At 2106Z on 12/18/17 a AWS ElastiCache node failed which triggered a failover to the secondary node. A previous configuration error had caused incorrect settings to be left on the secondary node. These settings broke the rate limiting logic in our indexing system when the secondary ElastiCache node became the primary. We applied a fix at 2130Z that allowed the indexers to resume syncing Close.io data. All backlogged data was processed 11 minutes later.