📝 8 新 Issues
✅ 5 关闭
✨ 7 新 PRs
🎉 5 合并
🔥 核心要点
- 🎉 PR #2780 支持 Coordinator Leader 故障时自动切换，提升系统高可用性
- 🎉 PR #2947 修复 Azure 文件系统 Token 委托认证问题
- 🎉 PR #2922 添加 AI Agent 编码指南文档 AGENTS.md
- 🐛 Issue #2977 发现 PeriodicSnapshotManager 当 initialDelay 为 0 时静默跳过快照调度
- ✨ PR #2976 引入 KV Scan RPC 层，正在 Review 中
⚠️ 重点关注
- ⚠️ Issue #2975 不稳定测试: CoordinatorEventProcessorTest.testDoBucketReassignment
- ⚠️ Issue #2974 不稳定测试: RebalanceManagerITCase.testRebalanceWithRemoteLog
💬 Issue/PR 动态
Issue 讨论
Issue #2973 [server] Allow enabling lakehouse for log tables without bucket key created before cluster-level lakehouse is enabled
- @Copilot: 指出新增注释存在语法错误，建议重写以更清晰表达意图
- @beryllw: LGTM，认可该 PR 的实现方案
Issue #2900 [fluss-client] Support Complex Data Types on the Java API (NestedRow/ROW)
- @polyzos: 指出写路径每个元素重复创建 PojoToRowConverter 的性能问题，以及 Map 类型不安全的往返问题
Issue #2922 [docs] Add AGENTS.md - AI agent coding guide
- @vaibhavk1992: 询问是否需要进一步 PR 审批即可合并
- @platinumhamburg: 表示需要等待 wuchong 作为最终审阅者确认
Issue #2947 [fs/azure] Set auth.type=Custom for token delegation
- @affo: 修复了文档中的拼写错误并改进了注释，解释 auth.type 与 provider.type 配对使用
Issue #2976 Introduce RPC Layer for KV Scan
- @polyzos: 修复了部分不稳定测试，请求 wuchong 等人审阅
Issue #2780 [Feature] Support coordinator switch when leader fail
- @wuchong: 重构了代码：移除 @RequireCoordinatorLeader 注解、优化 leaderReadyFuture 逻辑、重用线程池
Issue #2933 [lake/tiering] add total bytes read tracking and metrics for tiering service
- @luoyuxia: 指出 TieringSplitReader 的快照字节数统计存在 ClassCastException 风险，建议重命名方法
- @leekeiabstraction: 建议使用更精确的断言，询问是否期望其他行类型
Issue #2730 [blog] Fluss rust / python / c++ client release blog draft
- @luoyuxia: 建议修改博客标题以对齐其他发布博客格式，添加 releases 标签
- @leekeiabstraction: 更新了内容，询问今天是否可以发布
Issue #2669 [docs] Add Real-Time User Profile quickstart tutorial
- @Prajwal-banakar: 同意简化快速入门指南方案，将完整的 RoaringBitmap 示例移至博客文章
Issue #2920 [lake/tiering] add table dropped handling and nullable write results
- @luoyuxia: 提出多个改进建议：表删除应作为失败事件统一处理、询问 LakeWriter 为 null 的场景、Paimon/Iceberg 同样需要处理
Issue #2855 [kv] Support downloading CDC log segments from remote storage during KV table recovery
- @platinumhamburg: 指出多个问题：迭代器资源泄露、remoteLogEndOffset 包含性边界错误、缺少边界条件测试
Issue #2769 [server] Integrate remote directory selector into table/partition creation
- @Copilot: 指出多处潜在问题：配置合并保留已删除键、删除配置检测失败、数组越界风险、不可变列表包装问题
Issue #2948 [rpc] Client zero-copy lazy parse ByteBuf to avoid deep memory copy
- @polyzos: 指出 retain() 在无记录情况下也会调用导致资源浪费，应仅在 hasRecords 为 true 时调用
Issue #2972 [flink] Implement SupportsWatermarkPushDown for FlinkTableSource
- @Copilot: 指出全局配置影响其他测试、执行计划断言过于脆弱、测试注释误导等问题
Issue #2594 [TASK-2293] use column id instead of position in Partial Update
- @polyzos: 指出 RowEncoder 未关闭导致资源泄露，createFieldGetters 每条记录调用性能问题
PR Review
暂无
📝 新建 Issue/PR
Issues
- #2977 [Server] PeriodicSnapshotManager.start() silently skips scheduling when initialDelay is 0 @platinumhamburg
- #2975 [test] Unstable test CoordinatorEventProcessorTest.testDoBucketReassignment @app/github-actions
- #2974 [test] Unstable test RebalanceManagerITCase.testRebalanceWithRemoteLog @app/github-actions
- #2971 [flink] FlinkTableSource should implement SupportsWatermarkPushDown @LiebingYu
- #2968 Fluss Introduces ChangelogNormalize Operator In WAL Image Mode @polyzos
- #2967 Fix OutOfBoundException in writeChangeType when the number of records is large @gyang94
- #2966 TieringSourceReader adjust to scan arrow record batch and write arrow record batch to lake @luoyuxia
- #2965 support log scanner scan data as arrow reacord batch @luoyuxia
Pull Requests
- #2978 [Server] Fix PeriodicSnapshotManager.start() silently disables snapshot @platinumhamburg
- #2976 Introduce RPC Layer for KV Scan @polyzos
- #2973 [server] Allow enabling lakehouse for log tables without bucket key created before cluster-level lakehouse is enabled @luoyuxia
- #2972 [flink] Implement SupportsWatermarkPushDown for FlinkTableSource @LiebingYu
- #2970 [WIP] feat: log sub batch poc @gyang94
- #2969 [kv] Fix non-target columns not being nulled during partial update on first insert @binary-signal
- #2964 [server] allow user to alter partition nun retention. @loserwang1024
✅ 关闭 Issue/PR
已关闭 Issues
- #2968 Fluss Introduces ChangelogNormalize Operator In WAL Image Mode @polyzos
- #2946 Azure FS token delegation does not take effect @affo
- #2932 [lake/tiering] add total bytes read tracking and metrics for tiering service @beryllw
- #2770 [Feature] Support coordinator switch when leader fail @zcoo
- #1246 [file system] Add configuration option to enable/disable token delegation process for S3 file system (allows support for MinIO and other S3-compatible object stores) @michaelkoepf component=filesystem
已合并 PRs
- #2953 [client][backport] Skip blocked bucket instead of stopping drain loop @fresh-borzoni
- #2947 [fs/azure] Set auth.type=Custom for token delegation @affo
- #2933 [lake/tiering] add total bytes read tracking and metrics for tiering service @beryllw
- #2922 [docs] Add AGENTS.md - AI agent coding guide @vaibhavk1992
- #2780 [Feature] Support coordinator switch when leader fail @zcoo