Indices

Index Configuration (Global Defaults)

Indices control how the local database indexes transcribed data for efficient querying. Global indices defined here apply to all transcribes as defaults. Transcribe-level indices (at transcribes.<type>.indices.*) take precedence when both are defined.

All three index types can be defined at the root level:

• value-match: For equality (eq, ne, in) and starts-with (sw) queries
• partial-match: For contains, ends-with, and fuzzy queries (expensive)
• sortable: For sort and range (lt, gt, le, ge) queries

Starts-with queries do not need partial-match — value-match covers them. Reach for partial-match only when callers actually search by contains, ends-with, or fuzzy match.

Attributes listed in any indices.* block of a transcribe are automatically fetched and stored — they do not also need to be listed in ldap.attributes. For relationship fields, choose the stored side before adding coverage. See Attribute Modeling.

Example:

indices {
  partial-match = "cn, sn, description"
  value-match = "uid, mail, employeeNumber"
  sortable = "cn, displayName"
}

Attributes for contains, ends-with, and fuzzy queries. Creates trigram indexes for substring matching. Trigram indexes are typically 5–15× larger than equality indexes and slow writes proportionally — enable only for attributes that need this coverage. Starts-with (sw) queries do NOT require partial-match; configure those attributes under value-match instead. No baseline defaults — substring/typeahead is always explicit. Common "typeahead pack" for person directories:

indices.partial-match = "cn, sn, givenName, displayName"

Priority: SCRIBE_INDICES_PARTIAL_MATCH > config

indices.partial-match = null
indices.partial-match = ${?SCRIBE_INDICES_PARTIAL_MATCH}

Attributes for sort and range (≥, ≤) queries. Baseline defaults are automatically included for observed attributes: cn, sn, displayName, mail, uid. Covers paged-list sort keys and identity lookups. Group-membership attributes (memberOf, nrfMemberOf) are equality-only via value-match; add them here only if you actually sort by memberOf or range-scan group DNs. For relationship direction guidance, see Attribute Modeling.

Example: "cn, displayName, memberOf"

Priority: SCRIBE_INDICES_SORTABLE > config

indices.sortable = null
indices.sortable = ${?SCRIBE_INDICES_SORTABLE}

Attributes for equality (eq, ne, in) and starts-with (sw) queries. Baseline defaults are automatically included for observed attributes: objectClass, structuralObjectClass, cn, sn, ou, uid, mail, displayName, member, memberOf, nrfMemberOf, nrfAssignedRoles, manager. Add custom equality filters here. Choose the stored relationship side before adding coverage: Attribute Modeling.

Example: "cn, uid, mail, nrfMemberOf"

Priority: SCRIBE_INDICES_VALUE_MATCH > config

indices.value-match = null
indices.value-match = ${?SCRIBE_INDICES_VALUE_MATCH}