time boundaries are agreed.
• The approved model service returns a usable Scout analysis when AI analysis is in scope.
• Decisions, validation, approval and release results are recorded.
• The workflow produces its expected dataset version, model references, observations or application result.
• Source freshness and historical coverage are reported separately.
• Remaining items have owners, required evidence and target follow-up times.
• Recovery limits and temporary-data cleanup responsibilities are documented.

Prepare a Support Request

Include the task name, tenant context through the approved support channel, page and action, time, visible error or request reference, affected scope, product version and last successful step. Follow the customer's support policy for records and attachments. Keep passwords, API secrets and model credentials in the approved secret-management system.

For related setup, see Administration, Private Deployment and Brick Models.