Argo CD continuously comparing desired state in Git with live Kubernetes state

Argo CD in Production: GitOps Sync, ApplicationSets, Rollbacks, and Security

A green sync does not prove the delivery system is safe When Argo CD turns an application green, it proves one narrow thing: at that moment, the cluster matches the desired manifests Argo CD calculated. It does not prove that the image passed its tests, that a deletion is safe, or that the next Git change belongs in production. That boundary is the right place to begin. Argo CD is not a pipeline that makes release judgments for a team. It is a Kubernetes controller that repeatedly compares, reports, and—when policy permits—reconciles state. Its value is not another attractive dashboard. Its value is turning deployment intent from scattered commands into something reviewable and reproducible. ...

May 9, 2025 · 13 min · 2705 words · Xinwei Xiong, Me
Several model pipelines converge at an AI Gateway where cost and governance are weighed

AI Gateway Guide: LiteLLM, Kong, APISIX, Cloudflare, or Portkey?

An AI Gateway is not merely a reverse proxy placed in front of a language model. Once an application reaches production, every model call carries several kinds of uncertainty at once: long-lived streaming connections, token-based billing, provider quotas, sensitive inputs, changing model behavior, and outputs that cannot be trusted by default. A conventional API gateway can handle part of this traffic. It may authenticate clients, terminate TLS, enforce request limits, and route HTTP requests. It does not automatically answer the questions that matter most to an AI product: ...

April 16, 2025 · 16 min · 3223 words · Xinwei Xiong, Me
A layered search system connecting multimodal data, embeddings, reranking, and grounded answers

Jina AI in 2026: Embeddings, Reranking, Reader, and Jina Serve

Jina used to be easiest to explain as a cloud-native neural-search framework. That description is still true, but it is no longer sufficient. In 2026, the name covers two related products: Jina Search Foundation: hosted APIs and model families for embeddings, reranking, web reading, search, and research. Jina Serve: the open-source framework for turning Python AI components into services and composing them into distributed flows. The distinction matters. Search Foundation gives an application retrieval intelligence; Jina Serve gives a team control over how its own services run. One is a set of capabilities, the other an orchestration layer. Treating them as interchangeable usually leads either to unnecessary infrastructure or to a hosted dependency that was never consciously chosen. ...

April 12, 2025 · 12 min · 2381 words · Xinwei Xiong, Me

Kubernetes Resources and Learning Path Summary

Requirements 🔥 We need to further study and advance in kubernetes, reading source code is a necessary path. 👀 More importantly, it’s the collection of kubernetes resources. ⚠️ All resources use external links, book resources may not have links, others are personal experiences and summaries. Please contact for removal if there’s copyright infringement. 🚧 ⚠️ **Note: This article will be permanently stored in notion and will be continuously updated, providing a write channel. If you have better resources, welcome to add them on notion ~** CloudNative Learning Path ...

August 6, 2024 · 6 min · 1208 words · Xinwei Xiong, Me
Three interchangeable runtime, network, and storage modules connect to one orchestration chassis through standardized couplings

Deep Dive into Kubernetes CNI, CRI, CSI Components

Container Runtime Container Runtime runs on every node in a Kubernetes (k8s) cluster and is responsible for the entire container lifecycle. Docker is currently the most widely used. With the development of container cloud, more and more container runtimes have emerged. To solve the integration problems between these container runtimes and Kubernetes, the community introduced CRI (Container Runtime Interface) in Kubernetes version 1.5 to support more container runtimes. ⌀ Image no longer available · image-20230308203547983 ...

September 28, 2023 · 32 min · 6766 words · Xinwei Xiong, Me

OpenIM clustering design Kubernetes deploy concludes

Conference and Reference Links Conference reference documents: https://nsddd.notion.site/2899028707604b8090b36677c031cdf8?pvs=4 Video playback: Bilibili: https://www.bilibili.com/video/BV1s8411q7Um/?spm_id_from=333.999.0.0 Comment: I think that middleware can be replaced with https://kubeblocks.io which can help you manage multiple database middlewares. im reads the configuration information, and reads the config/ directory. The config.yaml hard-coded in the code can be automatically divided into rpc for different services, and then unified directories. By default, the directory is read from the binary running path. two floors openim version: https://github.com/openimsdk/open-im-server/blob/main/docs/conversions/version.md Storage can consider using: https://github.com/openebs/openebs https://github.com/rook/rook Core target: ...

September 17, 2023 · 14 min · 2771 words · Xinwei Xiong, Me