How a Failed Rebalance Revealed 8.6 TB of Orphan Databases on a Production SingleStore Cluster
If you run a self-hosted SingleStore cluster in production, there is a specific flavor of problem you hope you never see: the database is serving reads and writes,