# LogReducer **Type**: `object` ## Properties - name: attributeMergeStrategyEntries (optional) type: array of `AttributesMergeStrategyEntry` description: List of attribute paths and their corresponding merge strategies. If no strategy is defined for a path, the terminal merge strategy will be used. - name: attributeTokenization (optional) type: enum: `KEYS_ONLY`, `KEYS_AND_VALUES` description: Use attribute shape only, or shape and whole-value similarity tokens. - name: dedupThreshold (optional) type: integer description: Number of duplicates we send for a pattern before we start deduplication. - name: defaultAttributesMergeStrategy (optional) type: `AttributesMergeStrategy` - name: delimiters (required) type: array of string description: Characters to use for splitting tokens. Default is missing '@' to avoid splitting email addresses. Also missing '.' to avoid splitting identifiers using it. - name: emitNewPatternSignals (optional) type: boolean description: Whether this operation may emit LogPatternSignal events to the agent-service when new patterns are detected or replayed. Must be enabled before an agent can subscribe to this pipeline. - name: enabledMasks (required) type: array of string description: Masks that are enabled for tokenizing messages. These are specified as the names of the masks to use. The default enabled masks are: ipport, timestamp, uuid, number. - name: excludedAttributePrefixes (optional) type: array of array of string description: Excluded attribute prefixes as arrays of literal key segments. Does not affect partition-by values or output attributes. - name: includedAttributePrefixes (optional) type: array of array of string description: Included attribute prefixes as arrays of literal key segments. Empty selects all. Array elements are traversed without indices. - name: integrationExceptionConfigs (optional) type: array of `IntegrationExceptionConfig` description: Exception configs that define which exceptions to use from the ones imported from the observability vendor integrations. - name: logReducerExceptions (optional) type: array of `EventPredicate` description: Predicate for skipping events. Logs matching these predicates won't be aggregated. - name: masks (required) type: array of array of string description: Set of masks that can be applied to messages before tokenizing them. These masks are specified as pairs of (name, regex). The default masks are: ipport, timestamp, uuid, number, awsarn, awstoken. - name: name (required) type: string - name: partitionByAttributePaths (optional) type: array of array of string description: Attribute paths as arrays of keys. Use for unambiguous nested vs flat attributes. Example: [["user", "id"]] for nested user.id, [["user.id"]] for flat attribute named 'user.id' - name: partitionByAttributes (optional) type: array of string - name: partitionByTags (optional) type: array of string - name: reductionTimeWindow (optional) type: string description: Time window in which logs are considered for deduplication. Should not be modified without explicit guidance. - name: repeatCountInSummaryMessageEnabled (optional) type: boolean description: Whether to include the repeat count prefix in the summary message. Grepr will add a prefix to the summary message with the number messages aggregated. However, this might break downstream parsing if the message is expected to be in a specific format. If you want to include the repeat count, set this to true. - name: samplingConfig (optional) type: `SamplingConfig` - name: similarityThreshold (optional) type: number description: Controls how close a set of tokens are to a pattern before it is considered a match. Calculated as the percent of matching tokens between a pattern and a message. - name: tokenizationInput (optional) type: enum: `MESSAGE_ONLY`, `ATTRIBUTES_ONLY`, `MESSAGE_AND_ATTRIBUTES`, `AUTO` description: Selects message text, attributes, both, or message text with an attributes fallback. - name: type (required) type: enum: `log-reducer` description: Reduces logs by combining messages with similar patterns.