400 Recipes Milestone: What We Built and What's Next

💡 Quick Answer: Kubernetes Recipes has reached 400 production-ready articles covering Kubernetes, OpenShift, NVIDIA GPU infrastructure, AI model serving, and platform engineering. Every recipe includes working YAML, troubleshooting steps, and best practices drawn from real production environments.

The Problem

Kubernetes documentation is scattered across vendor docs, GitHub issues, blog posts, and Stack Overflow answers. Finding a single, tested, production-ready solution means stitching together fragments from dozens of sources. This is especially true for advanced topics like GPU infrastructure, RDMA networking, and AI model serving.

The Solution

We’ve built a single, searchable library of 400 recipes. Here’s what’s new.

AI & GPU Infrastructure (50+ recipes)

The fastest-growing section. From deploying a single model to managing multi-tenant GPU clusters:

• Model Deployment — Deploy Llama 2 70B, Phi-4, Whisper, Stable Diffusion XL, and 10+ more models with production-ready manifests
• Inference Serving — Triton + vLLM, TensorRT-LLM, multi-model serving, autoscaling on GPU metrics
• Model Storage — hostPath vs PVC patterns for fast model loading with NVMe caching
• Benchmarking — AIPerf and GenAI-Perf for TTFT, ITL, and throughput measurement
• Training — NeMo training, distributed training with Kubeflow, MPI Operator
• Multi-Tenant GPU — namespace isolation, ResourceQuotas, time-slicing vs MIG, chargeback monitoring

NVIDIA Networking (20+ recipes)

Deep-dive infrastructure recipes for high-performance networking:

• NNCP — 10 recipes for static IPs, bonds, VLANs, Linux bridges, OVS bridges, and jumbo frames
• SR-IOV — VF configuration, AI workload binding, mixed NIC generations, troubleshooting
• NFSoRDMA — dedicated NIC setup, bonding, PV integration, performance tuning
• DOCA/MOFED — driver containers on OpenShift, NIC driver entrypoint deep-dive

ArgoCD GitOps (10 recipes)

Complete ArgoCD coverage from basic setup to multi-cluster fleet management:

• Sync waves ordering, App-of-Apps pattern, pre/post-sync hooks, database migrations, canary deployments, and multi-cluster management

OpenShift Platform Engineering (15+ recipes)

• Image Mirroring — IDMS, ITMS, MCP rollout management
• Security — RHACS full install, custom policies, network segmentation, compliance scanning, CI/CD integration
• Node Management — RHCOS MachineConfig, custom CA certificates
• Registry — Quay robot accounts, default permissions, token rotation
• Serverless — KnativeServing with KPA/HPA

Core Kubernetes (100+ recipes)

The foundation — battle-tested patterns for every cluster:

• Troubleshooting — CrashLoopBackOff, OOMKilled, ImagePullBackOff, DNS issues, scheduling failures
• Autoscaling — HPA, VPA, Cluster Autoscaler, KEDA
• Deployments — rolling updates, canary, blue-green, probes
• Security — NetworkPolicies, Pod Security, RBAC, secrets management

What’s Next

• More AI model recipes — covering the latest HuggingFace trending models as they release
• Performance benchmarking — MLPerf, AIPerf, and GenAI-Perf deep-dives
• Platform engineering patterns — multi-cluster GitOps, policy-as-code, developer portals
• Community contributions — we welcome PRs at github.com/greg4fun/kubernetesrecipes.github.io

About the Book

These online recipes complement the Kubernetes Recipes book published by Apress. The book covers core concepts in depth; this site extends it with cutting-edge topics in AI infrastructure, NVIDIA networking, and OpenShift platform engineering.

Key Takeaways

• 400 recipes covering 10 categories: AI, networking, security, deployments, autoscaling, troubleshooting, configuration, storage, observability, and Helm
• Every recipe includes working YAML, a Quick Answer, common issues, and best practices
• AI/GPU infrastructure is the fastest-growing section with 50+ recipes
• All content is open source and free — contributions welcome