Hello all, I am facing an issue where a specific datasource (XYZ) becomes queryable only intermittently and frequently drops into an "Inactive/Unused" state. The Issue: The datasource appears in the "Unused" list in the console. When clicked on datasource and , I receive the following error: Plan validation failed / org.apache.calcite.runtime.CalciteContextException: Object 'XYZ' not found / org.apache.calcite.tools.ValidationException Current Workaround: If I manually "Mark segments as used" in the console, the table becomes visible and queryable again. However, after some time, it reverts to the "Unused" state automatically. Key Context & Environment: Ingestion: Data is ingested via manual batch processes. The most recent ingestion was several weeks ago, and no manual ingestion was running when the segments became unused. Compaction: Auto-compaction is enabled for this datasource with skipOffsetFromLatest: PT0S. Retention: There are no active drop rules or retention policies configured. Deep Storage: S3 Druid Version: 25.0.0 Requesting Help with: Since no manual ingestion was active, could Auto-Compaction with a PT0S offset be marking segments as unused but failing to successfully publish/load the new compacted versions? Are there specific Coordinator or Overlord log signatures I should search for to see why these segments are being deactivated? Are there known issues where PT0S offset causes race conditions that lead to all segments being marked as unused? Any guidance on which logs to check or common pitfalls with this setup would be greatly appreciated!