Operations

TL;DR The .github/copilot-instructions.md file started as 10 lines of project description and grew into a 99-line “operating system” for AI assistants. Then it split: failure patterns moved into docs/ai-lessons.md (now 482 lines across 20+ categories), and file-type-specific rules moved into .github/instructions/ with applyTo glob patterns. The same template structure was standardized across 5 repositories. This post traces the three generations of AI instruction architecture and shows how every production incident permanently improves AI reliability. ...

TL;DR Grafana went completely dark for about 26 hours on my home k3s cluster. Two things broke simultaneously: Loki entered CrashLoopBackOff, and Prometheus silently stopped ingesting metrics — its pods showed as healthy and 2/2 Running the whole time. The actual cause was Longhorn’s auto-balancer migrating replicas onto a freshly-added cluster node (k3s-agent-4) that had unstable storage during its first 48 hours. The replica I/O errors propagated directly into the workloads, corrupting mid-write files: a Prometheus WAL segment and a Loki TSDB index file. Both required offline surgery via a busybox pod to delete the corrupted files before the services could recover. ...

TL;DR I run Wiki.js on k3s as the cluster’s internal knowledge base. It is not a place I write documentation — it is a place the AI writes documentation after completing work. When Claude finishes deploying a service, debugging an incident, or refactoring infrastructure, it commits the results to the wiki with architecture diagrams, decision rationale, and operational notes. I am the primary reader. When I want to understand how something works, or why a specific decision was made three weeks ago, I go to the wiki instead of digging through git history or re-reading code. ...

TL;DR The AI memory system I built three weeks ago started causing the problem it was designed to solve: context window exhaustion. Five generic Claude skills — duplicated identically across all 5 repositories in my workspace — consumed 401KB (~100K tokens) of potential context. The gh-cli skill alone was 40KB per copy, accounting for 42% of all skill content. I ran a full audit, deleted 25 duplicate files, and documented the anti-pattern to prevent recurrence. ...

TL;DR Upgraded a production k3s cluster from v1.29.0+k3s1 to v1.34.4+k3s1 across 8 nodes — 3 control plane servers, 4 amd64 worker agents, and 1 arm64 Lima VM agent. The upgrade stepped through every minor version (v1.29 → v1.30 → v1.31 → v1.32 → v1.33 → v1.34) with etcd snapshots between each step. Longhorn was upgraded from v1.6.0 to v1.8.2 in two stages (v1.7.3 as an intermediate step). SSH was broken to all cluster nodes, so the entire upgrade was done via Proxmox QEMU Guest Agent (qm guest exec) and Lima CLI (limactl shell). Discovered that k3s intentionally pins Traefik to v2.11.24 even when bundling Helm chart v27 — Traefik v3 migration is a separate effort. ...

TL;DR Two weeks of building a production Kubernetes cluster with AI pair programming. Claude Opus 4.6 handles complex multi-step infrastructure work via the CLI. GitHub Copilot provides inline code completion in VS Code. AWS Bedrock (Nova Micro, Claude Sonnet 4.5) powers runtime AI services inside the cluster. The key discovery: AI tools without persistent memory are dangerous. Every session starts from zero. The same bugs get recreated, the same anti-patterns get suggested, the same cluster-specific constraints get forgotten. The solution is the “Memory Protocol” – a set of documentation files the AI reads before every session and updates after every discovery. ...

TL;DR After running the cluster for nearly two weeks, today I took a step back to document and optimize the monitoring stack. This covers kube-prometheus-stack (Prometheus + Grafana + AlertManager), Loki for log aggregation, custom dashboards for every service, alert tuning to reduce noise, and the cluster-wide performance benchmarks I ran to establish baseline metrics. The Monitoring Architecture ┌──────────────────────────────────────────────────┐ │ Grafana │ │ ┌──────────┐ ┌──────────┐ ┌──────────┐ │ │ │ Metrics │ │ Logs │ │ Alerts │ │ │ │ Explorer │ │ Explorer │ │ Rules │ │ │ └──────┬───┘ └──────┬───┘ └──────┬───┘ │ └─────────┼──────────────┼─────────────┼───────────┘ │ │ │ ┌─────┴─────┐ ┌─────┴─────┐ │ │Prometheus │ │ Loki │ │ │ (metrics) │ │ (logs) │ │ └─────┬─────┘ └─────┬─────┘ │ │ │ ┌─────┴──────┐ ┌──────┴──────┐ ┌─────┴────┐ │AlertManager│ │ Exporters │ │Promtail │ │ → Slack │ │ node │ │(log │ └────────────┘ │ kube-state │ │ shipper) │ │ cAdvisor │ └──────────┘ │ custom │ └─────────────┘ kube-prometheus-stack The foundation is kube-prometheus-stack, deployed via Helm. This single chart installs: ...

TL;DR After one week of operating this cluster with real workloads, I have accumulated a healthy list of production failures. Each one taught me something about Kubernetes, infrastructure, or my own assumptions. Here are the top 10, ranked by how much time they cost to investigate and fix. 1. The Longhorn S3 Backup Credential Rotation Impact: All Longhorn backups silently failed for 12 hours. What happened: I rotated the IAM credentials used for S3 backups and updated the Kubernetes secret. But Longhorn caches credentials at startup — it does not re-read the secret dynamically. All backup jobs continued using the stale credentials and failing silently. ...