Logo
NeoArc Studio

Search Drift Detection

Compare design-time search schemas against deployed index schemas to identify field removals, additions, type changes, and analyser mismatches.

Search drift detection compares the search schema defined in NeoArc Studio's graph model against the actual schema deployed to your search engine. The SearchDriftDetectionService identifies fields that have been removed, added, or changed between design-time and deployment, producing a structured drift report.

How Drift Detection Works

Design-Time Schema Export

The exportDesignSchema method generates the expected index schema from the graph model and active search profile.

interface DesignTimeSchema {
  indexName: string;       // Resolved from entity name via profile's indexNamingConvention
  entityName: string;      // Source entity name
  engine: string;          // Search engine vendor
  fields: DesignTimeField[];
}

Design-Time Field Structure

Each field in the design-time schema contains the following properties.

interface DesignTimeField {
  fieldName: string;        // Resolved field name (after casing transformation)
  canonicalName: string;    // Original canonical property name
  canonicalType: string;    // Canonical type from the model
  resolvedType: string;     // Engine-specific type (via profile type mapping)
  searchFieldType?: string; // Search field type override (if set)
  isKey: boolean;           // Whether this is the document key
  behaviours: string[];     // Search behaviours
  analyzer?: string;        // Analyser
}

Field Name Resolution

When resolving a field's name in the search index, the following priority applies.

PrioritySourceDescription
1 (highest)Explicit fieldName overrideThe fieldName value set directly on the search projection takes priority over all other resolution
2Profile naming conventionThe profile's namingConvention is applied to the canonical property name (e.g. camelCase, snake_case)

Drift Item Types

The drift detection engine produces four types of drift items, each with a severity level that indicates the impact on your deployed index.

TypeSeverityDescription
field-removedErrorField exists in the design-time schema but is missing from the deployed index
field-addedInfoField exists in the deployed index but is not present in the design-time schema
type-changedWarningField type differs between the design-time schema and the deployed index
analyzer-changedWarningField analyser differs between the design-time schema and the deployed index

Drift Report Structure

interface DriftReport {
  indexName: string;         // Entity/index name
  totalDrifts: number;       // Total number of drift items
  errors: DriftItem[];       // Breaking changes (field-removed)
  warnings: DriftItem[];     // Potential issues (type-changed, analyzer-changed)
  infos: DriftItem[];        // Informational items (field-added)
  items: DriftItem[];        // All drift items combined
  isClean: boolean;          // True when no drift exists
}

Deployed Field Input

The deployed schema is provided as an array of DeployedField objects, normalised from the engine's native mapping format.

interface DeployedField {
  fieldName: string;   // Field name in the deployed index
  type: string;        // Engine-specific type
  analyzer?: string;   // Analyser name
}

Interpreting Results