Hi all, Here’s the update about our work at Timecho on IoTDB in the past 2 weeks, which were mainly issue fixes:
Issue Fixes *-*Fixed array index out of bounds in tree mode during cross-region aggregation with ALIGN BY DEVICE queries. -Fixed precise memory estimation issue with BinaryColumn objects. -Fixed errors when executing cross-region, multi-device aggregation queries with ORDER BY TIME + ALIGN BY DEVICE. -Fixed issue where regions on mistakenly removed DataNodes (DN) required manual cleanup. -Fixed issue with default replica number for root.__system not following user-defined parameters. -Fixed non-zero CachedRequest count after IoTV2 sync completion. -Fixed thread leak in IoTV2's TsFileChecker. -Fixed several issues related to IoTV2 behavior during cluster scale-in. -Fixed array index overflow during load balancing after scale-in. -Fixed WAL stuck issue when a DN exits but its WAL logs aren’t deleted. -Fixed issue in IoTV2 where Follower Load TsFile did not skip updating the LastCache. -Fixed write failures caused by insufficient memory in Pipe during ingestion. -Fixed failure to manually drop subscriptions under table mode. -Fixed metadata sync issue where metadata was still synced after user password change in sink under table mode. -Fixed potential NPE when tablet contains nulls and sink format=tsfile. -Optimized sync retry strategy: now retries locally on receiver side in case of memory shortage or system rejection. Pipe Monitoring & Streaming Enhancements: -Added metrics: end-to-end latency from request entering Pipe framework to successful reception on receiver. -Added metric: backlog queue length of Pipe InsertNode. -Enabled automatic WAL compression disablement when streaming mode is turned on. Have a nice weekend :) Best regards, Pengcheng The Timecho Team
