Hi all, Here’s the update about our work at Timecho on IoTDB this week:
- Enhanced IT framework to better support restart testing. - Improved error messages for cases when Session is not opened. - Introduced memory control for the pipe sending queue (hot loading not supported). - Fixed NPE issue in monitoring when data synchronization is enabled via cross-gateway connector. - Addressed potential deadlock when concurrent delete timeseries operations occur at the receiver side. - Changed data type conversion during Load from asynchronous to synchronous execution for better stability. - Reduced the reporting interval from DataNode (DN) to ConfigNode (CN) to 3 seconds or less. - Improved asynchronous cleanup thread: if waiting for query execution thread exceeds 5 seconds, log with corresponding QueryId. - During pipe creation, check WAL compression status. If enabled, switch to WAL disable after pipe flush, while keeping single-point WAL load. - Refined TsFile resend condition after restart: changed from "pipe progress ≥ recorded progress" to a stricter "pipe progress > recorded progress". - Added safeguard on DN side to ensure heartbeat/meta sync requests from CN are only processed after the previous request completes. - Adjusted pipe restart logic: if a pipe remains stopped for over 5 minutes, it switches to a new file list. The 5-minute threshold has been extended. Have a nice weekend! Best regards, Pengcheng The Timecho Team
