apache/beam

[Feature Request]: Add metrics tracking for offset commit failures and retry attempts in KafkaCommitOffset

已关闭

#38,587 创建于 2026年5月21日

在 GitHub 查看贡献者指南
 (4 条评论) (0 个反应) (0 位负责人)Java (4,097 个派生)batch import
P3good first issueiojavanew feature

仓库指标

星标
 (7,313 个星标)
PR 合并指标
 (平均合并 6天) (30 天内合并 243 个 PR)

描述

What would you like to happen?

Currently in KafkaCommitOffset.CommitOffsetDoFn there is no visibility into how often offset commits are failing or how many retries are being attempted. If commits are silently failing in production you have no way to know without digging through logs. Adding Beam metrics here would make it easy to monitor commit health through existing monitoring dashboards. Proposed approach

Add a counter for total commit failures Add a counter for retry attempts Add a counter for commits that exhausted all retries Use Beam's built in Metrics.counter() so it works with any runner

Why this is useful If you're running a high throughput pipeline and offset commits are struggling, right now you'd have no idea until you see duplicate processing. This gives you an early warning signal.

Issue Priority

Priority: 3 (nice-to-have improvement)

Issue Components

  • Component: Python SDK
  • Component: Java SDK
  • Component: Go SDK
  • Component: Typescript SDK
  • Component: IO connector
  • Component: Beam YAML
  • Component: Beam examples
  • Component: Beam playground
  • Component: Beam katas
  • Component: Website
  • Component: Infrastructure
  • Component: Spark Runner
  • Component: Flink Runner
  • Component: Prism Runner
  • Component: Twister2 Runner
  • Component: Hazelcast Jet Runner
  • Component: Google Cloud Dataflow Runner

贡献者指南

研究方向
The implementation should be in the Java SDK's Kafka IO connector, specifically in the `KafkaCommitOffset.java` file within the `CommitOffsetDoFn` class. Examine the existing commit logic to identify where failures and retries occur. Use Beam's `Metrics.counter()` to add counters for total failures, retry attempts, and exhausted retries. The issue has no linked PRs or assignees, so it is available for contribution.
技术栈
javakafka
领域
backenddataobservability
议题类型
功能
难度面向新贡献者的预计实现难度，1 表示很小改动，5 表示专家级工作。
3
预计时间有经验贡献者完成调查、实现、测试并准备 pull request 的粗略时间范围。
半天
活动状态议题当前的可参与程度：新鲜、活跃、陈旧、阻塞或等待维护者输入。
新近可参与
清晰度议题是否清楚说明期望改动、验收标准和下一步。
清晰
前置要求
Apache Beam SDK knowledgeKafka offset commit mechanismJava
新手友好度1-100 的估计分数，表示该议题对首次贡献者的友好程度。
65