[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"similar-SeldonIO--seldon-core":3,"tool-SeldonIO--seldon-core":64},[4,17,27,35,43,56],{"id":5,"name":6,"github_repo":7,"description_zh":8,"stars":9,"difficulty_score":10,"last_commit_at":11,"category_tags":12,"status":16},3808,"stable-diffusion-webui","AUTOMATIC1111\u002Fstable-diffusion-webui","stable-diffusion-webui 是一个基于 Gradio 构建的网页版操作界面，旨在让用户能够轻松地在本地运行和使用强大的 Stable Diffusion 图像生成模型。它解决了原始模型依赖命令行、操作门槛高且功能分散的痛点，将复杂的 AI 绘图流程整合进一个直观易用的图形化平台。\n\n无论是希望快速上手的普通创作者、需要精细控制画面细节的设计师，还是想要深入探索模型潜力的开发者与研究人员，都能从中获益。其核心亮点在于极高的功能丰富度：不仅支持文生图、图生图、局部重绘（Inpainting）和外绘（Outpainting）等基础模式，还独创了注意力机制调整、提示词矩阵、负向提示词以及“高清修复”等高级功能。此外，它内置了 GFPGAN 和 CodeFormer 等人脸修复工具，支持多种神经网络放大算法，并允许用户通过插件系统无限扩展能力。即使是显存有限的设备，stable-diffusion-webui 也提供了相应的优化选项，让高质量的 AI 艺术创作变得触手可及。",162132,3,"2026-04-05T11:01:52",[13,14,15],"开发框架","图像","Agent","ready",{"id":18,"name":19,"github_repo":20,"description_zh":21,"stars":22,"difficulty_score":23,"last_commit_at":24,"category_tags":25,"status":16},1381,"everything-claude-code","affaan-m\u002Feverything-claude-code","everything-claude-code 是一套专为 AI 编程助手（如 Claude Code、Codex、Cursor 等）打造的高性能优化系统。它不仅仅是一组配置文件，而是一个经过长期实战打磨的完整框架，旨在解决 AI 代理在实际开发中面临的效率低下、记忆丢失、安全隐患及缺乏持续学习能力等核心痛点。\n\n通过引入技能模块化、直觉增强、记忆持久化机制以及内置的安全扫描功能，everything-claude-code 能显著提升 AI 在复杂任务中的表现，帮助开发者构建更稳定、更智能的生产级 AI 代理。其独特的“研究优先”开发理念和针对 Token 消耗的优化策略，使得模型响应更快、成本更低，同时有效防御潜在的攻击向量。\n\n这套工具特别适合软件开发者、AI 研究人员以及希望深度定制 AI 工作流的技术团队使用。无论您是在构建大型代码库，还是需要 AI 协助进行安全审计与自动化测试，everything-claude-code 都能提供强大的底层支持。作为一个曾荣获 Anthropic 黑客大奖的开源项目，它融合了多语言支持与丰富的实战钩子（hooks），让 AI 真正成长为懂上",138956,2,"2026-04-05T11:33:21",[13,15,26],"语言模型",{"id":28,"name":29,"github_repo":30,"description_zh":31,"stars":32,"difficulty_score":23,"last_commit_at":33,"category_tags":34,"status":16},2271,"ComfyUI","Comfy-Org\u002FComfyUI","ComfyUI 是一款功能强大且高度模块化的视觉 AI 引擎，专为设计和执行复杂的 Stable Diffusion 图像生成流程而打造。它摒弃了传统的代码编写模式，采用直观的节点式流程图界面，让用户通过连接不同的功能模块即可构建个性化的生成管线。\n\n这一设计巧妙解决了高级 AI 绘图工作流配置复杂、灵活性不足的痛点。用户无需具备编程背景，也能自由组合模型、调整参数并实时预览效果，轻松实现从基础文生图到多步骤高清修复等各类复杂任务。ComfyUI 拥有极佳的兼容性，不仅支持 Windows、macOS 和 Linux 全平台，还广泛适配 NVIDIA、AMD、Intel 及苹果 Silicon 等多种硬件架构，并率先支持 SDXL、Flux、SD3 等前沿模型。\n\n无论是希望深入探索算法潜力的研究人员和开发者，还是追求极致创作自由度的设计师与资深 AI 绘画爱好者，ComfyUI 都能提供强大的支持。其独特的模块化架构允许社区不断扩展新功能，使其成为当前最灵活、生态最丰富的开源扩散模型工具之一，帮助用户将创意高效转化为现实。",107662,"2026-04-03T11:11:01",[13,14,15],{"id":36,"name":37,"github_repo":38,"description_zh":39,"stars":40,"difficulty_score":23,"last_commit_at":41,"category_tags":42,"status":16},3704,"NextChat","ChatGPTNextWeb\u002FNextChat","NextChat 是一款轻量且极速的 AI 助手，旨在为用户提供流畅、跨平台的大模型交互体验。它完美解决了用户在多设备间切换时难以保持对话连续性，以及面对众多 AI 模型不知如何统一管理的痛点。无论是日常办公、学习辅助还是创意激发，NextChat 都能让用户随时随地通过网页、iOS、Android、Windows、MacOS 或 Linux 端无缝接入智能服务。\n\n这款工具非常适合普通用户、学生、职场人士以及需要私有化部署的企业团队使用。对于开发者而言，它也提供了便捷的自托管方案，支持一键部署到 Vercel 或 Zeabur 等平台。\n\nNextChat 的核心亮点在于其广泛的模型兼容性，原生支持 Claude、DeepSeek、GPT-4 及 Gemini Pro 等主流大模型，让用户在一个界面即可自由切换不同 AI 能力。此外，它还率先支持 MCP（Model Context Protocol）协议，增强了上下文处理能力。针对企业用户，NextChat 提供专业版解决方案，具备品牌定制、细粒度权限控制、内部知识库整合及安全审计等功能，满足公司对数据隐私和个性化管理的高标准要求。",87618,"2026-04-05T07:20:52",[13,26],{"id":44,"name":45,"github_repo":46,"description_zh":47,"stars":48,"difficulty_score":23,"last_commit_at":49,"category_tags":50,"status":16},2268,"ML-For-Beginners","microsoft\u002FML-For-Beginners","ML-For-Beginners 是由微软推出的一套系统化机器学习入门课程，旨在帮助零基础用户轻松掌握经典机器学习知识。这套课程将学习路径规划为 12 周，包含 26 节精炼课程和 52 道配套测验，内容涵盖从基础概念到实际应用的完整流程，有效解决了初学者面对庞大知识体系时无从下手、缺乏结构化指导的痛点。\n\n无论是希望转型的开发者、需要补充算法背景的研究人员，还是对人工智能充满好奇的普通爱好者，都能从中受益。课程不仅提供了清晰的理论讲解，还强调动手实践，让用户在循序渐进中建立扎实的技能基础。其独特的亮点在于强大的多语言支持，通过自动化机制提供了包括简体中文在内的 50 多种语言版本，极大地降低了全球不同背景用户的学习门槛。此外，项目采用开源协作模式，社区活跃且内容持续更新，确保学习者能获取前沿且准确的技术资讯。如果你正寻找一条清晰、友好且专业的机器学习入门之路，ML-For-Beginners 将是理想的起点。",84991,"2026-04-05T10:45:23",[14,51,52,53,15,54,26,13,55],"数据工具","视频","插件","其他","音频",{"id":57,"name":58,"github_repo":59,"description_zh":60,"stars":61,"difficulty_score":10,"last_commit_at":62,"category_tags":63,"status":16},3128,"ragflow","infiniflow\u002Fragflow","RAGFlow 是一款领先的开源检索增强生成（RAG）引擎，旨在为大语言模型构建更精准、可靠的上下文层。它巧妙地将前沿的 RAG 技术与智能体（Agent）能力相结合，不仅支持从各类文档中高效提取知识，还能让模型基于这些知识进行逻辑推理和任务执行。\n\n在大模型应用中，幻觉问题和知识滞后是常见痛点。RAGFlow 通过深度解析复杂文档结构（如表格、图表及混合排版），显著提升了信息检索的准确度，从而有效减少模型“胡编乱造”的现象，确保回答既有据可依又具备时效性。其内置的智能体机制更进一步，使系统不仅能回答问题，还能自主规划步骤解决复杂问题。\n\n这款工具特别适合开发者、企业技术团队以及 AI 研究人员使用。无论是希望快速搭建私有知识库问答系统，还是致力于探索大模型在垂直领域落地的创新者，都能从中受益。RAGFlow 提供了可视化的工作流编排界面和灵活的 API 接口，既降低了非算法背景用户的上手门槛，也满足了专业开发者对系统深度定制的需求。作为基于 Apache 2.0 协议开源的项目，它正成为连接通用大模型与行业专有知识之间的重要桥梁。",77062,"2026-04-04T04:44:48",[15,14,13,26,54],{"id":65,"github_repo":66,"name":67,"description_en":68,"description_zh":69,"ai_summary_zh":69,"readme_en":70,"readme_zh":71,"quickstart_zh":72,"use_case_zh":73,"hero_image_url":74,"owner_login":75,"owner_name":76,"owner_avatar_url":77,"owner_bio":78,"owner_company":79,"owner_location":79,"owner_email":80,"owner_twitter":79,"owner_website":81,"owner_url":82,"languages":83,"stars":122,"forks":123,"last_commit_at":124,"license":125,"difficulty_score":126,"env_os":127,"env_gpu":128,"env_ram":129,"env_deps":130,"category_tags":135,"github_topics":136,"view_count":23,"oss_zip_url":79,"oss_zip_packed_at":79,"status":16,"created_at":145,"updated_at":146,"faqs":147,"releases":177},2388,"SeldonIO\u002Fseldon-core","seldon-core","An MLOps framework to package, deploy, monitor and manage thousands of production machine learning models","Seldon Core 是一款专为 Kubernetes 环境设计的 MLOps 与 LLMOps 框架，旨在帮助用户高效地打包、部署、监控及管理成千上万个生产级机器学习模型。它解决了 AI 应用从实验阶段走向大规模生产时面临的部署复杂、资源利用率低及监控困难等痛点，支持在本地或任意云端以标准化方式运行各类模型。\n\n这款工具非常适合需要构建模块化、数据驱动型 AI 系统的开发者、算法工程师及运维团队。无论是处理单一模型还是复杂的组合应用，Seldon Core 都能提供开箱即用的生产级支持。其独特的技术亮点包括：利用 Kafka 实现组件间实时数据流的灵活管道（Pipelines）；基于原生或自定义逻辑的自动扩缩容能力；通过多模型服务共享推理服务器以节省基础设施成本；以及支持“超量部署”（Overcommit），允许部署超出当前内存限制的模型数量，从而优化闲置资源。此外，它还内置了强大的实验功能，支持 A\u002FB 测试和影子部署，方便用户在候选模型间进行流量路由与效果验证，是打造高可用、可扩展 AI 应用的理想选择。","\u003Cdiv align=\"center\">\n\n \u003Ca href=\"https:\u002F\u002Fwww.seldon.io\u002Fsolutions\u002Fcore\u002F\">\n  \u003Cimg alt=\"Core 2 Logo\" src=\"\u002F.images\u002Fcore-2-logo.png\" alt=\"Core 2 Logo\" style=\"max-width: 100%; height: auto; width: 400px;\">\n \u003C\u002Fa>\n\u003C\u002Fdiv>\n\n# Deploy Modular, Data-centric AI applications at scale\n\n##  💡 About\nSeldon Core 2 is an MLOps and LLMOps framework for deploying, managing and scaling AI systems in Kubernetes - from singular models, to modular and data-centric applications. With Core 2 you can deploy in a standardized way across a wide range of model types, on-prem or in any cloud, and production-ready out of the box. \n\n\u003C\u002Fbr>\n \u003Cdiv align=\"center\">\n   \u003Ca href=\"https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=ar5lSG_idh4\">\n     \u003Cimg src=\"\u002F.images\u002FCore-intro-thumbnail.png\" alt=\"Introductory Youtube Video\" style=\"max-width: 100%; width: 500px; height: auto;\">\n   \u003C\u002Fa>\n \u003C\u002Fdiv>\n\u003C\u002Fbr>\n\nTo reach out to Seldon regarding commercial use, visit our [website](https:\u002F\u002Fwww.seldon.io\u002F).\n\n## 📚 Documentation  \n\nThe Seldon Core 2 Docs can be found [here](https:\u002F\u002Fdocs.seldon.ai\u002Fseldon-core-2). For most specific sections, see here:\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fdocs.seldon.ai\u002Fseldon-core-2\u002Finstallation\u002Finstallation\">🔧 Installation\u003C\u002Fa>  &nbsp • &nbsp\n  \u003Ca href=\"https:\u002F\u002Fdocs.seldon.ai\u002Fseldon-core-2\u002Fuser-guide\u002Fservers\"> ⛽ Servers\u003C\u002Fa>  &nbsp • &nbsp\n  \u003Ca href=\"https:\u002F\u002Fdocs.seldon.ai\u002Fseldon-core-2\u002Fuser-guide\u002Fmodels\">🤖 Models\u003C\u002Fa>  &nbsp •  &nbsp\n  \u003Ca href=\"https:\u002F\u002Fdocs.seldon.ai\u002Fseldon-core-2\u002Fuser-guide\u002Fpipelines\"> 🔗 Pipelines\u003C\u002Fa>  &nbsp • &nbsp\n  \u003Ca href=\"https:\u002F\u002Fdocs.seldon.ai\u002Fseldon-core-2\u002Fuser-guide\u002Fexperiment\">🧑‍🔬 Experiments\u003C\u002Fa>  &nbsp • &nbsp\n  \u003Ca href=\"https:\u002F\u002Fdocs.seldon.ai\u002Fseldon-core-2\u002Fuser-guide\u002Fperformance-tuning\">📊 Performance Tuning\u003C\u002Fa>  \n\u003C\u002Fp>\n\n## 🧩 Features\n\n * **Pipelines**: Deploy composable AI applications, leveraging Kafka for realtime data streaming between components\n * **Autoscaling** for models and application components based on native or custom logic\n * **Multi-Model Serving**: Save infrastructure costs by consolidating multiple models on shared inference servers\n * **Overcommit**: Deploy more models than available memory allows, saving infrastructure costs for unused models\n * **Experiments**: Route data between candidate models or pipelines, with support for A\u002FB tests and shadow deployments\n * **Custom Components**: Implement custom logic, drift & outlier detection, LLMs and more through plug-and-play integrate with the rest of Seldon's ecosytem of ML\u002FAI products!\n \n## 🔬 Research\n\nThese features are influenced by our position paper on the next generation of ML model serving frameworks: \n\n👉 [Desiderata for next generation of ML model serving](http:\u002F\u002Farxiv.org\u002Fabs\u002F2210.14665)\n\n## 📜 License\n\nSeldon is distributed under the terms of the The Business Source License. A complete version of the license is available in the [LICENSE file](LICENSE) in this repository. Any contribution made to this project will be licensed under the Business Source License.\n\n\n\n","\u003Cdiv align=\"center\">\n\n \u003Ca href=\"https:\u002F\u002Fwww.seldon.io\u002Fsolutions\u002Fcore\u002F\">\n  \u003Cimg alt=\"Core 2 Logo\" src=\"\u002F.images\u002Fcore-2-logo.png\" alt=\"Core 2 Logo\" style=\"max-width: 100%; height: auto; width: 400px;\">\n \u003C\u002Fa>\n\u003C\u002Fdiv>\n\n# 大规模部署模块化、以数据为中心的AI应用\n\n##  💡 关于\nSeldon Core 2 是一个 MLOps 和 LLMOps 框架，用于在 Kubernetes 中部署、管理和扩展 AI 系统——从单个模型到模块化、以数据为中心的应用。借助 Core 2，您可以以标准化的方式部署各种类型的模型，无论是在本地还是在任何云环境中，并且开箱即用即可投入生产。\n\n\u003C\u002Fbr>\n \u003Cdiv align=\"center\">\n   \u003Ca href=\"https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=ar5lSG_idh4\">\n     \u003Cimg src=\"\u002F.images\u002FCore-intro-thumbnail.png\" alt=\"介绍性 YouTube 视频\" style=\"max-width: 100%; width: 500px; height: auto;\">\n   \u003C\u002Fa>\n \u003C\u002Fdiv>\n\u003C\u002Fbr>\n\n如需就商业用途与 Seldon 联系，请访问我们的 [网站](https:\u002F\u002Fwww.seldon.io\u002F)。\n\n## 📚 文档  \n\nSeldon Core 2 的文档可以在这里找到：[https:\u002F\u002Fdocs.seldon.ai\u002Fseldon-core-2](https:\u002F\u002Fdocs.seldon.ai\u002Fseldon-core-2)。对于最具体的章节，请参阅以下内容：\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fdocs.seldon.ai\u002Fseldon-core-2\u002Finstallation\u002Finstallation\">🔧 安装\u003C\u002Fa>  &nbsp • &nbsp\n  \u003Ca href=\"https:\u002F\u002Fdocs.seldon.ai\u002Fseldon-core-2\u002Fuser-guide\u002Fservers\"> ⛽ 服务器\u003C\u002Fa>  &nbsp • &nbsp\n  \u003Ca href=\"https:\u002F\u002Fdocs.seldon.ai\u002Fseldon-core-2\u002Fuser-guide\u002Fmodels\">🤖 模型\u003C\u002Fa>  &nbsp •  &nbsp\n  \u003Ca href=\"https:\u002F\u002Fdocs.seldon.ai\u002Fseldon-core-2\u002Fuser-guide\u002Fpipelines\"> 🔗 流水线\u003C\u002Fa>  &nbsp • &nbsp\n  \u003Ca href=\"https:\u002F\u002Fdocs.seldon.ai\u002Fseldon-core-2\u002Fuser-guide\u002Fexperiment\">🧑‍🔬 实验\u003C\u002Fa>  &nbsp • &nbsp\n  \u003Ca href=\"https:\u002F\u002Fdocs.seldon.ai\u002Fseldon-core-2\u002Fuser-guide\u002Fperformance-tuning\">📊 性能调优\u003C\u002Fa>  \n\u003C\u002Fp>\n\n## 🧩 特性\n\n * **流水线**：部署可组合的 AI 应用，利用 Kafka 在组件之间进行实时数据流传输。\n * **自动扩缩容**：根据原生或自定义逻辑对模型和应用组件进行自动扩缩容。\n * **多模型服务**：通过将多个模型整合到共享的推理服务器上，降低基础设施成本。\n * **超量部署**：部署超出可用内存容量的模型数量，从而节省未使用模型的基础设施成本。\n * **实验**：在候选模型或流水线之间路由数据，支持 A\u002FB 测试和影子部署。\n * **自定义组件**：通过即插即用的方式与其他 Seldon 的机器学习\u002F人工智能产品生态系统集成，实现自定义逻辑、漂移与异常检测、大语言模型等功能！\n\n## 🔬 研究\n\n这些特性受到我们关于下一代 ML 模型服务框架的立场论文的影响：\n\n👉 [下一代 ML 模型服务的需求](http:\u002F\u002Farxiv.org\u002Fabs\u002F2210.14665)\n\n## 📜 许可证\n\nSeldon 根据商业源代码许可证条款进行分发。完整的许可证版本可在本仓库的 [LICENSE 文件](LICENSE) 中找到。对该项目的任何贡献都将依据商业源代码许可证进行授权。","# Seldon Core 2 快速上手指南\n\nSeldon Core 2 是一个专为 Kubernetes 设计的 MLOps 和 LLMOps 框架，支持从单一模型到模块化、数据驱动型 AI 应用的大规模部署与管理。\n\n## 环境准备\n\n在开始之前，请确保满足以下系统要求和前置依赖：\n\n*   **操作系统**: Linux 或 macOS (Windows 用户建议使用 WSL2 或 Docker Desktop)。\n*   **Kubernetes 集群**: 需要一个正在运行的 Kubernetes 集群 (版本 1.23+)。\n    *   本地开发推荐：[Kind](https:\u002F\u002Fkind.sigs.k8s.io\u002F), [Minikube](https:\u002F\u002Fminikube.sigs.k8s.io\u002F), 或 [Docker Desktop](https:\u002F\u002Fwww.docker.com\u002Fproducts\u002Fdocker-desktop\u002F)。\n    *   生产环境推荐：EKS, GKE, AKS 或自建集群。\n*   **kubectl**: 已安装并配置好与集群的通信。\n*   **Helm**: 版本 3.0+ (用于安装 Seldon Core)。\n*   **容器镜像源 (国内加速)**:\n    *   由于 Seldon 官方镜像位于 Docker Hub，国内访问可能较慢。建议配置国内镜像加速器（如阿里云、腾讯云等）或在 `values.yaml` 中指定国内镜像仓库地址。\n\n## 安装步骤\n\n使用 Helm 是安装 Seldon Core 2 最便捷的方式。\n\n1.  **添加 Seldon Helm 仓库**\n\n    ```bash\n    helm repo add seldon https:\u002F\u002Fstorage.googleapis.com\u002Fseldon-charts\n    helm repo update\n    ```\n\n2.  **创建命名空间**\n\n    ```bash\n    kubectl create namespace seldon-system\n    ```\n\n3.  **安装 Seldon Core**\n\n    执行以下命令进行默认安装：\n\n    ```bash\n    helm install seldon-core seldon\u002Fseldon-core \\\n      --namespace seldon-system \\\n      --set prefix=seldon\n    ```\n\n    > **提示**：若需自定义配置（如指定镜像源以加速下载），可先拉取 values 文件修改后再安装：\n    > ```bash\n    > helm show values seldon\u002Fseldon-core > values.yaml\n    > # 编辑 values.yaml 修改 image.repository 等字段\n    > helm install seldon-core seldon\u002Fseldon-core -f values.yaml --namespace seldon-system\n    > ```\n\n4.  **验证安装**\n\n    等待所有 Pod 状态变为 `Running`：\n\n    ```bash\n    kubectl get pods -n seldon-system\n    ```\n\n## 基本使用\n\n以下示例展示如何部署一个简单的机器学习模型（例如一个预训练的 SkLearn 模型）。\n\n### 1. 准备模型定义文件\n\n创建一个名为 `model.yaml` 的文件，定义一个 `SeldonDeployment` 资源。此示例使用 Seldon 提供的内置测试镜像。\n\n```yaml\napiVersion: mlops.seldon.io\u002Fv1\nkind: SeldonDeployment\nmetadata:\n  name: example-model\n  namespace: default\nspec:\n  predictors:\n  - graph:\n      implementation: SKLEARN_SERVER\n      modelUri: gs:\u002F\u002Fseldon-models\u002Fv1.21.0-dev\u002Fsklearn\u002Fmnist\n      name: classifier\n    name: default\n    replicas: 1\n```\n\n> **注意**: `modelUri` 指向 Google Cloud Storage。国内用户若无法访问，可替换为托管在国内对象存储（如 OSS\u002FCOS）上的模型路径，或使用本地构建的容器镜像 (`type: \"container\"`)。\n\n### 2. 部署模型\n\n应用配置文件到 Kubernetes 集群：\n\n```bash\nkubectl apply -f model.yaml\n```\n\n### 3. 检查部署状态\n\n查看部署状态，直到 `STATUS` 显示为 `Available`：\n\n```bash\nkubectl get sdep example-model\n```\n\n### 4. 发送预测请求\n\n获取服务的入口地址（假设已配置 Ingress 或使用端口转发）。若使用端口转发进行测试：\n\n```bash\nkubectl port-forward svc\u002Fexample-model-default-0 8000:8000\n```\n\n使用 `curl` 发送测试请求：\n\n```bash\ncurl -X POST http:\u002F\u002Flocalhost:8000\u002Fv2\u002Fmodels\u002Fclassifier\u002Finfer \\\n  -H \"Content-Type: application\u002Fjson\" \\\n  -d '{\"inputs\": [{\"name\": \"input\", \"shape\": [1, 64], \"datatype\": \"FP32\", \"data\": [[0.0]*64]}]}'\n```\n\n如果返回包含 `outputs` 的 JSON 响应，则说明模型部署成功并开始提供服务。","某大型电商平台的推荐算法团队需要在 Kubernetes 集群上管理数百个实时预测模型，并支持频繁的 A\u002FB 测试与动态扩缩容。\n\n### 没有 seldon-core 时\n- **部署碎片化**：每个模型都需要编写独立的 Dockerfile 和 K8s YAML 文件，维护成本极高且容易出错。\n- **资源浪费严重**：无法在同一个推理服务中混合部署多个小模型，导致大量 CPU 和内存闲置，基础设施成本居高不下。\n- **实验流程繁琐**：进行 A\u002FB 测试或灰度发布时，需手动修改网关配置和流量路由规则，耗时且风险大。\n- **监控盲区**：缺乏统一的指标收集标准，难以实时监控模型漂移（Drift）或异常值，故障排查依赖人工日志检索。\n- **弹性不足**：面对大促流量洪峰，模型服务无法基于自定义业务指标自动扩缩容，常出现响应超时或服务宕机。\n\n### 使用 seldon-core 后\n- **标准化交付**：利用其统一的模型打包规范，团队可将不同框架训练的模型以标准化方式一键部署至任何云环境或本地集群。\n- **多模型复用**：通过“多模型服务”特性，将数十个小模型 Consolidate 到共享的推理服务器中，显著降低资源占用和云账单。\n- **原生实验支持**：直接配置 A\u002FB 测试和影子部署（Shadow Deployment），seldon-core 自动处理复杂的流量切分与路由，让实验迭代从数天缩短至分钟级。\n- **全链路可观测**：内置监控组件自动暴露标准指标，结合自定义逻辑轻松实现模型漂移检测与异常报警，保障生产稳定性。\n- **智能弹性伸缩**：基于原生或自定义逻辑实现细粒度自动扩缩容，从容应对流量波动，确保高并发下的低延迟响应。\n\nseldon-core 将原本杂乱无章的模型运维工作转化为模块化、数据驱动的自动化流程，极大提升了企业级 AI 应用的交付效率与运行可靠性。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002FSeldonIO_seldon-core_390ac228.png","SeldonIO","Seldon","https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002FSeldonIO_17a4841a.png","Machine Learning Deployment for Kubernetes",null,"hello@seldon.io","https:\u002F\u002Fseldon.io","https:\u002F\u002Fgithub.com\u002FSeldonIO",[84,88,92,96,100,103,107,111,115,119],{"name":85,"color":86,"percentage":87},"Go","#00ADD8",55.1,{"name":89,"color":90,"percentage":91},"Jupyter Notebook","#DA5B0B",20.2,{"name":93,"color":94,"percentage":95},"Kotlin","#A97BFF",13.3,{"name":97,"color":98,"percentage":99},"JavaScript","#f1e05a",3.1,{"name":101,"color":86,"percentage":102},"Go Template",2.5,{"name":104,"color":105,"percentage":106},"Java","#b07219",1.6,{"name":108,"color":109,"percentage":110},"Makefile","#427819",1.5,{"name":112,"color":113,"percentage":114},"Gherkin","#5B2063",1.1,{"name":116,"color":117,"percentage":118},"Python","#3572A5",0.7,{"name":120,"color":121,"percentage":118},"HTML","#e34c26",4738,861,"2026-04-02T08:36:27","NOASSERTION",4,"未说明 (基于 Kubernetes，通常支持 Linux)","未说明 (取决于部署的具体模型需求)","未说明",{"notes":131,"python":129,"dependencies":132},"Seldon Core 2 是一个运行在 Kubernetes 上的 MLOps\u002FLLMOps 框架，而非直接的 Python 库。其核心运行环境依赖是 Kubernetes 集群。它支持模块化 AI 应用部署，利用 Kafka 进行实时数据流传输。具体的 GPU、内存、Python 版本及深度学习库（如 PyTorch, TensorFlow）需求完全取决于用户在平台上部署的具体模型或组件，框架本身不强制指定这些底层硬件或软件版本。",[133,134],"Kubernetes","Kafka",[13],[137,138,139,140,141,142,143,144],"kubernetes","machine-learning","deployment","serving","mlops","aiops","machine-learning-operations","production-machine-learning","2026-03-27T02:49:30.150509","2026-04-06T06:52:07.914872",[148,153,158,162,167,172],{"id":149,"question_zh":150,"answer_zh":151,"source_url":152},10985,"创建 SeldonDeployment 时遇到 'service seldon-webhook-service not found' 错误怎么办？","该错误通常表示 Seldon Webhook 服务未正确安装或未在指定命名空间中运行。请检查 seldon-system 命名空间下是否存在名为 'seldon-webhook-service' 的服务。可以通过运行 'kubectl get svc -n seldon-system' 确认。如果服务不存在，可能需要重新安装 Seldon Core Operator 或检查 Helm Chart 部署是否成功。确保 webhook 配置中的服务地址与实际部署的服务名称和命名空间匹配。","https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fissues\u002F952",{"id":154,"question_zh":155,"answer_zh":156,"source_url":157},10986,"使用 s2i 构建 Seldon 镜像时出现 'ImportError: libcublas.so.9.0: cannot open shared object file' 错误如何解决？","此错误表明容器内缺少 CUDA 库文件。需要在 .s2i\u002Fenvironment 文件中正确设置 LD_LIBRARY_PATH 环境变量，指向 CUDA 库所在路径（例如：LD_LIBRARY_PATH=\u002Fusr\u002Flocal\u002Fcuda-9.0\u002Flib64）。同时确保基础镜像中包含对应版本的 CUDA 运行时库。如果使用 tensorflow-gpu，需确认镜像版本与 CUDA 版本兼容。还可以尝试在 requirements.txt 中明确指定与 CUDA 版本匹配的 tensorflow-gpu 版本。","https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fissues\u002F215",{"id":159,"question_zh":160,"answer_zh":161,"source_url":157},10987,"发送预测请求时收到 JSON 格式无效的错误，正确的 payload 格式是什么？","Seldon Core 期望特定的 JSON 格式。对于 ndarray 类型的数据，正确的格式应为：{\"data\":{\"ndarray\":[[1,2,3,...]]}}。注意外层是 'data' 对象，内部包含 'ndarray' 键，其值为二维数组。避免直接发送原始数组或错误的嵌套结构。如果使用 Python 客户端，可让 SeldonClient 自动处理格式转换；若手动构造请求，请严格遵循此 schema。",{"id":163,"question_zh":164,"answer_zh":165,"source_url":166},10988,"如何将大型 numpy 数组转换为 JSON 时减少 payload 大小以提高响应速度？","将 numpy 数组直接通过 tolist() 转为 JSON 会导致数据膨胀。建议方案包括：1) 使用二进制格式（如 protobuf\u002Ftensor_proto）而非 JSON 传输数据，Seldon 支持 gRPC 协议可高效处理张量；2) 对图像等数据进行压缩（如 JPEG\u002FPNG 编码）后再传输；3) 在模型服务端实现 predict_raw 接口直接处理 proto 对象，避免中间转换为列表带来的开销。若必须用 REST，可考虑分块传输或降低数据精度。","https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fissues\u002F365",{"id":168,"question_zh":169,"answer_zh":170,"source_url":171},10989,"本地使用 REST 调用 Seldon 微服务时延迟很高，如何优化？","高延迟主要源于 Proto 与 numpy 数组之间的序列化\u002F反序列化转换。优化建议：1) 改用 gRPC 协议替代 REST，gRPC 在二进制传输和序列化效率上更优；2) 在模型代码中实现 predict_raw 方法，直接操作 protobuf 对象，跳过默认的 array\u003C->proto 转换步骤；3) 确保使用最新版本的 seldon-core 镜像，其中包含性能改进（参考 PR #739）；4) 检查网络配置，本地测试时确保使用 localhost 而非外部 IP。","https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fissues\u002F803",{"id":173,"question_zh":174,"answer_zh":175,"source_url":176},10990,"如何在 Helm Chart 中为 seldon-webhook-service 和 controller-manager 添加自定义 labels 和 annotations？","可以通过修改 Helm values.yaml 文件来实现。在 seldon-core-operator 的 chart 中，添加如下配置：\noperator:\n  service:\n    labels:\n      prometheus.io\u002Fscrape: \"false\"\n    annotations:\n      custom.annotation\u002Fkey: \"value\"\n  deployment:\n    labels:\n      app.kubernetes.io\u002Fpart-of: seldon\n    annotations:\n      sidecar.istio.io\u002Finject: \"false\"\n这样可以在部署时为服务和 Pod 注入指定的元数据，用于控制 Prometheus 抓取、Istio 边车注入等行为。具体字段名需参考 chart 模板中的定义。","https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fissues\u002F2860",[178,183,188,193,198,203,208,213,218,223,228,233,238,243,248,253,258,263,268,273],{"id":179,"version":180,"summary_zh":181,"released_at":182},53461,"v1.19.0","## 摘要\n\n### 概述\n本次发布重点在于平台维护、安全加固和生态系统的现代化改造。主要工作包括大规模修复 CVE 漏洞、全面将镜像和工具链升级至 Python 3.12、更新依赖项和基础镜像、优化 CI 和工作流，以及清理和迁移文档。虽然未引入重大新用户功能，但此次发布显著提升了安全态势、测试覆盖率和长期可维护性。\n\n### 安全与 CVE 修复\n- 在 Go、Python 和容器镜像中修复了多个 CVE，涉及 operator、executor、alibi-detect 服务器、python-builder 和核心构建器镜像。\n- 将 Go 运行时更新至较新的补丁版本，以修复已知漏洞。\n- 更新自动伸缩依赖项，修复 KEDA 库中的 CVE。\n- 升级 RClone 并刷新基础镜像，以解决容器层面的 CVE。\n\n### 依赖项与运行时升级\n- Python 全面升级至 3.12，涵盖：\n  - 预打包服务器\n  - Alibi Detect 和 Alibi Explain 服务器\n  - Python 封装库\n- Conda 基础镜像从 Miniconda 迁移至 Miniforge（开源版）。\n- UBI 基础镜像统一并更新至最新版本（适用时为 UBI 9.7）。\n- 更新 Protobuf、gRPC、Poetry 等构建时依赖项。\n- Kubernetes 兼容性测试范围扩展至支持 1.35 版本（1.23.x 至 1.35.x）。\n\n### 镜像与构建系统\n- 在 Docker 镜像构建中新增 SBOM 生成，以提升许可证和供应链的可见性。\n- 弃用 Python 封装库及两个 Alibi 服务器相关的 GPU 相关 Docker 镜像。\n\n### 许可与合规\n- 重新生成并清理许可证元数据。\n\n### Helm、Operator 及配置变更\n- 已弃用的分析 Helm 图表从构建工作流中移除。\n- Keda 依赖项升级至 2.17.3，并向前兼容 2.18.3（最新版本）。\n\n### 文档\n文档已迁移到 GitBook，访问地址为 [这里](https:\u002F\u002Fdocs.seldon.ai\u002Fseldon-core-1)。\n\n### 弃用与清理\n- 正式弃用与 GPU 相关的文件。\n- 移除或禁用若干示例和图表。\n\n### 向后兼容性\n- 用户应针对 Python 3.12 验证自定义镜像和笔记本。\n- 从旧版本升级的 Operator 应审查镜像和基础操作系统的变化。\n\n## 升级说明\n\n**Python 3.12 的采用**\n所有核心 Python 组件——包括预打包服务器、Alibi Detect\u002FExplain 服务器、封装库、笔记本和端到端测试——现已面向 Python 3.12 构建。\n\n这意味着：\n- 尚未兼容 Python 3.12 的轮子包或依赖项可能在运行时失败。\n- 基于旧版 Python 构建的自定义推理镜像必须重新构建并进行测试。\n\n**容器基础镜像变更**\n- executor、operator 及相关支持镜像全面迁移至较新的 UBI 基础镜像。\n- Conda 基础镜像切换至 Miniforge。\n- 出于安全考虑，RClone 及其他第三方工具均已升级。\n\n影响：\n- 镜像大小和层组成将发生轻微变化。","2026-01-23T19:10:04",{"id":184,"version":185,"summary_zh":186,"released_at":187},53462,"v1.19.0-rc.1","## 摘要\n\n### 概述\n此版本候选主要聚焦于平台维护、安全加固和生态系统的现代化改造。核心主题包括大规模修复 CVE 漏洞、全面将各镜像及工具链升级至 Python 3.12、更新依赖和基础镜像、优化 CI 和工作流，以及清理和迁移文档。本次发布未引入重大新用户功能，但显著提升了安全态势、测试覆盖率和长期可维护性。\n\n### 安全与 CVE 修复\n- 在 Go、Python 和容器镜像中修复了多个 CVE 漏洞，涉及 operator、executor、alibi-detect 服务器、python-builder 和核心构建器镜像。\n- 将 Go 运行时更新至较新的补丁版本，以修复已知漏洞。\n- 更新自动伸缩依赖项，修复 KEDA 库中的 CVE 漏洞。\n- 升级 RClone 并刷新基础镜像，以解决容器层面的 CVE 漏洞。\n\n### 依赖与运行时升级\n- Python 全面升级至 Python 3.12，涵盖：\n  - 预打包服务器\n  - Alibi Detect 和 Alibi Explain 服务器\n  - Python 封装库\n- Conda 基础镜像从 Miniconda 迁移至 Miniforge（OSS）。\n- UBI 基础镜像统一并更新至 UBI 9.7（如适用）。\n- 更新 Protobuf、gRPC、Poetry 等构建时依赖。\n- 扩展 Kubernetes 兼容性测试范围，支持 Kubernetes 1.35（1.23.x 至 1.35.x）。\n\n### 镜像与构建系统\n- 在 Docker 镜像构建中新增 SBOM 生成，以提升许可证和供应链的可见性。\n- 弃用 Python 封装库及两个 Alibi 服务器相关的 GPU 相关 Docker 镜像。\n\n### 许可与合规\n重新生成并清理许可证元数据。\n\n### Helm、Operator 及配置变更\n- 已弃用的分析 Helm 图表从构建工作流中移除。\n- Keda 依赖升级至 2.17.3，并向前兼容 2.18.3（最新版本）。\n\n### 文档\n文档已迁移到 GitBook，访问地址为 [这里](https:\u002F\u002Fdocs.seldon.ai\u002Fseldon-core-1)。\n\n### 弃用与清理\n- 正式弃用与 GPU 相关的文件。\n- 移除或禁用若干示例和图表。\n\n### 向后兼容性\n- 本版本候选未引入有意破坏性的 API 变更。\n- 用户应针对 Python 3.12 验证自定义镜像和笔记本。\n- 从旧版本升级的 Operator 应审查镜像和基础操作系统的变化。\n\n## 升级说明\n\n**Python 3.12 的采用**\n所有核心 Python 组件——包括预打包服务器、Alibi Detect\u002FExplain 服务器、封装库、笔记本和端到端测试——现已面向 Python 3.12 构建。\n\n这意味着：\n- 尚未兼容 Python 3.12 的轮子包或依赖项可能在运行时失败。\n- 基于旧版 Python 构建的自定义推理镜像必须重新构建并进行测试。\n\n**容器基础镜像变更**\n- executor、operator 及相关支持镜像全面迁移到更新的 UBI 基础镜像。\n- Conda 基础镜像切换至 Miniforge。\n- RClone 及其他第三方工具升级","2026-01-08T12:00:59",{"id":189,"version":190,"summary_zh":191,"released_at":192},53463,"v2.10.2","概述\r\n\r\nCore `2.10.2` 是一个补丁版本，修复了多个长期存在的问题，这些问题在 `2.10.x` 系列版本中变得更加突出。\r\n\r\n- 流水线无法创建或删除，会一直停留在该状态，即使集群随后恢复正常。\r\n- 代理与调度器之间的 gRPC 流可能被阻塞，导致模型无法加载。\r\n- Operator 被阻止对自定义资源进行协调，从而使得管理员无法进行任何更改。\r\n- MLServer 的并行工作进程未正确设置。\r\n- MLServer 访问日志出现大量输出，造成日志泛滥。\r\n\r\n错误修复详情：\r\n\r\n如果 Kafka 集群处于不健康状态，且 `model-gateway` 等组件无法连接到 Broker，流水线就会处于失败状态。调度器会收到这一通知，但当 Kafka 集群恢复健康后，它并不会尝试在必要的服务上重新创建该流水线。此次修复将按照配置的周期性间隔，自动重试创建或删除流水线。这由调度器上的两个环境变量控制：\r\n\r\n- `RETRY_CREATING_FAILED_PIPELINES_TICK` 默认值为 `60s`，表示调度器每隔多久尝试一次创建失败的流水线；\r\n- `RETRY_DELETING_FAILED_PIPELINES_TICK` 默认值为 `60s`，表示调度器每隔多久尝试一次删除失败的流水线；\r\n- `MAX_RETRY_FAILED_PIPELINES` 默认值为 `10`，表示调度器最多重试次数。\r\n\r\n在尝试发送数据时，gRPC 流未能得到妥善处理。Go 上下文未检查接收方是否已关闭流，这导致了阻塞问题：调度器试图在代理上加载模型，但此时代理已经关闭了流，从而阻止了调度器继续执行其他模型加载操作。我们还发现类似的问题出现在其他多个场景中。现在，在尝试发送数据之前，我们会先验证流是否仍然处于活动状态。\r\n\r\n由于各种原因（如网络连接不良、调度器因存活探针失败而重启等），Operator 有时需要多次重试向调度器发送集群状态。如果这种情况反复发生，Operator 可能会长时间持续重试，进而导致其无法对任何资源进行协调。为此，我们在指数退避重试机制中设置了最大重试次数限制，并对 Operator 的所有网络调用强制设置了超时时间，以解决这一问题。\r\n\r\n在配置 MLServer 的并行工作进程数量时，Helm Chart 设置了错误的环境变量。这会导致那些将该数值设置为大于 1 的客户（默认值为 1）面临延迟增加和吞吐量下降的问题。目前，Helm Chart 已改用正确的环境变量 `MLSERVER_PARALLEL_WORKERS`，其默认值仍为 1。请注意，此变量适用于 MLServer 版本 ≥ `1.1.0`。如果您使用的是低于该版本的 MLServer，则应手动设置环境变量 `MLSERVER_MODEL_PARALLEL_WORKERS`，因为 Helm Chart 已不再支持该变量。","2025-12-19T10:55:25",{"id":194,"version":195,"summary_zh":196,"released_at":197},53464,"v2.10.1","### 概述\n\nCore 2.10.1 是一个补丁版本，修复了一个自 2.9.0 版本以来一直存在、但从 2.10.0 版本开始对用户可见的重大部分调度错误。\n\n此外，我们还消除了调度器因等待从未创建的服务器副本连接而出现启动缓慢的一系列场景。这些场景通常是由于配置错误引起的，例如在已部署了引用该配置且 `.spec.replicas > 0` 的 StatefulSet 类型的 `Servers` 后，再更新 `ServerConfig`。\n\n### 错误修复详情：\n\n从 2.10.0 版本开始，当托管模型的推理服务器 Pod 被重启后（无论原因如何），模型最终只会被调度到大约 `model.spec.minReplicas` 个服务器副本上，而不是请求的（也是预期的）`model.spec.replicas` 个副本。实际被调度的模型副本数量会因服务器副本在重启后与调度器建立连接的时间顺序而有所不同。\n\n这一回归问题的出现，是因为自 2.9.0 版本以来就存在的部分调度逻辑中的一个现有缺陷，在 2.10 版本中修复了一个与部分调度无关的数据竞争问题后，开始持续显现出来。在此之前，由于大多数集群运行场景下难以触发该数据竞争问题，因此用户并未遇到此问题。\n\n在 2.10.1 版本中，我们修复了根本性的问题，使部分调度能够按预期工作。\n\n### 从早期 Core 2 版本升级\n\n此补丁版本未引入任何 CRD 或配置变更。不过，如果您是从 2.10.0 之前的版本升级，建议您先阅读 [2.10.0 版本的发布说明](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Freleases\u002Ftag\u002Fv2.10.0)。\n\n### 更改日志\n\n由 [`auto-changelog`](https:\u002F\u002Fgithub.com\u002FCookPete\u002Fauto-changelog) 生成。\n\n#### [v2.10.1](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fcompare\u002Fv2.10.0...v2.10.1)\n\n> 2025年10月20日\n\n- 修复（operator）：错误的预期副本数通知 [`#6890`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F6890)\n- 修复（scheduler）：当 Model 的 minReplicas 被设置时，并非所有模型都会部署到 Servers 上 [`#6885`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F6885)","2025-10-20T16:20:24",{"id":199,"version":200,"summary_zh":201,"released_at":202},53465,"v2.10.0","### 概述\n\nCore 2.10.0 是一个包含重大新功能的版本，重点在于可扩展性、易用性和错误修复。\n\n### 从之前的 Core 2 版本升级\n\n- 所有 CRD 变更均保持与现有 CR 的向后兼容性。\n- 我们在 SeldonConfig 中引入了新的 Core 2 扩展配置选项（`config.ScalingConfig.*`），旨在集中管理 Core 2 的配置，并允许在 Core 2 集群部署后进行配置更改。为确保平稳过渡，部分配置选项将仅在后续版本中生效，但建议终端用户在升级到下一个版本（2.11）之前，将其设置为所需值。\n\n使用 Helm 升级时过程无缝，现有的 Helm 值将用于填充新的配置选项。如果不使用 Helm，之前的 SeldonConfig CR 仍然有效，但扩展配置将采用限制性的默认值。其中，`maxShardCountMultiplier` 参数[[文档](https:\u002F\u002Fdocs.seldon.ai\u002Fseldon-core-2\u002Fv2.10\u002Fuser-guide\u002Fperformance-tuning\u002Fpipelines\u002Fscalability-pipelines#id-1.-how-scaling-works-at-a-glance)]需要被设置，才能充分利用新的管道可扩展性特性。该参数可以随时更改，其值的变化将传播到所有使用该配置的组件。\n\n### 新特性\n\n- 管道可扩展性功能：所有管道组件（`dataflow-engine`、`pipelinegateway`、`modelgateway`）现在均可水平扩展，不再受限于每个主题的 Kafka 分区数量。[[文档](https:\u002F\u002Fdocs.seldon.ai\u002Fseldon-core-2\u002Fv2.10\u002Fuser-guide\u002Fperformance-tuning\u002Fpipelines\u002Fscalability-pipelines)]\n- 与 Kafka Schema Registry 集成：在部署管道时，提供对模型和管道输入\u002F输出主题的 Schema 合约的可见性，从而将 Core 2 管道与更广泛的 Kafka 生态系统连接起来。此功能有助于与 Kafka Connect 和 ksqlDB 等产品集成，以构建针对机器学习工作流定制的数据流、处理和日志记录解决方案。[[文档](https:\u002F\u002Fdocs.seldon.ai\u002Fseldon-core-2\u002Fintegrations\u002Fconfluent\u002Fschema-registry)]\n- 新的翻译层：可在 OpenAI API REST 与 OIP 之间进行转换。这使得在与通过 Seldon LLM 模块部署的 LLM 模型通信时，可以使用标准的 OpenAI 库和客户端。\n\n#### 实验性功能（早期预览，尚未达到生产就绪状态）\n- 将推理服务器配置为 Kubernetes Deployment 而非 StatefulSet。\n\n### 易用性改进\n\n- 管道控制平面现在对中断更具鲁棒性，细粒度的状态更新会传播到 Pipeline CR 的状态中。\n- 管道数据平面在组件重启后恢复速度更快。\n- 消除了推理服务器副本重启期间的停机原因，并实现了所有组件更为优雅的关闭流程。\n- 所有 Core 2 组件现在都具有关联…","2025-10-08T02:25:52",{"id":204,"version":205,"summary_zh":206,"released_at":207},53466,"v2.9.1","### 概述\n\nCore 2.9.1 是一个以修复 bug 和提升安全性为主的补丁版本。同时，我们还引入了一系列与循环流水线、易用性和成本效益相关的重要功能：\n\n### Bug 修复\n\n* 允许 Core 2 稳定地与 Pod 中断预算（PDB）协同工作 (#6560)。此前，终止或排空中的 Pod 仍显示 `Ready: True`，这导致从 PDB 的角度来看，这些 Pod 仍然被视为未中断。\n* 使 `model-gateway` 的请求超时时间可配置（之前为 10 分钟，新默认值为 2 分钟）(#6522)。这是为了解决特定场景下推理请求压垮可用的推理服务器副本，从而导致显著延迟增加的问题。此前，在这种情况下，Kafka 模型输入主题中会积压大量未处理的条目，进而导致新请求总是超时（因为在负载均衡器层面，队列中已有足够多的请求先于它们到达超时时间）。新的默认值对于大多数用例来说可能偏高，但为了不阻止慢速模型的推理，我们将其设置得较为保守。不过，您仍可以在 `model-gateway` Pod 级别通过 `MODELGATEWAY_WORKER_TIMEOUT_MS` 环境变量，根据自身的工作负载调整超时时间。\n* 修复了 `dataflow-engine` 在与调度器通信时的边缘情况 (#6506)。在存在多个 `dataflow-engine` 副本，或者单个副本正在重启的情况下，流水线的状态会基于最后接收到的消息来设定。有时，由于锁机制的实现方式，已终止的 `dataflow-engine` 副本发送的状态更新，会在 Kubernetes 启动用于替换它的新副本之后才被处理。这就导致流水线以非确定性的方式进入 `PipelineTerminated` 状态。\n\n### 新特性\n\n* 允许流水线包含循环，并限制迭代次数 (#6413, #6480，[文档](https:\u002F\u002Fdocs.seldon.ai\u002Fseldon-core-2\u002Fuser-guide\u002Fexamples\u002Fpipeline-cyclic))。此功能可通过 CR 中新增的 `spec.allowCycles` 字段，按流水线粒度进行启用。\n* 允许将 Core 2 Operator 安装在其各自的命名空间内，但管理 Core 2 CR 的范围可以扩展到其他多个命名空间 (#6434，[文档](https:\u002F\u002Fdocs.seldon.ai\u002Fseldon-core-2\u002Finstallation\u002Fproduction-environment))。\n* 允许终端用户在删除模型或流水线时，一并删除与其关联的 Kafka 主题 (#6353, #6383，[文档](https:\u002F\u002Fdocs.seldon.ai\u002Fseldon-core-2\u002Finstallation\u002Fadvanced-configurations\u002Fmanaging-kafka-topics))。使用此功能时需谨慎，因为它会导致无法再观测历史推理请求和响应数据，而这些数据可能正是针对流水线中的某些模型所收集的。\n\n### 文档改进\n\n* 扩展了关于 [自动伸缩](https:\u002F\u002Fdocs.seldon.ai\u002Fseldon-core-2\u002Fuser-guide\u002Fscaling) 和 [性能调优](https:\u002F\u002Fdocs.seldon.ai\u002Fseldon-core-2\u002Fuser-guide\u002Fperformance-tuning) 的文档。\n* [快速入门指南](https:\u002F\u002Fdocs.seldon.ai\u002Fseldon-core-2\u002Fuser-guide\u002Fquickstart)\n\n### Core 2 发布镜像\n\n* Core 2 镜像","2025-07-09T14:02:10",{"id":209,"version":210,"summary_zh":211,"released_at":212},53467,"v2.9.0","### 概述\n\nCore 2.9 是一个功能丰富的版本，旨在提升易用性、通过自动伸缩和调度改进简化运维，并支持流式用例（通过为 REST 和 gRPC 客户端提供模型响应流式传输）。\n\nCore 2 还推出了全新的 [文档](https:\u002F\u002Fdocs.seldon.ai\u002Fseldon-core-2)，内容和结构均经过重新设计。文档将持续优化，以覆盖高级配置和用例。\n\n### CRD 更新：\n\n本版本中的所有 CRD 变更均保持向后兼容性，因此包含现有 CR 的集群可以无缝迁移。\n\n* 在 Model CRD 中新增 `status.availableReplicas` 字段 (#5873)。这是部分调度功能的一部分。该字段不由最终用户直接设置，而是由 Seldon Kubernetes Operator 自动更新。\n* 在 Model CRD 中新增 `spec.llm` 字段 (#6234)。该字段由 Seldon LLM 模块中的 PromptRuntime 使用，用于引用 LLM 模型。在任何给定时间，`spec.llm` 和 `spec.explainer` 只能设置其中一个。这允许部署多个充当同一 LLM 提示生成器的“模型”。\n\n### （主要）特性：\n\n* 我们为支持流式的 MLServer 模型新增了 REST（通过 [SSE](https:\u002F\u002Fhtml.spec.whatwg.org\u002Fmultipage\u002Fserver-sent-events.html)）和 gRPC 的 **推理响应流式传输** 支持 (#6293, #6292)。这要求 MLServer 版本 ≥ 1.6.0。\n* 我们引入了 **模型副本的部分调度** (#6221, [文档](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fblob\u002Frelease-2.9\u002Fdocs-gb\u002Fmodels\u002Fscheduling.md))，从而改善 Core 2 在自动伸缩过程中的行为。借助此新功能，Core 2 调度器将尽可能多地加载请求的模型副本，即使没有推理服务器具备满足该请求所需的足够副本数。\n\t**部分调度仅在最终用户在模型清单中提供 `spec.minReplicas` 时生效**（作为判断模型是否“可用”的用户指定最小值），并且当存在至少具有该数量副本的合适推理服务器时才会起作用。使用部分调度，模型的状态可以是：\n\t- 完全调度：`spec.replicas == status.availableReplicas`；`ModelReady` 条件为 `True`，消息为 `ModelAvailable`。所有请求的副本均可处理推理请求。\n\t- 部分调度：`status.availableReplicas >= spec.minReplicas` 但 `status.availableReplicas \u003C spec.replicas`；`ModelReady` 条件为 `True`，消息为 `ModelAvailable`。Core 2 未能找到足够的服务器副本以加载该模型的所有请求副本。此状态可能是暂时的，例如当新的服务器副本正在创建但尚未就绪时。此时，可用的模型副本仍可处理推理请求。\n\t- 无法调度：未找到具备不少于模型 `spec.minReplicas` 副本数的合适推理服务器。`ModelReady` 条件为 `False`，消息为 `ScheduleFailed`。尽管如此，部分模型副本可能仍然可用于处理推理请求（例如……","2025-04-07T23:21:13",{"id":214,"version":215,"summary_zh":216,"released_at":217},53468,"v2.8.5","### 概述\n\n本次发布引入了与从旧版本升级以及在网络问题发生时通过保活模式优雅恢复相关的稳定性修复。此外，还增加了一些系统可用性方面的改进。\n\n我们还优化了关于使用 [HPA 自动扩缩容](https:\u002F\u002Fdocs.seldon.ai\u002Fseldon-core-2\u002Fkubernetes\u002Fhpa-rps-autoscaling) 和 Core 2 [架构](https:\u002F\u002Fdocs.seldon.ai\u002Fseldon-core-2\u002Farchitecture) 的文档。\n\n### （主要）特性：\n\n- 添加保活设置（服务器端：[`#6016`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F6016)，客户端：[`#5978`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5978)）\n- 公开有状态集的持久卷声明保留策略 [`#5946`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5946)\n- 为已删除的流水线和实验设置生存时间（TTL）[`#5948`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5948)\n- 改进控制平面同步 [`#6029`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F6029)、[`#6021`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F6021) 和 [`#6020`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F6020)\n\n### （主要）已修复的缺陷：\n\n- 将 `status.selector` 字段设为可选 [`#5985`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5985)\n- 关闭 Envoy 管理接口 [`#5936`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5936)\n\n### 更改日志\n\n本项目的所有重要更改都将记录在此文件中。日期以 UTC 时间显示。\n\n由 [`auto-changelog`](https:\u002F\u002Fgithub.com\u002FCookPete\u002Fauto-changelog) 生成。\n\n#### [v2.8.5](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fcompare\u002Fv2.8.4...v2.8.5)\n\n> 2024年11月7日\n\n- fix(ci): v2 版本针对 2.8.5 版本的变更 [`#6034`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F6034)\n- 为所有 Envoy 配置添加保活设置 [`#6033`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F6033)\n- fix(ci): 合并 v2 版本针对 2.8.5 版本的变更（4）[`#6032`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F6032)\n- fix: 使用生成 ID 来引导版本 ID [`#6029`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F6029)\n- feat(docs): 在 HPA 文档中添加更多配置细节 [`#6019`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F6019)\n- fix(scheduler): 始终为已删除的模型发送模型事件 [`#5992`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5992)\n- feat(scheduler): 将已删除资源的 TTL 作为 CLI 参数公开 [`#5994`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5994)\n- fix(ci): v2 版本针对 2.8.5 版本的变更（3）[`#6023`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F6023)\n- fix(dataflow): 使每个副本使用唯一的订阅名称 [`#6021`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F6021)\n- fix(ci): 修复不稳定测试 [`#6022`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F6022)\n- fix(agent): 允许按实例对服务器连接进行串行排序 [`#6020`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F6020)\n- fix(ci): v2 版本针对 2.8.5 版本的变更（2）[`#6018`](https:","2024-11-07T16:16:27",{"id":219,"version":220,"summary_zh":221,"released_at":222},53469,"v2.8.4","### 概述\n\n本次发布引入了多项功能，帮助用户在动态负载下运行 Core 2。我们在单模型服务场景中添加了对 HPA 模型和基于自定义指标的服务器自动扩缩容的支持。展示如何根据模型 RPS 进行扩缩容的文档已在此处说明：[这里](https:\u002F\u002Fdocs.seldon.ai\u002Fseldon-core-2\u002Fkubernetes\u002Fhpa-rps-autoscaling)。\n\n此外，本次发布还降低了控制平面出现问题时（特别是 `seldon-scheduler` 重启时）数据平面中断的风险。这是通过将调度器的（重新）启动过程与系统其他部分同步来实现的。\n\n我们还新增了多项功能，其中最值得注意的是，用户现在可以将模型升级到正在进行的实验中的新版本。\n\n目前，文档正在迁移到新的站点 [这里](https:\u002F\u002Fdocs.seldon.ai\u002Fseldon-core-2)（实验性）。\n\n### 主要功能：\n\n- 添加 GitBook 支持（实验性）[`#5943`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5943)\n- 向 `seldonconfig` 组件添加元数据对象 [`#5918`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5918)\n- `seldon-scheduler` 启动同步 [`#5930`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5930)\n- Envoy Grafana 仪表板 [`#5894`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5894)\n- 升级实验中的模型 [`#5874`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5874)\n- 添加流水线和模型名称验证 [`#5872`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5872)\n- 添加重试参数 [`#5875`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5875)\n\n### 主要修复的 Bug：\n\n- 在 `seldon-agent` 中跳过乱序的控制消息 [`#5969`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5969)\n- 将 `StatefulSet.Spec.Replicas` 用作服务器状态的参考 [`#5945`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5945)\n- 清理无法加载的模型 [`#5857`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5857) 和 [`#5830`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5830)\n\n### 更改日志\n\n本项目的所有重要更改都将记录在此文件中。日期以 UTC 时间显示。\n\n由 [`auto-changelog`](https:\u002F\u002Fgithub.com\u002FCookPete\u002Fauto-changelog) 生成。\n\n#### [v2.8.4](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fcompare\u002Fv2.8.3...v2.8.4)\n\n> 2024年10月11日\n\n- 修复(ci): v2 版本为 2.8.4 版本所做的更改（5）[`#5970`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5970)\n- 修复(docs): 恢复 changelog.md 的更改 [`#5971`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5971)\n- 修复(agent): 跳过代理中的乱序控制消息 [`#5969`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5969)\n- 更新 README.md [`#5955`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5955)\n- 修复(docs): 将名称从 core v2 改为 Core 2 [`#5963`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5963)\n- 修复(ci): v2 版本为 2.8.4 版本所做的更改（4）[`#5968`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5968)\n- 增加排水宽限期等待时间 [`#5967`]","2024-10-11T15:53:32",{"id":224,"version":225,"summary_zh":226,"released_at":227},53470,"v2.8.3","### 概述\n\n本次发布旨在修复实验工作流，使系统在组件（特别是 `seldon-scheduler`）重启时能够正确处理已删除的实验。作为此次改进的一部分，系统现在可以：\n\n- 处理不同版本的 `seldon-scheduler` 本地嵌入式数据库（BadgerDB），并支持无缝迁移到最新版本。\n- 在本地嵌入式数据库损坏或丢失的情况下，从 Kubernetes 中恢复调度器的状态，包括恢复 Kubernetes 中存在的实验和流水线。\n\n### （主要）已修复的缺陷：\n\n- scheduler\u002Foperator：在从缓存恢复时处理已删除的实验 [`#5726`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5726)\n- scheduler：不在状态报告中反馈正在排水的服务器 [`#5761`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5761)\n\n### （主要）新增功能：\n\n- operator：为 Seldon 特有的 CR 添加自定义打印列 [`#5736`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5736)\n\n### （主要）升级内容：\n\n- 使用 mlserver 1.6.0 [`#5724`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5724)\n\n### 更改日志\n\n本项目的所有重要更改都将记录在此文件中。日期以 UTC 时间显示。\n\n由 [`auto-changelog`](https:\u002F\u002Fgithub.com\u002FCookPete\u002Fauto-changelog) 自动生成。\n\n#### [v2.8.3](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fcompare\u002Fv2.8.2...v2.8.3)\n\n> 2024年7月17日\n\n- ci：合并 v2 的变更以用于 2.8.3 版本的发布 [`#5762`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5762)\n- fix(scheduler)：不在状态报告中反馈正在排水的服务器 [`#5761`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5761)\n- ci：来自 `v2` 的变更，用于 `2.8.3` 版本的发布 [`#5751`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5751)\n- fix：在从缓存恢复时处理已删除的实验 [`#5726`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5726)\n- fix(ansible)：显式添加对 community.docker 集合的依赖 [`#5746`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5746)\n- fix(golang-lint)：更新 lint 配置文件，移除已弃用的设置 [`#5747`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5747)\n- build(grpc)：更新负责代码生成的 Makefile [`#5749`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5749)\n- 将 \u002Fscheduler 中的 ubi9\u002Fopenjdk-17-runtime 从 1.18 升级到 1.20 [`#5666`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5666)\n- 将 \u002Fscheduler 中的 github.com\u002Fgo-playground\u002Fvalidator\u002Fv10 升级 [`#5712`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5712)\n- 将 \u002Fscheduler\u002Fdata-flow 中的 org.junit.jupiter:junit-jupiter-params 升级 [`#5729`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5729)\n- 重新生成许可证信息 [`#5737`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5737)\n- feat(operator)：为 Seldon 特有的 CR 添加自定义打印列 [`#5736`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5736)\n- 重新生成许可证信息 [`#5735`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5735)\n- 重新生成许可证信息 [`#5734`](https:\u002F\u002Fgi","2024-07-17T11:23:52",{"id":229,"version":230,"summary_zh":231,"released_at":232},53471,"v2.8.2","### Bugs Fixed:\r\n\r\n- Pipeline terminating fix on rebalance [`#5703`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5703)\r\n- Do not create new KafkaStreams app for existing pipelines [`#5550`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5550)\r\n- Handle unload too quick after load [`#5504`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5504)\r\n\r\n### Features:\r\n\r\n-  Add pipeline version to Kafka headers [`#5493`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5493)\r\n\r\n### Changelog\r\n\r\nAll notable changes to this project will be documented in this file. Dates are displayed in UTC.\r\n\r\nGenerated by [`auto-changelog`](https:\u002F\u002Fgithub.com\u002FCookPete\u002Fauto-changelog).\r\n\r\n#### [v2.8.2](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fcompare\u002Fv2.8.1...v2.8.2)\r\n\r\n> 26 June 2024\r\n\r\n- ci: Merge changes from v2 to release 2.8 branch [`#5705`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5705)\r\n- fix(scheduler): Pipeline terminating fix on rebalance [`#5703`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5703)\r\n- Bump ubi9\u002Fubi-micro from 9.4-6.1716471860 to 9.4-9 in \u002Foperator [`#5682`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5682)\r\n- Bump ubi9\u002Fubi-micro from 9.4-6.1716471860 to 9.4-9 in \u002Fhodometer [`#5687`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5687)\r\n- Bump github.com\u002Fgo-logr\u002Flogr from 1.4.1 to 1.4.2 in \u002Foperator [`#5686`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5686)\r\n- Bump google.golang.org\u002Fprotobuf from 1.34.1 to 1.34.2 in \u002Fscheduler [`#5691`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5691)\r\n- Bump ubi9\u002Fubi-micro from 9.4-6.1716471860 to 9.4-9 in \u002Fscheduler [`#5695`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5695)\r\n- Bump ubi9\u002Fubi-minimal from 9.4-949.1717074713 to 9.4-1134 in \u002Fscheduler [`#5696`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5696)\r\n- Bump rclone\u002Frclone from 1.66.0 to 1.67.0 in \u002Fscheduler [`#5697`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5697)\r\n- Bump ubi9\u002Fubi-minimal in \u002Fscheduler [`#5667`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5667)\r\n- Bump envoyproxy\u002Fenvoy from v1.30.1 to v1.30.2 in \u002Fscheduler [`#5668`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5668)\r\n- Bump go.opentelemetry.io\u002Fotel\u002Fsdk from 1.26.0 to 1.27.0 in \u002Fscheduler [`#5677`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5677)\r\n- Bump go.opentelemetry.io\u002Fcontrib\u002Finstrumentation\u002Fgithub.com\u002Fgorilla\u002Fmux\u002Fotelmux [`#5679`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5679)\r\n- Bump ubi9\u002Fubi-micro from 9.4-6 to 9.4-6.1716471860 in \u002Fhodometer [`#5654`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5654)\r\n- Bump go.opentelemetry.io\u002Fotel\u002Ftrace from 1.26.0 to 1.27.0 in \u002Fscheduler [`#5640`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5640)\r\n- Bump grafana\u002Fgrafana from 10.4.3 to 11.0.0 in \u002Fscheduler [`#5644`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5644)\r\n- Bump ubi9\u002Fubi-micro from 9.4-6 to 9.4-6.1716471860 in \u002Foperator [`#5656`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5656)\r\n- Bump github.com\u002Fhashicorp\u002Fgo-retryablehttp in \u002Fhodometer [`#5657`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5657)\r\n- Bump ubi9\u002Fubi-minimal in \u002Fscheduler [`#5659`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5659)\r\n- Bump ubi9\u002Fubi-micro from 9.4-6 to 9.4-6.1716471860 in \u002Fscheduler [`#5660`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5660)\r\n- bump(deps): major version updates in \u002Fscheduler\u002Fdata-flow [`#5639`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5639)\r\n- fix(ansible): allow multiple custom secrets without namespaces [`#5638`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5638)\r\n- feat: Add pipelines to core2 qa control plane tests [`#5636`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5636)\r\n- test(k6): Add k8s-based control plane tests  [`#5609`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5609)\r\n- Bump go.opentelemetry.io\u002Fcontrib\u002Finstrumentation\u002Fgithub.com\u002Fgorilla\u002Fmux\u002Fotelmux [`#5627`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5627)\r\n- Bump google.golang.org\u002Fprotobuf from 1.34.0 to 1.34.1 in \u002Fscheduler [`#5628`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5628)\r\n- Bump grafana\u002Fgrafana from 10.4.2 to 10.4.3 in \u002Fscheduler [`#5611`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5611)\r\n- build(ansible): add options to install grafana and dashboards [`#5589`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5589)\r\n- fix(deps): Bump org.apache.kafka:kafka-streams-test-utils from 7.6.0-ccs to 7.6.1-ccs in \u002Fscheduler\u002Fdata-flow [`#5538`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5538)\r\n- Bump google.golang.org\u002Fprotobuf from 1.33.0 to 1.34.1 in \u002Foperator [`#5590`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5590)\r\n- Bump github.com\u002Fconfluentinc\u002Fconfluent-kafka-go\u002Fv2 in \u002Foperator [`#5591`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5591)\r\n- Bump github.com\u002Fconfluentinc\u002Fconfluent-kafka-go\u002Fv2 in \u002Fscheduler [`#5596`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5596)\r\n- docs: Reference MLServer `infer` deprecation [`#5606`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5606)\r\n- Bump go.opentelemetry.io\u002Fotel\u002Fexporters\u002Fotlp\u002Fotlptrace in \u002Fscheduler [`#","2024-06-26T11:28:17",{"id":234,"version":235,"summary_zh":236,"released_at":237},53472,"v1.18.2","## What's Changed\r\n* docs(openapi): Use command with file by @jesse-c in https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5349\r\n* fix(logging): adserver is very noisy for feedback requests, possibly affecting performance by @majolo in https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5351\r\n* feat(build): use github actions to build docker images by @RafalSkolasinski in https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5380\r\n* fix(build): fix pipeline by @RafalSkolasinski in https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5389\r\n* Create license.yml dummy for v2 by @sakoush in https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5466\r\n* add licensing update to contributing.md by @paulb-seldon in https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5510\r\n* Update oss_commits.md by @paulb-seldon in https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5529\r\n* Feature: support multi hosts in vs by @ooooona in https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5430\r\n* Removed Seldon Deploy and added Seldon Enterprise Platform where necessary by @ramonpzg in https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5552\r\n* add generated dependencies licence update by @gsunner in https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5637\r\n* fix(helm): helm-docs installation + version bump by @RafalSkolasinski in https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5635\r\n\r\n## New Contributors\r\n* @ooooona made their first contribution in https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5430\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fcompare\u002Fv1.18.1...v1.18.2","2024-06-05T10:53:25",{"id":239,"version":240,"summary_zh":241,"released_at":242},53473,"v2.8.1","### Changelog\r\n\r\nAll notable changes to this project will be documented in this file. Dates are displayed in UTC.\r\n\r\nGenerated by [`auto-changelog`](https:\u002F\u002Fgithub.com\u002FCookPete\u002Fauto-changelog).\r\n\r\n#### [v2.8.1](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fcompare\u002Fv2.8.0...v2.8.1)\r\n\r\n> 21 March 2024\r\n\r\n- ci: Merge changes from v2 to release 2.8 branch (for 2.8.1) [`#5470`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5470)\r\n- do not mark source in terminate events of old pipelines [`#5469`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5469)\r\n- Re-generate license info [`#5468`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5468)\r\n- feat(ci): Add license workflow [`#5465`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5465)\r\n- ci: Merge changes from v2 to release 2.8 branch (for 2.8.1) [`#5467`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5467)\r\n- fix: update 3rd party licenses [`#5462`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5462)\r\n- ci: Merge changes from v2 to release 2.8 branch (for 2.8.1) [`#5460`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5460)\r\n- fix(dataflow) CVEs related to com.microsoft.azure:adal4j [`#5458`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5458)\r\n- Bump ubi9\u002Fubi-micro from 9.3-13 to 9.3-15 in \u002Fscheduler [`#5457`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5457)\r\n- fix: Add pipelines in stress test [`#5437`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5437)\r\n- ci: Merge changes from v2 to release 2.8 branch (for 2.8.1) [`#5436`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5436)\r\n- fix(scheduler): Introduce `sendMsg` timeout for grpc streams (scheduler -> controller) [`#5434`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5434)\r\n- build(deps): bump dataflow dependency versions [`#5429`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5429)\r\n- Bump ubi9\u002Fopenjdk-17-runtime from 1.17 to 1.18 in \u002Fscheduler [`#5401`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5401)\r\n- fix(modelgateway): KafkaAdmin error when fetching OIDC token [`#5428`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5428)\r\n- fix(modelgateway): kafka topics with correct number of partitions [`#5427`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5427)\r\n- Bump github.com\u002Fsignalfx\u002Fsplunk-otel-go\u002Finstrumentation\u002Fgithub.com\u002Fconfluentinc\u002Fconfluent-kafka-go\u002Fv2\u002Fkafka\u002Fsplunkkafka [`#5420`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5420)\r\n- Bump google.golang.org\u002Fprotobuf from 1.32.0 to 1.33.0 in \u002Fscheduler [`#5421`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5421)\r\n- Bump github.com\u002Fgolang\u002Fprotobuf from 1.5.3 to 1.5.4 in \u002Fscheduler [`#5422`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5422)\r\n- Bump google.golang.org\u002Fprotobuf from 1.32.0 to 1.33.0 in \u002Foperator [`#5424`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5424)\r\n- Bump rclone\u002Frclone from 1.65.2 to 1.66.0 in \u002Fscheduler [`#5410`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5410)\r\n- Bump google.golang.org\u002Fgrpc from 1.62.0 to 1.62.1 in \u002Fhodometer [`#5412`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5412)\r\n- Bump ubi9\u002Fubi-minimal from 9.3-1552 to 9.3-1612 in \u002Fscheduler [`#5409`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5409)\r\n- Bump envoyproxy\u002Fenvoy from v1.29.1 to v1.29.2 in \u002Fscheduler [`#5408`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5408)\r\n- Bump grafana\u002Fgrafana from 10.3.3 to 10.4.0 in \u002Fscheduler [`#5407`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5407)\r\n- Bump ubi9\u002Fubi-micro from 9.3-13 to 9.3-15 in \u002Foperator [`#5417`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5417)\r\n- Bump ubi9\u002Fubi-micro from 9.3-13 to 9.3-15 in \u002Fhodometer [`#5418`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5418)\r\n- Bump google.golang.org\u002Fgrpc from 1.62.0 to 1.62.1 in \u002Fscheduler [`#5423`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5423)\r\n- Bump google.golang.org\u002Fgrpc from 1.62.0 to 1.62.1 in \u002Foperator [`#5425`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5425)\r\n- fix(controller): reload models upon reconnect to the scheduler [`#5411`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5411)\r\n- Bump go.opentelemetry.io\u002Fotel\u002Fexporters\u002Fotlp\u002Fotlptrace\u002Fotlptracegrpc [`#5395`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5395)\r\n- fix(scheduler): Increase event hub buffer inline with other components [`#5406`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5406)\r\n- fix(dataflow): catch all exceptions when creating a KafkaStreams object [`#5405`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5405)\r\n- docs(SeldonRuntime): Add an example for overriding PodSpec in SeldonRuntime [`#5404`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5404)\r\n- fix: Use mlserver 1.5.0 [`#5403`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5403)\r\n- Bump google.golang.org\u002Fgrpc from 1.61.1 to 1.62.0 in \u002Fscheduler [`#5393`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5393)\r\n- Bump github.com\u002Fgo-playground\u002Fvalidator\u002Fv10 in \u002Fscheduler [`#5392`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5392)\r\n- Bump go.opentelemetry.io\u002Fcontrib\u002Finstrumentation\u002Fgoogle.golang.org\u002Fgrpc\u002Fotelgrpc [`#5394`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseld","2024-03-21T16:18:59",{"id":244,"version":245,"summary_zh":246,"released_at":247},53474,"v2.8.0","### Changelog\r\n\r\nAll notable changes to this project will be documented in this file. Dates are displayed in UTC.\r\n\r\nGenerated by [`auto-changelog`](https:\u002F\u002Fgithub.com\u002FCookPete\u002Fauto-changelog).\r\n\r\n#### [v2.8.0](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fcompare\u002Fv2.7.0...v2.8.0)\r\n\r\n> 28 February 2024\r\n\r\n- ci: Merge changes from v2 to release 2.8 branch [`#5379`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5379)\r\n- update 3rd party licenses [`#5377`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5377)\r\n- fix(gateway): wait for kafka topic creation [`#5359`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5359)\r\n- fix(dataflow): wait for kafka topic creation [`#5375`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5375)\r\n- build(deps) update dataflow dependencies [`#5360`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5360)\r\n- fix(dataflow): handle pipeline errors and clear kafka streams state [`#5358`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5358)\r\n- use k8s 1.29.2 in kind [`#5357`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5357)\r\n- update Inactive check [`#5355`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5355)\r\n- Bump github.com\u002Fgo-playground\u002Fvalidator\u002Fv10 in \u002Fscheduler [`#5327`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5327)\r\n- fix(dataflow): set default OTEL_EXPORTER_OTLP_PROTOCOL in compose setup [`#5353`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5353)\r\n- Bump google.golang.org\u002Fgrpc from 1.61.0 to 1.61.1 in \u002Fscheduler [`#5326`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5326)\r\n- Bump grafana\u002Fgrafana from 10.3.1 to 10.3.3 in \u002Fscheduler [`#5328`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5328)\r\n- fix(scheduler): Send server statuses on controller reconnect [`#5350`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5350)\r\n- Bump google.golang.org\u002Fgrpc from 1.61.0 to 1.61.1 in \u002Fhodometer [`#5347`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5347)\r\n- bump(librdkafka): from v1.9.2 to v2.3.0 [`#5321`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5321)\r\n- fix(agent): set context deadline for grpc model server control plane [`#5329`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5329)\r\n- fix: Experiments and Models state fixes when reconnecting to scheduler [`#5320`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5320)\r\n- enable isotime in the logs [`#5319`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5319)\r\n- feat(operator): Expose PodSpec in OverrideSpec for SeldonRuntime [`#5296`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5296)\r\n- Bump go.opentelemetry.io\u002Fotel\u002Fsdk from 1.22.0 to 1.23.1 in \u002Fscheduler [`#5311`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5311)\r\n- Bump go.opentelemetry.io\u002Fotel\u002Fexporters\u002Fotlp\u002Fotlptrace in \u002Fscheduler [`#5312`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5312)\r\n- Bump go.opentelemetry.io\u002Fcontrib\u002Finstrumentation\u002Fgithub.com\u002Fgorilla\u002Fmux\u002Fotelmux [`#5310`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5310)\r\n- Bump github.com\u002Fsignalfx\u002Fsplunk-otel-go\u002Finstrumentation\u002Fgithub.com\u002Fconfluentinc\u002Fconfluent-kafka-go\u002Fkafka\u002Fsplunkkafka [`#5313`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5313)\r\n- Bump go.opentelemetry.io\u002Fcontrib\u002Finstrumentation\u002Fnet\u002Fhttp\u002Fotelhttp [`#5314`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5314)\r\n- fix: Add kubectl option to pipeline-tests script [`#5307`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5307)\r\n- fix(logging): Update otel and tracing config [`#5291`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5291)\r\n- fix: Pipeline state during disconnects [`#5298`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5298)\r\n- fix: Connection retries when scheduler restarts for dataflow and controller [`#5292`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5292)\r\n- fix(pipelinegateway): Use composite key for Kafka when x-request-id header is specified [`#5275`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5275)\r\n- build(deps): Update dataflow dependencies [`#5278`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5278)\r\n- add sleep to script [`#5277`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5277)\r\n- Bump google.golang.org\u002Fgrpc from 1.56.3 to 1.61.0 in \u002Fcomponents\u002Ftls [`#5272`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5272)\r\n- Bump github.com\u002Fotiai10\u002Fcopy from 1.7.0 to 1.14.0 in \u002Fcomponents\u002Ftls [`#5274`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5274)\r\n- Bump github.com\u002Fgrpc-ecosystem\u002Fgo-grpc-middleware in \u002Fhodometer [`#5270`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5270)\r\n- Bump github.com\u002Fstretchr\u002Ftestify from 1.7.0 to 1.8.4 in \u002Fhodometer [`#5271`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5271)\r\n- Bump github.com\u002Fhashicorp\u002Fgo-retryablehttp in \u002Fhodometer [`#5269`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5269)\r\n- Bump google.golang.org\u002Fprotobuf from 1.31.0 to 1.32.0 in \u002Foperator [`#5268`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5268)\r\n- Bump github.com\u002Fgrpc-ecosystem\u002Fgo-grpc-middleware in \u002Foperator [`#5266`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5266)\r\n- Bump emperror.dev\u002Ferrors from 0.8.0 to 0.8.1 in \u002Foperator [`#5265`](https:\u002F\u002Fgithub","2024-02-28T16:34:44",{"id":249,"version":250,"summary_zh":251,"released_at":252},53475,"v1.18.1","All notable changes to this project will be documented in this file. Dates are displayed in UTC.\r\n\r\nGenerated by [`auto-changelog`](https:\u002F\u002Fgithub.com\u002FCookPete\u002Fauto-changelog).\r\n\r\n#### [v1.18.1](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fcompare\u002Fv1.18.0...v1.18.1)\r\n\r\n> 29 February 2024\r\n\r\n- fix pipeline [`#5389`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5389)\r\n- feat(build): use github actions to build docker images [`#5380`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5380)\r\n- fix(logging): adserver is very noisy for feedback requests, possibly affecting performance [`#5351`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5351)\r\n-  docs(openapi): Use command with file [`#5349`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5349)\r\n- fix(adserver): adserver returns cloudevents compatible response [`#5348`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5348)\r\n- fix(docs): Change V2 Inference Protocol to OIP [`#5297`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5297)\r\n- fix(docs): Update docs w latest K8s compatibility [`#5295`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5295)\r\n- Update README.md [`#5293`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5293)\r\n- build(ansible): fix link to minio chart [`#5276`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5276)\r\n- feat(ci): dependabot worfklow for v2 [`#5194`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5194)\r\n- rewording [`#5192`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5192)\r\n- Update index.rst [`#5191`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5191)\r\n- Update README.md [`#5190`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5190)\r\n- bump versions to 1.19.0-dev [`#7`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F7)\r\n- Release 1.18.0 [`9771f06`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fcommit\u002F9771f06bb5f0380f81531db5a3cb473b7a42e02c)","2024-02-29T16:44:51",{"id":254,"version":255,"summary_zh":256,"released_at":257},53476,"v2.8.0-rc2","### Changelog\r\n\r\nAll notable changes to this project will be documented in this file. Dates are displayed in UTC.\r\n\r\nGenerated by [`auto-changelog`](https:\u002F\u002Fgithub.com\u002FCookPete\u002Fauto-changelog).\r\n\r\n#### [v2.8.0-rc2](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fcompare\u002Fv2.8.0-rc1...v2.8.0-rc2)\r\n\r\n> 27 February 2024\r\n\r\n- ci: Merge changes from v2 to release 2.8 branch [`#5379`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5379)\r\n- update 3rd party licenses [`#5377`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5377)\r\n- fix(gateway): wait for kafka topic creation [`#5359`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5359)\r\n- fix(dataflow): wait for kafka topic creation [`#5375`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5375)\r\n- build(deps) update dataflow dependencies [`#5360`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5360)\r\n- fix(dataflow): handle pipeline errors and clear kafka streams state [`#5358`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5358)\r\n- use k8s 1.29.2 in kind [`#5357`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5357)\r\n- update Inactive check [`#5355`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5355)\r\n- Bump github.com\u002Fgo-playground\u002Fvalidator\u002Fv10 in \u002Fscheduler [`#5327`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5327)\r\n- fix(dataflow): set default OTEL_EXPORTER_OTLP_PROTOCOL in compose setup [`#5353`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5353)\r\n- Bump google.golang.org\u002Fgrpc from 1.61.0 to 1.61.1 in \u002Fscheduler [`#5326`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5326)\r\n- Bump grafana\u002Fgrafana from 10.3.1 to 10.3.3 in \u002Fscheduler [`#5328`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5328)\r\n- fix(scheduler): Send server statuses on controller reconnect [`#5350`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5350)\r\n- Bump google.golang.org\u002Fgrpc from 1.61.0 to 1.61.1 in \u002Fhodometer [`#5347`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5347)\r\n- bump(librdkafka): from v1.9.2 to v2.3.0 [`#5321`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5321)\r\n- Generating changelog for v2.8.0-rc2 [`d2f4375`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fcommit\u002Fd2f4375ceac57afa3466ef6aef58eef09fab6528)\r\n- Setting version for helm charts [`cdd1f3a`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fcommit\u002Fcdd1f3a759bfa1bceef5f7353ad60936e3a1fa10)\r\n- Setting version for yaml manifests [`89aa658`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fcommit\u002F89aa65822887a5c497dfe5e156b65010367e7fd3)\r\n","2024-02-27T11:23:54",{"id":259,"version":260,"summary_zh":261,"released_at":262},53477,"v2.8.0-rc1","### Changelog\r\n\r\nAll notable changes to this project will be documented in this file. Dates are displayed in UTC.\r\n\r\nGenerated by [`auto-changelog`](https:\u002F\u002Fgithub.com\u002FCookPete\u002Fauto-changelog).\r\n\r\n#### [v2.8.0-rc1](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fcompare\u002Fv2.7.0...v2.8.0-rc1)\r\n\r\n> 19 February 2024\r\n\r\n- fix(agent): set context deadline for grpc model server control plane [`#5329`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5329)\r\n- fix: Experiments and Models state fixes when reconnecting to scheduler [`#5320`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5320)\r\n- enable isotime in the logs [`#5319`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5319)\r\n- feat(operator): Expose PodSpec in OverrideSpec for SeldonRuntime [`#5296`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5296)\r\n- Bump go.opentelemetry.io\u002Fotel\u002Fsdk from 1.22.0 to 1.23.1 in \u002Fscheduler [`#5311`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5311)\r\n- Bump go.opentelemetry.io\u002Fotel\u002Fexporters\u002Fotlp\u002Fotlptrace in \u002Fscheduler [`#5312`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5312)\r\n- Bump go.opentelemetry.io\u002Fcontrib\u002Finstrumentation\u002Fgithub.com\u002Fgorilla\u002Fmux\u002Fotelmux [`#5310`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5310)\r\n- Bump github.com\u002Fsignalfx\u002Fsplunk-otel-go\u002Finstrumentation\u002Fgithub.com\u002Fconfluentinc\u002Fconfluent-kafka-go\u002Fkafka\u002Fsplunkkafka [`#5313`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5313)\r\n- Bump go.opentelemetry.io\u002Fcontrib\u002Finstrumentation\u002Fnet\u002Fhttp\u002Fotelhttp [`#5314`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5314)\r\n- fix: Add kubectl option to pipeline-tests script [`#5307`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5307)\r\n- fix(logging): Update otel and tracing config [`#5291`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5291)\r\n- fix: Pipeline state during disconnects [`#5298`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5298)\r\n- fix: Connection retries when scheduler restarts for dataflow and controller [`#5292`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5292)\r\n- fix(pipelinegateway): Use composite key for Kafka when x-request-id header is specified [`#5275`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5275)\r\n- build(deps): Update dataflow dependencies [`#5278`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5278)\r\n- add sleep to script [`#5277`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5277)\r\n- Bump google.golang.org\u002Fgrpc from 1.56.3 to 1.61.0 in \u002Fcomponents\u002Ftls [`#5272`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5272)\r\n- Bump github.com\u002Fotiai10\u002Fcopy from 1.7.0 to 1.14.0 in \u002Fcomponents\u002Ftls [`#5274`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5274)\r\n- Bump github.com\u002Fgrpc-ecosystem\u002Fgo-grpc-middleware in \u002Fhodometer [`#5270`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5270)\r\n- Bump github.com\u002Fstretchr\u002Ftestify from 1.7.0 to 1.8.4 in \u002Fhodometer [`#5271`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5271)\r\n- Bump github.com\u002Fhashicorp\u002Fgo-retryablehttp in \u002Fhodometer [`#5269`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5269)\r\n- Bump google.golang.org\u002Fprotobuf from 1.31.0 to 1.32.0 in \u002Foperator [`#5268`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5268)\r\n- Bump github.com\u002Fgrpc-ecosystem\u002Fgo-grpc-middleware in \u002Foperator [`#5266`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5266)\r\n- Bump emperror.dev\u002Ferrors from 0.8.0 to 0.8.1 in \u002Foperator [`#5265`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5265)\r\n- Bump github.com\u002Fonsi\u002Fgomega from 1.18.1 to 1.31.1 in \u002Foperator [`#5267`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5267)\r\n- Bump github.com\u002Fsirupsen\u002Flogrus from 1.8.1 to 1.9.3 in \u002Fscheduler [`#5264`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5264)\r\n- Bump github.com\u002Fgrpc-ecosystem\u002Fgo-grpc-middleware in \u002Fscheduler [`#5263`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5263)\r\n- Bump github.com\u002Frs\u002Fxid from 1.3.0 to 1.5.0 in \u002Fscheduler [`#5261`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5261)\r\n- Bump github.com\u002Fdgraph-io\u002Fbadger\u002Fv3 in \u002Fscheduler [`#5262`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5262)\r\n- Bump github.com\u002Fonsi\u002Fgomega from 1.18.1 to 1.31.1 in \u002Fscheduler [`#5256`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5256)\r\n- Bump go.opentelemetry.io\u002Fcontrib\u002Finstrumentation\u002Fgithub.com\u002Fgorilla\u002Fmux\u002Fotelmux [`#5258`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5258)\r\n- Bump go.opentelemetry.io\u002Fcontrib\u002Finstrumentation\u002Fgoogle.golang.org\u002Fgrpc\u002Fotelgrpc [`#5260`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5260)\r\n- Bump github.com\u002Fotiai10\u002Fcopy from 1.7.0 to 1.14.0 in \u002Fscheduler [`#5259`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5259)\r\n- Bump github.com\u002Ffsnotify\u002Ffsnotify from 1.5.1 to 1.7.0 in \u002Fscheduler [`#5253`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5253)\r\n- Bump go.opentelemetry.io\u002Fcontrib\u002Finstrumentation\u002Fnet\u002Fhttp\u002Fotelhttp [`#5254`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5254)\r\n- Bump github.com\u002Fgoogle\u002Fuuid from 1.4.0 to 1.6.0 in \u002Fscheduler [`#5255`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5255)\r\n- Bump github.com\u002Fenvoyproxy\u002Fgo-control-plane in \u002Fscheduler [`#5247`](https:\u002F\u002Fgithub.com\u002FSeld","2024-02-20T11:24:23",{"id":264,"version":265,"summary_zh":266,"released_at":267},53478,"v1.18.0","Release 1.18.0\r\n    \r\n- build(license): Change license to BSL 1.1 ([read announcement](https:\u002F\u002Fwww.seldon.io\u002Fstrengthening-our-commitment-to-open-core))\r\n- fix(CVEs): Fix Critical CVEs in Operator & Executor\r\n- fix(CVEs): Fix Critical CVEs in MLFlow Server\r\n- fix(CVEs): Fix Critical CVEs in Rclone Storage Initializer\r\n- fix(CVEs): Fix Critical CVEs in Alibi Explain & Alibi Detect Servers\r\n- fix(CVEs): Fix Critical CVEs in TFServing Proxy\r\n- feat(alibi-detect-server): Make Alibi Detect artifact folder configurable\r\n- build(controller): Update to latest version of controller-runtime\r\n- fix(Operator & Executor): Upgrade to UBI9 and use as default image\r\n- fix(MLFlow Server): Fix permission issue in s2i wrapper image\r\n- fix(MLFlow Server): Upgrade MLFlow to 2.9.2\r\n- fix(Alibi Servers): Upgrade transformers dependency to 4.36.2\r\n- fix(Alibi Explain Server): Check-in alibi explain protos into VCS\r\n- fix(examples): Fix MLFlow model uri in server examples\r\n- build(docs): Misc docs build \u002F lint improvements\r\n- build(docs): Re-generate OIP v2 README\r\n- build(docs): Use reproducible requirements.txt\r\n- fix(docs): Correct various issues for `linkcheck` target\r\n- fix(docs): Add to-do to match existing approach in other repositories\r\n- fix(docs): Add missing scalar types\r\n- fix(docs): Correct links and spelling mistakes\r\n- fix(docs): Generate missing code cell IDs\r\n- fix(docs): Upgrade packages\r\n- fix(docs): Add missing 1.17.x mentions\r\n- fix(docs): Suppress valid reference warnings","2024-01-22T14:02:25",{"id":269,"version":270,"summary_zh":271,"released_at":272},53479,"v2.7.0","### Changelog\r\n\r\nAll notable changes to this project will be documented in this file. Dates are displayed in UTC.\r\n\r\nGenerated by [`auto-changelog`](https:\u002F\u002Fgithub.com\u002FCookPete\u002Fauto-changelog).\r\n\r\n#### [v2.7.0](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fcompare\u002Fv2.6.0...v2.7.0)\r\n\r\n> 22 January 2024\r\n\r\n- build(license): Change license to BSL 1.1 ([read announcement](https:\u002F\u002Fwww.seldon.io\u002Fstrengthening-our-commitment-to-open-core))\r\n- fix(envoy): add readiness probe to Envoy [`#5158`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5158)\r\n- Initial commit to remove finalizer - also fixes name typo to fix SeldonRuntime Ready status for dataflow engine [`#5109`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5109)\r\n- fix: Start triton server via bash -c tritonserver instead of just tritonserver [`#5030`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5030)\r\n- fix(scheduler): Fix deleting models that are still progressing [`#5143`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5143)\r\n- fix(docs): Auto-detected typos [`#5135`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5135)\r\n- refactor(crds): Use built-in OpenAPI validation [`#5129`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5129)\r\n- fix(docs): Use consistent shell style [`#5133`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5133)\r\n- feat(kafka): Add support for SASL OAUTHBEARER mechanism for Kafka [`#5127`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5127)\r\n- fix(podmonitor): Pipelinegateway podmonitor label fix [`#5120`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5120)\r\n- build(kafka): change message.max.bytes in broker side to align with producer and consumer [`#5126`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5126)\r\n- V2D-1253 Use smaller model in demo but still reference bigger one [`#5115`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5115)\r\n- test: Ensure agent client starts [`#5112`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5112)\r\n- feat(docs): [SCv2] Add a section about loading custom HuggingFace models from Seldon CLI [`#5106`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5106)\r\n- build(dataflow): bump grpc-stub and grpc-protobuf to 1.57.2 [`#5110`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5110)\r\n- feat(scheduler): Match models requirements for servers not replicas & improve status handling [`#5107`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5107)\r\n- feat(docs): [SCv2] Automatically create and upload a custom HF model to seldon-models in GCS on every new MLServer version [`#5103`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5103)\r\n- ci(security): Remove linting steps for GHA security workflow [`#5102`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5102)\r\n- Add ns env var for local docker deployments [`#5101`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5101)\r\n- remove timoeut in envoy [`#5099`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5099)\r\n- feat: Allow kafka consumer group id prefix configuration [`#5072`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5072)\r\n- fix(scheduler): Manual trigger envoy update [`#5074`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5074)\r\n- feat(scheduler): Report lack of dataflow engines in pipeline statuses [`#5080`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5080)\r\n- adding sphinx-youtube extension and embedding intro video [`#5065`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5065)\r\n- fix(docs): add note that Basic tier in Event Hub does not support Kafka protocol [`#5018`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5018)\r\n- fix(operator): consistent pod service monitors reconcilor app labels [`#5073`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5073)\r\n- ci(github): Add GitHub PR template for Core v2 [`#5081`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5081)\r\n- fix(docs): typo in pipeline page [`#5056`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5056)\r\n- fix(manifests): rename components to allow install along core v1 [`#5055`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5055)\r\n- fix: Allow number of replicas in k6 tests [`#5053`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5053)\r\n- Add kafka producer compression example for configuration customization [`#5044`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5044)\r\n- Allow digits in topic prefix [`#5051`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5051)\r\n- fix(docs): Fix various broken links in docs [`#5047`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5047)\r\n- fix: hodometer enable flag and docs [`#5025`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5025)\r\n- fix helm docs runtime name typo [`#5019`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5019)\r\n- feat(helm): Make SeldonConfig configurable in SeldonRuntime Helm chart [`#5014`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5014)\r\n- Generating changelog for v2.7.0 [`227ed04`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fcommit\u002F227ed04eb2d6ae5c4d28369fd67a5f03c41b9a80)\r\n- feat: LPL release [`7a83250`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fcommit\u002F7a8325018bf323702dc6ef01594db5177c112c3d)\r\n- Gen","2024-01-22T14:02:17",{"id":274,"version":275,"summary_zh":276,"released_at":277},53480,"v2.7.0-rc1","### Changelog\r\n\r\nAll notable changes to this project will be documented in this file. Dates are displayed in UTC.\r\n\r\nGenerated by [`auto-changelog`](https:\u002F\u002Fgithub.com\u002FCookPete\u002Fauto-changelog).\r\n\r\n#### [v2.7.0-rc1](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fcompare\u002Fv2.6.0...v2.7.0-rc1)\r\n\r\n> 21 August 2023\r\n\r\n- remove timoeut in envoy [`#5099`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5099)\r\n- feat: Allow kafka consumer group id prefix configuration [`#5072`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5072)\r\n- fix(scheduler): Manual trigger envoy update [`#5074`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5074)\r\n- feat(scheduler): Report lack of dataflow engines in pipeline statuses [`#5080`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5080)\r\n- adding sphinx-youtube extension and embedding intro video [`#5065`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5065)\r\n- fix(docs): add note that Basic tier in Event Hub does not support Kafka protocol [`#5018`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5018)\r\n- fix(operator): consistent pod service monitors reconcilor app labels [`#5073`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5073)\r\n- ci(github): Add GitHub PR template for Core v2 [`#5081`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5081)\r\n- fix(docs): typo in pipeline page [`#5056`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5056)\r\n- fix(manifests): rename components to allow install along core v1 [`#5055`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5055)\r\n- fix: Allow number of replicas in k6 tests [`#5053`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5053)\r\n- Add kafka producer compression example for configuration customization [`#5044`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5044)\r\n- Allow digits in topic prefix [`#5051`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5051)\r\n- fix(docs): Fix various broken links in docs [`#5047`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5047)\r\n- fix: hodometer enable flag and docs [`#5025`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5025)\r\n- fix helm docs runtime name typo [`#5019`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5019)\r\n- feat(helm): Make SeldonConfig configurable in SeldonRuntime Helm chart [`#5014`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fpull\u002F5014)\r\n- Generating changelog for v2.7.0-rc1 [`cf6fc83`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fcommit\u002Fcf6fc83561d6f57a42ee0cb0e6eef7dbbdf6c76e)\r\n- Setting version for yaml manifests [`4bd4b18`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fcommit\u002F4bd4b187adc256d125200bf73cfdbcf6a3c58057)\r\n- Setting version for helm charts [`30bcd42`](https:\u002F\u002Fgithub.com\u002FSeldonIO\u002Fseldon-core\u002Fcommit\u002F30bcd42fba0531792aec268aba7526dcb9aa716e)\r\n","2023-08-25T09:18:01"]