Last week in mysql-server development (34)

[henrikingo-oracle]

This is the second weekly newsletter giving you a comprehensive overview of what is going on in the MySQL open development project. This newsletter was generated with the help of an AI tool, but significant manual editing was still involved. Note that the timeframe covered really is last week, ending Sunday.

• Repositories: mysql/mysql-server and mysql/mysql-community
• Date range: 2026-08-17 through 2026-08-23 inclusive.
• Coverage: 8 mysql-server PRs; 13 issues; 16 mysql-community Discussions.

Repository	Pull requests	Issues	Discussions
mysql/mysql-server	8	3	0
mysql/mysql-community	0	10	16

Executive Summary

1. InnoDB B+Tree Performance Optimization, Part 1 28 by @zhao-song sprung to life with a discussion how to phase this significant work and posting a PoC patch for review.
2. Several new proposals in binlog and replication: switchover observability, asynchronous crash-recovery rollback, large-transaction/DDL replication, and binlog commit optimizations. @SongLibing from Alibaba was active as one of his PR already is merged and several new proposals from him and other Alibaba engineers were lined up. The PR from @wxueting-aws is now in review, and @HugoWenTD posted a desingn for In-Memroy Relaylog. 684
3. Vector indexing remained active, and we also received a new proposal that nicely complements existing work in by @klmckeigIntegrate SVS (Vamana ANN) as a Vector Search Index Option for MySQL 116.

Merged

• Shrink NET::buff to reduce memory usage … 714 by @SongLibing

Implementing / Review

• [POC] Prototyping - MySQL User Defined Types … 707 by @marcalff
  Experimental UDT work. A rebase was requested; automated review raised authorization, lifecycle, metadata-persistence, and correctness concerns.

• Add DELETE ... RETURNING clause support … 725 by @evgenyp-azm implementation of DELETE … RETURNING, including parser, executor, and MTR changes.

• GROUP BY ALL … #660 by @hlosukwakha PR still has issues with including binary files.

• Fix typo in multi-statement announcement comment … 671 by @kazanzhy Minor correction.

Planning / PoC

• InnoDB B+Tree Performance Optimization, Part 1: Insert Path Improvements and Concurrent Split Handling … 28 by @zhao-song
  @KernelMaker posted a PoC patch. Maintainers accepted collaborative review but require use of the current table-type architecture.

• In-Memory Relaylog … 684 by @HugoWenTD
  AWS Aurora MySQL contributors presented an in-memory relay-log and group-boundary-queue design.

• Precise and observable primary switchover in Group Replication / InnoDB Cluster … [Proposal]: Precise and observable primary switchover in Group Replication / InnoDB Cluster #97 by @vgrippa
  Proposes excluding read-only and idle transactions from controlled-switchover waits and adding observability.

• Rollback Prepared Transactions Asynchronously During Binlog Crash Recovery … [Proposal]: Rollback Prepared Transactions Asynchronously During Binlog Crash Recovery #100 by @SongLibing from Alibaba
  Proposes moving recovered prepared-DML rollback off the startup-critical path.

• Optimizing Replication Lag for Large Transactions and DDL … 115 by @genze-wu from Alibaba
  Proposes binlog real-time replication for concurrent replica application of large transactions and DDL.

• Binlog Commit Optimization For Large Transaction … 77 by @SongLibing from Alibaba
  Proposes renaming an on-disk binlog-cache file into the next binary log rather than copying it under LOCK_log.

• Performance Optimization for Binlog Commit … 114 by @SongLibing from Alibaba
  Proposes binlog-in-redo and parallel binlog-flush designs.

• [VI] PoC Design Details … 701 by @mayprasa
  Active design discussion continues around vector-index storage, filtering, MVCC, latching, DML, redo, and interface design.

New

• Integrate SVS (Vamana ANN) as a Vector Search Index Option for MySQL … 116 by @klmckeig from Intel
  Proposes SVS/Vamana as an additional implementation behind the native vector-index interface.

Bug Fixes

• Bug#112704: Fix TempTable CTE clone handler mismatch when TempTable falls back to InnoDB … 672 by @catalinbp
  Open correctness fix for TempTable-to-InnoDB fallback and CTE failures; review and CI completion remain pending.

Misc / GitHub-platform and process fixes

• Update PR workflows for actions/checkout v7 compatibility … 715 by @RidhaOracle Closed as fixed

• Return a clearer error when account SSL requirements are not met … 99 by @Taeknology Expose SSL-policy failures to clients or limit detail to DBA-facing observability.

Steering group

The first members of the project's steering group have been announced

Contributor Summit talks available

The video recordings from the August 2026 Contributor Summit are now uploaded to Youtube

---

The goal for this newsletter is to give a high level status overview and link to all the different wikis and threads and code, with minimal commentary. If something significant is missing from this newsletter, please tell henrik.ingo@oracle.com

[jfg956]

Thanks for the Newsletter Henrik. I see there is a bug section, with only links to GitHub: have you considered adding significant activity from bugs.mysql.com ? Cheers, J-F

[henrikingo-oracle]

No, I think the scope of this newsletter is that bugs that were fixed by a contribution that came in as a pull request would be captured (they do end up getting their own bugs.mysql.com number in the end).

But since you mentioned bugs database, I did notice your social media post about attributing credit to authors of patches that fixed bugs. Someone is looking into adding the missing credits if/where necessary. This is an issue with the old path to contributing patches, so again kind of out of scope for this newsletter, but close enough and especially important enough that I thought you are entitled to a reply and I should mention it just in case nobody had reached out to you yet.

[jfg956]

I thought you are entitled to a reply

Moving this discussion to DMs in MySQL Community Slack.