12.2.272 Platform Release

Released October 14, 2026

This release is a patch update for the 12.2 platform release.

Addressed issues

• Automation flows: We've improved automation flow reliability by preventing a duplicate scheduled trigger on an in-progress flow step from terminating the entire flow when the original step was healthy and completed successfully.

• BQL: We've improved result accuracy for BQL queries that combine GROUP BY with an aggregate over a related-entity traversal, so counts now reflect your full data.

• Consolidation: We've improved consolidation performance by moving certain read operations to a dedicated connection pool, reducing overhead on primary database connections.

• Consolidation: We've enhanced consolidation reliability with an improvement to how retry operations are processed, helping consolidation run smoothly without interruption.

• Consolidation: We've improved consolidation reliability by making sure every queued work item reliably records its completion, so consolidation always moves forward instead of waiting on unfinished rows.

• Data integration: We've improved sync reliability by resolving a database connection-pool exhaustion issue that could occur during data integration.

• Data integration: We've improved data integration sync performance by skipping full-row writes for source rows that haven't changed.

• Performance: We've improved memory efficiency for BQL cluster-condition queries, preventing excessive database memory usage that could affect shared infrastructure during large relationship deletes.

• Platform: We've improved platform stability by making shared database connection pools safe to use with transaction auto-commit disabled, preventing compute and data-integration operations from failing or silently not completing.

• Platform: We've enhanced compute reliability so records with an invalid stored risk score now recompute automatically instead of repeatedly retrying.

• Tickets: We've improved ticketing automation reliability by changing how changed tickets are detected for incremental reconcile, preventing a scheduled job from running out of time before completing.