知识卡片
重试与新任务分队调度
依赖故障恢复后,积压重试可能让新到任务继续超时。将未尝试工作与重试工作分队可分别分配服务机会,优先处理仍有时效价值的新工作。调度应明确配额、优先级与防饥饿规则;分两队本身既不保证公平,也不自动解决总处理能力不足。
这张卡片看不懂?
这张卡片的反馈已满 10 条,暂时无法继续提交。
已有反馈 · 0 / 10 条
还没有人提交反馈。
在整书笔记中的位置
原书出处 · 1 处
- Chapter 12. Event-Driven Batch Processing · 查看原书依据
The first queue is used for all work items which have never had an attempt made at processing. The second queue is the “retry” queue, which is used for any message which has had at least one (unsuccessful) attempt at processing. Using this dual queue approach, in
参考来源
- 位置:《Designing Distributed Systems, 2nd Edition》第12章《Chapter 12. Event-Driven Batch Processing》(来源项目 book-sources;源文件:extracted/Designing Distributed Systems, 2nd Edition/_epub-src/OEBPS/ch12.html)
- 结论依据:原文双队列示例解释恢复后积压如何继续伤害新请求。
- 原始内容:The first queue is used for all work items which have never had an attempt made at processing. The second queue is the “retry” queue, which is used for any message which has had at least one (unsuccessful) attempt at processing. Using this dual queue approach, in