One ServiceAccount, Two Namespaces: A Limited Kubernetes Kubeconfig
Create a namespace-limited kubeconfig with one ServiceAccount token, a reusable ClusterRole, and explicit RoleBindings.
Read articleInfrastructure · AI systems · Operations
Field notes.
Small, practical write-ups about making systems safer to change, easier to understand, and calmer to operate.
-
-
Linux Disk Full: Docker overlay2 and Elasticsearch Cleanup Runbook
Find disk consumers, diagnose damaged Docker layers, and reclaim Elasticsearch storage through supported APIs.
Read article -
Kubernetes Namespace and PVC Stuck in Terminating — Cleanup Runbook
A command-first runbook for namespace deletion, PVC protection, Longhorn finalizers, and deliberate storage cleanup.
Read article -
Troubleshooting High Memory Usage and OOMKilled Pods in Kubernetes
A Kubernetes and Node.js memory runbook: cgroups, process RSS, V8 headroom, Next.js retention, heap snapshots, and GC diagnostics.
Read article -
Stock Options in the Iranian Market: A Practical Knowledge Base
An options reference with verified arithmetic, rial/toman checks, conditional payoff scenarios, and exercise-funding calculations.
Read article -
Restoring a Gzip-Compressed MongoDB Backup in Docker
A MongoDB restore runbook covering Docker, gzip archives, authentication, verification, and common restore failures.
Read article -
Archiving and Compressing Directories in Linux with tar, gzip, and zip
A Linux command reference for creating, extracting, listing, and checking tar, gzip, and zip archives.
Read article -
Building a Floating IP with Keepalived for Linux High Availability
A practical two-VM Keepalived setup with VRRP, VIP failover drills, application health checks, and network troubleshooting.
Read article -
Why my active-active HAProxy setup failed: ARP, shared VIPs, and the right way to design it
A real-world look at why one shared loopback VIP caused uneven traffic and stale ARP paths—and how VRRP, ECMP/BGP, or DNS change the design.
Read article -
Making a Kubernetes workload boring to operate
A practical reliability baseline for probes, graceful shutdowns, rollout safety, resource limits, and failure drills.
Read article -
Treat Terraform state like production data
A safer workflow for remote state, locking, reviewed plans, refactors, and recovery without risky manual edits.
Read article -
A repeatable path from alert to cause
A simple incident workflow that connects service-level metrics, deployments, structured logs, and traces without dashboard guesswork.
Read article -
A small RAG service I would actually deploy
A compact reference architecture for ingestion, permission-aware retrieval, cited answers, evaluation, and safe operations.
Read article -
One controlled gateway for LLM traffic
A practical gateway pattern for model routing, budgets, retries, telemetry, and guardrails shared across applications.
Read article