root_cause_category: replication_lag
required_keywords:
  - replication lag
  - write-heavy workload
  - replica
  - wal
  - replay
required_evidence_sources:
  - aws_cloudwatch_metrics
  - aws_performance_insights
equivalent_root_cause_categories:
  - replication_lag_wal_volume
forbidden_categories:
  - cpu_saturation
  - connection_exhaustion
optimal_trajectory:
  - query_grafana_metrics
  - query_grafana_logs
  - query_grafana_alert_rules
  - describe_rds_instance
  - describe_rds_events
max_investigation_loops: 4
model_response: |
  ROOT_CAUSE: The database hit replication lag because a write-heavy workload on the primary generated WAL faster than the read replica could replay it. The symptom is stale replica reads rather than a primary availability outage.
  ROOT_CAUSE_CATEGORY: replication_lag

  VALIDATED_CLAIMS:
  - ReplicaLag stayed above 900 seconds on payments-prod-replica-1. [evidence: aws_cloudwatch_metrics]
  - WriteIOPS and transaction log generation spiked on the primary during the same window. [evidence: aws_cloudwatch_metrics]
  - Performance Insights shows bulk UPDATE statements dominated DB load with WAL write waits. [evidence: aws_performance_insights]

  NON_VALIDATED_CLAIMS:
  - Reducing batch size or adding replica capacity would likely reduce the lag more quickly.

  CAUSAL_CHAIN:
  - A write-heavy workload drove sustained bulk updates on the primary.
  - WAL generation increased faster than the replica could replay.
  - Replication lag accumulated on payments-prod-replica-1 and stale reads triggered the alert.