2 个仓库
Patterns and mechanisms for managing technical failures in distributed processes using retries and incident states.
Distinct from Exception Handling Pipelines: Different from API error mapping or shell exceptions; focuses on the long-running nature of workflow recovery.
Explore 2 awesome GitHub repositories matching software engineering & architecture · Workflow Exception Handling. Refine with filters or upvote what's useful.
Zeebe 是一个云原生工作流引擎和分布式状态机,旨在通过 BPMN 和 DMN 标准进行业务流程编排。它作为一个高性能 gRPC 工作流运行时,通过分区事件流架构执行复杂的业务流程。该系统还作为大语言模型代理的编排器,在确定性业务流程中协调 AI 推理和工具使用。 该引擎通过其点对点代理网络和确保高可用性和容错性的基于共识的数据复制模型而脱颖而出。它采用分区代理集群来实现水平扩展,并利用自适应请求背压来调节传入的命令流并防止系统过载。 该平台涵盖了广泛的操作功能,包括带有性能热力图的实时执行监控、通过决策表的自动化业务决策,以及通过基于轮询的作业工作者模型进行的分布式任务执行。它还提供用于多租户资源隔离、基于身份的访问控制以及集成外部 Web API 和无服务器函数的工具。 该系统可部署在 Kubernetes 和 Docker 等各种环境中,并通过命令行界面和程序化 REST API 的组合进行管理。
Manages technical failures using retries and incidents to ensure reliability and recovery.
Spring Retry is a Java library that provides exception handling, automated operation reattempts, backoff strategies, and recovery mechanisms for application workflows. It handles transient operation failures through declarative annotations, programmatic templates, stateful context storage, and fallback recovery paths. The library controls retry behavior by specifying attempt counts, timeouts, and inclusion or exclusion rules for specific exception types. It pauses execution between attempts using fixed, exponential, or randomized backoff intervals, and caches retry state outside the call stac
Provides a framework for managing failures in transactional workflows through retry limits and strategies.