🌊 Fluss 每日动态

📅 2026-04-22

🔗 apache/fluss

📝 12 新 Issues

✅ 8 关闭

✨ 10 新 PRs

🎉 16 合并

🔥 核心要点

🎉 PR #3169 Flink Lake 表非分区扫描支持过滤器下推优化
🎉 PR #3171 Helm Chart 支持 existingSecret 配置 SASL 凭证
🐛 PR #3157 修复 KV 表部分更新时非目标列未置空问题
📚 PR #3150 重新引入 Quickstart Docker 镜像优化开发者体验
🚀 PR #3165 S3 委托令牌支持 path-style-access 配置传递

⚠️ 重点关注

✅ 今日无重大问题

💬 Issue/PR 动态

Issue 讨论

Issue #2683 [docs] Reintroduce the fluss-quickstart-flink Docker image

@luoyuxia: 将提议拆分为三个后续问题：恢复 Flink quickstart、移动手动设置到合适位置、规划 Spark quickstart

Issue #3161 [docs] Reintroduce a prebuilt fluss-quickstart-flink image for the main quickstart

@matrixsparse: 表示有兴趣贡献，询问是否可以分配此任务
@luoyuxia: 表示自己已经在处理此问题，因为之前是她移除的

Issue #3145 [test] Unstable test RemoteLogDownloaderTest.testPrefetchNum

@ankit-khare-2015: 表示愿意解决此问题，询问后续步骤，并在 fork 中测试了构建

Issue #3048 [lake][iceberg] Iceberg does not support union read for primary key table

@morazow: 讨论 PK 表 union read 方案，建议预排序写入或引入 deletion vectors

Issue #3063 [test] Unstable test RebalanceManagerITCase.testBuildClusterModel

@github-actions[bot]: 报告 CI 失败，RebalanceManagerITCase 测试抛出 Bucket leader and isr is empty 异常

Issue #3162 [docs] Move the manual JAR + Docker Compose Flink setup into Engine Flink getting started

@binary-signal: 支持提供预打包的 Docker 镜像，配合 Docker Compose 和 sidecar 容器可进一步简化体验

Issue #3163 [docs] Plan a fluss-quickstart-spark image for Spark Quickstart

@xi377266: 表示愿意处理此问题

Issue #3140 [POC][client] Support log scanner for multiple tables

@loserwang1024: 解释当前方法的问题：阻塞延迟高、多个 LogScanner 实例导致内存开销大

PR Review

PR #3132 [server] Optimize RemoteLogFetcher with async prefetch for recovery

@fresh-borzoni: 详细审查指出多个问题：资源泄漏、竞态条件、缺少重试机制、预取深度应可配置等

PR #3169 [flink] Push down lake filters for non-partitioned scans

@Copilot: 建议用 HashSet 替代 O(n²) 循环查找，并将注释集中到一处避免重复
@beryllw: 表示 LGTM

PR #3171 [helm] Add existingSecret support for SASL credentials

@affo: 建议拆分代码块添加注释、询问镜像选择是否可用 busybox
@fresh-borzoni: 确认已解决遗留的设计问题

PR #3141 [spark] Fix flaky SparkStreamingTest by handling interruption when close streaming execution

@beryllw: 指出 Thread.interrupt()、空行、fetchLatestOffsets 返回 None 时抛异常等问题
@Yohahaha: 建议重新抛出 InterruptedException 让 Spark streaming 捕获

PR #3154 [helm] scheduling primitives

@morazow: 建议使用占位符避免用户直接取消注释
@fresh-borzoni: 解释修改示例为 app.kubernetes.io/instance 以正确限定 anti-affinity 范围

PR #3150 Reintroduce Quickstart Image

@polyzos: 询问是否保持 Flink 1.20 版本
@luoyuxia: 解释 faker connector 不支持 Flink 2.2，暂保持 1.20 版本

PR #3008 [client] Fix corruption after zero-copy lazy parse ByteBuf

@Copilot: 指出 API 兼容性问题、建议添加单元测试验证 buffer 生命周期行为、修复 Javadoc 链接

PR #3030 [server] Implement JBOD Phase 1 Local Multi-Directory Support

@hanliu0830: 多处代码改进建议：异常处理改为抛出异常、路径需转绝对路径、trim 调用去重等

PR #3165 [s3] Propagate path-style-access config from server to client via delegation token

@Copilot: 建议优化代码避免重复 conf.get 调用、建议添加单元测试验证配置传递

PR #2786 [client] cancel pending log download when closing remote log downloader.

@fresh-borzoni: 指出多处资源管理和竞态问题：cancel、shutdownNow、close 时的状态管理
@loserwang1024: 确认已修改回原设计

📝 新建 Issue/PR

Issues

#3180 Add documentation for contributing a blog post @wuchong good first issue, component=website
#3179 Add documentation for FIP (Flink Improvement Proposal) contribution guide @wuchong good first issue, component=website
#3176 PK table messagesIn/bytesIn metrics double-counted due to CDC log @swuferhong
#3173 [server] Auto partition retention cleanup delayed by DAY partition creation jitter @swuferhong
#3172 [helm] Add existingSecret support for SASL credentials @fresh-borzoni
#3170 [tiering] Coordinator Should Push Lake Config to Tiering Service via Heartbeat RPC @Yohahaha
#3168 [flink] Lake filters shoud be pushed down for non-partitioned scans @luoyuxia
#3161 [docs] Reintroduce a prebuilt fluss-quickstart-flink image for the main quickstart @luoyuxia
#3162 [docs] Move the manual JAR + Docker Compose Flink setup into Engine Flink getting started @luoyuxia
#3163 [docs] Plan a fluss-quickstart-spark image for Spark Quickstart @luoyuxia
#3160 [spark] Improve spark ui more readable when reading fluss @Yohahaha
#3155 Introduce generic secondary index metadata foundation @platinumhamburg

Pull Requests

#3187 [lake] make compaction strategy min_input_files configurable @wattt3
#3186 [WIP] feat: system table @gyang94
#3184 [bp][helm] Support injecting environment variables and external secrets (backport to release-0.9) @morazow
#3182 [docs] add contribute a blog post guide (#3180) @tmchow
#3181 [flink] Push down lake filters for non-partitioned scans @luoyuxia
#3177 [server] Remove CDC log metrics from PK table write path to avoid dou… @swuferhong
#3174 [server] Fix auto partition drop delayed by DAY creation jitter @swuferhong
#3164 [spark] Exclude slf4j from spark connector shade @Yohahaha
#3158 [client] Fix race condition in RemoteLogDownloader causing flaky testPrefetchNum @ankit-khare-2015
#3156 Add generic secondary index metadata @platinumhamburg

✅ 关闭 Issue/PR

已关闭 Issues

#3172 [helm] Add existingSecret support for SASL credentials @fresh-borzoni
#3168 [flink] Lake filters shoud be pushed down for non-partitioned scans @luoyuxia
#3161 [docs] Reintroduce a prebuilt fluss-quickstart-flink image for the main quickstart @luoyuxia
#3153 [helm] Add pod scheduling primitives (affinity, nodeSelector, tolerations, topologySpreadConstraints) @fresh-borzoni
#3147 [server] Coordinator metrics not updated after restart when no events arrive @swuferhong
#3129 [Test] Unstable test SparkStreamingTest read: log table @wuchong component=connector/spark
#2848 [Feature] Provide built-in Flink SQL functions for RoaringBitmap construction and cardinality query (rbm32_build, rbm64_build, rbm32_cardinality, rbm64_cardinality) @matrixsparse
#2843 Invalid partial update for PK table @zuston version: 0.9.1

已合并 PRs

#3185 [bp][helm] Add pod annotations, labels, and pod disruption budget support (backport to release-0.9) @morazow
#3183 [helm] Backport helm chart improvements to release-0.9 @morazow
#3175 [helm] scheduling primitives (#3154) (backport to 0.9) @polyzos
#3171 [helm] Add existingSecret support for SASL credentials @fresh-borzoni
#3169 [flink] Push down lake filters for non-partitioned scans @luoyuxia
#3167 [s3] Propagate path-style-access via delegation token @luoyuxia
#3166 [cherry-pick][release-0.9] Reintroduce Quickstart Image (#3150) @luoyuxia
#3165 [s3] Propagate path-style-access config from server to client via delegation token @luoyuxia
#3157 [kv] Fix non-target columns not being nulled during partial update on first insert @luoyuxia
#3154 [helm] scheduling primitives @fresh-borzoni
#3150 Reintroduce Quickstart Image @polyzos
#3149 [client] Ensure UPDATE_BEFORE and UPDATE_AFTER are never split across poll batches @loserwang1024
#3148 [server] Fix coordinator metrics not updated when no events arrive @swuferhong
#3141 [spark] Fix flaky SparkStreamingTest by handling interruption when close streaming execution @Yohahaha
#3003 Fix buffer leak in LimitBatchScanner @fresh-borzoni
#2969 [kv] Fix non-target columns not being nulled during partial update on first insert @binary-signal