Attender: Fei Hui, Yiyang, Chung en, Peter, Wei-Chiu, Sammi Summary - 2.3 Release Target: Plan to release the first RC before December. - 2.3 Core Feature Scope: Includes JDK 25 support, Short-circuit, S3 lifecycle, STS, ZDU initial version, CV release, and some *TLA+*-reported bug fixes. - S3 Versioning Compatibility Strategy: First send an email to the dev community to confirm whether any users are using versioning; if no users are found, skip complex compatibility handling and address it cleanly in the first version. - Read Performance Optimization: *netty* streaming read performs better when reading large files sequentially and has no significant negative impact on random reads of small amounts of data; it is considered more universally applicable than the existing stream read implementation.
Progress & Status - 2.3 Feature Testing: Internal testing has been ongoing; overall stability is good, and testing is expected to be completed within one to two weeks. - Bug Fix Progress: Of approximately 400 bugs reported by the TLA+, 130–140 have been fixed so far, roughly one-third complete. - PR Throughput: PR submission volume has been high recently, exceeding 100 per week and approximately 300–400 per month. - DR (Disaster Recovery) Feature: Testing is mostly complete and the team is preparing to go live; the team is also interested in the community's trace feature and plans to adopt it. - S3 Versioning Development: Initial implementation is complete, but backward compatibility logic is complex; two approaches are under consideration: throwing exceptions directly or maintaining a compatibility layer. - Performance Test Findings: In certain sequential read scenarios, client CPU usage can reach 50%–90%, possibly related to IO submission patterns or connection count. - EC (Erasure Coding) Usage: The team is already using EC in production for cold data storage with no issues found so far. Items to Confirm - S3 Versioning Community Usage: Need to confirm via email whether any users are actually using the versioning feature. - S3 Versioning Legacy Data Migration Plan: If legacy data exists, need to determine how to map true/false to the new states and whether a separate migration tool is required. - Root Cause of High Client CPU Usage: Further analysis needed to determine whether the issue is related to IO submission patterns or whether the number of connections between client and server needs to be increased. - Ntty Streaming Read vs. Existing Implementation: More test data is needed to clarify performance differences across different workloads. Action Items - Wei-Chiu: Continue driving the 2.3 release, targeting the first RC before December. - Wei-Chiu: Serve as the 2.3 release manager. - *yiyang*: Send an email to the dev community today to inquire about actual usage of S3 versioning. - yiyang: Decide on the S3 versioning compatibility approach based on community feedback, and evaluate whether a migration tool is needed. - peter: Continue testing performance differences between *netty* streaming read and the existing implementation, and update related patches. - peter: Address remaining condition-related patches to ensure correctness. - Chung: Follow up on the implementation progress of the S3 checksum design. - Fermi: After the DR feature goes live, share practical experience with multi-DC single-cluster setups with the community.
