Field notes from the terminal.
Long-form, slowly written essays on Linux, networking, DevOps, and the boring infrastructure that keeps the rest of it standing. No guides written by SEO. No newsletter signup.
Linux tutorials and DevOps guides — Linux administration, server security, bash scripting, Docker, SSH security, cron jobs, file permissions.Essential Linux Tools List: 50+ Must-Have Commands for Every Admin
A comprehensive list of essential Linux tools and commands organized by category — networking, file management, monitoring, security, and more.
written by Khimananda Oli · sysadmin, recovering kubernaut
Showing all 47 articles
- devops
CI/CD Pipeline Design: The Stages Every Deployment Needs
Design a CI/CD pipeline that fails fast and deploys safely: lint, build once, test, scan, publish by digest, staged rollout, smoke tests, and a rollback that works.
11 minKhimananda OliKhimananda Oli11 min · est. read - security
SSH Key Types Compared: Ed25519 vs RSA vs ECDSA in 2026
Ed25519, RSA, ECDSA and FIDO2 SSH keys compared on security, speed and compatibility, plus SHA-1 deprecation, FIPS limits, post-quantum SSH and safe key rotation.
11 minKhimananda OliKhimananda Oli11 min · est. read - security
Applying CIS Benchmarks on Linux Without Breaking Production
A safe rollout plan for CIS Benchmarks on RHEL and Ubuntu: scan with OpenSCAP or USG, know which controls break servers, tailor exceptions, and remediate in stages.
12 minKhimananda OliKhimananda Oli12 min · est. read - ai
Migrating SysV Init Scripts to systemd Units with AI
Use an LLM to convert legacy /etc/init.d scripts into native systemd units: give it the right context, map every init idiom to a directive, and verify before cutover.
11 minKhimananda OliKhimananda Oli11 min · est. read - ai
Automating Postmortems: Feeding journald Logs to an LLM
Turn journald logs from a finished incident into a blameless postmortem draft: extract the window, merge hosts, build a cited timeline, and catch causal leaps.
11 minKhimananda OliKhimananda Oli11 min · est. read - ai
AI-Assisted SELinux Troubleshooting: Decoding AVC Denials
Use an LLM to read SELinux AVC denials without letting it reach for setenforce 0: find hidden denials, fix labels and booleans first, and keep audit2allow last.
11 minKhimananda OliKhimananda Oli11 min · est. read - ai
Natural Language to jq: Querying JSON on the Linux Command Line
Get correct jq filters from AI by sending the shape of your JSON, not the data — real idioms, the mistakes models make, and how to verify on sample data first.
10 minKhimananda OliKhimananda Oli10 min · est. read - ai
AI Code Review for Shell Scripts: Quoting Bugs and Unsafe Patterns
Run ShellCheck first, then use AI to review legacy bash for what it misses — temp-file races, set -e gaps, rm -rf traps — with CI hooks and bats tests for the fixes.
11 minKhimananda OliKhimananda Oli11 min · est. read - ai
Turning Bash History into Documentation with an LLM
Make bash history worth keeping, pull out one session by time window, scrub the secrets, and have an LLM turn the commands that worked into how-tos and onboarding notes.
11 minKhimananda OliKhimananda Oli11 min · est. read - ai
Explaining Kernel Panics and OOM Killer Logs with AI
Read OOM-killer reports and kernel panics with an LLM: collect evidence, decode gfp_mask and the task table, tell cgroup OOMs from global ones, and verify the cause.
11 minKhimananda OliKhimananda Oli11 min · est. read - ai
Auditing iptables and nftables Rules with AI (and Verifying It)
Dump your ruleset, have AI flag shadowed, redundant and overly broad rules, then prove each finding with counters and nmap, including the IPv6 and Docker gaps.
12 minKhimananda OliKhimananda Oli12 min · est. read - ai
Using AI CLIs Like Claude Code for Linux Server Maintenance
Agentic CLIs can investigate and fix servers, which is why they need guardrails: read-only first, a dedicated user, tight allowlists, dry-run plans and audit logs.
12 minKhimananda OliKhimananda Oli12 min · est. read - ai
Generating Linux Runbooks with AI (Without Shipping Fiction)
Turn shell history, scripts and incident notes into runbooks a tired engineer can follow at 3 AM — the structure that matters, verification gates, and what AI must not write.
11 minKhimananda OliKhimananda Oli11 min · est. read - ai
Writing find, awk and sed One-Liners with AI (and Verifying Them)
Get correct one-liners from a model on the first or second try — how to describe the task, the GNU vs BSD trap, and a verification loop that catches wrong answers safely.
12 minKhimananda OliKhimananda Oli12 min · est. read - ai
Using AI for Cron Expressions — and the Trap It Gets Wrong
Generate and explain cron schedules with an LLM, then verify them. Covers the day-of-month/day-of-week OR behaviour models routinely miss, plus systemd OnCalendar conversion.
10 minKhimananda OliKhimananda Oli10 min · est. read - ai
Using AI to Summarize Massive Log Files on Linux
Logs are far larger than any context window. Reduce first with grep, awk and uniq, then let a local model explain the aggregate — with working pipelines and cost-free privacy.
12 minKhimananda OliKhimananda Oli12 min · est. read - ai
AI-Assisted Incident Response: Triaging a Compromised Linux Server
Use a local model to accelerate triage on a suspected compromise without destroying evidence — what to capture first, what AI reads well, and the decisions it must never make.
13 minKhimananda OliKhimananda Oli13 min · est. read - ai
Reviewing AI-Generated Shell Scripts Before You Run Them
A concrete review process for AI-written bash — the seven mistakes models repeat, ShellCheck and dry-run gates, container testing, and a checklist to run before production.
12 minKhimananda OliKhimananda Oli12 min · est. read - ai
Debugging systemd Unit Failures with an LLM Assistant
Collect the right systemd context in one command, hand it to a local model, and verify what it tells you — plus the directives models routinely invent and how to catch them.
11 minKhimananda OliKhimananda Oli11 min · est. read - ai
Using AI to Explain a Linux Command Before You Run It
Build a local explain function that decodes any shell command before execution — prompt design, dangerous patterns to catch, and how to verify the answer against real documentation.
11 minKhimananda OliKhimananda Oli11 min · est. read - ai
Air-Gapped AI: Running Local Models on Linux with No Internet
Move models, Python wheels and container images onto an offline Linux host — Ollama blob transfer, GGUF plus Modelfile, pip wheelhouses, checksum verification and update process.
12 minKhimananda OliKhimananda Oli12 min · est. read - ai
Benchmarking LLM Inference on Linux: Measuring Tokens per Second Properly
Measure local LLM performance you can trust — TTFT vs tokens/sec, llama-bench, throughput under concurrency, and the confounders that make most published numbers useless.
11 minKhimananda OliKhimananda Oli11 min · est. read - ai
Fine-Tuning an LLM on a Single Linux GPU with LoRA and QLoRA
When fine-tuning is the right tool, how LoRA and QLoRA fit training onto one consumer GPU, dataset format, the hyperparameters that matter, and exporting to Ollama.
13 minKhimananda OliKhimananda Oli13 min · est. read - ai
Kubernetes GPU Scheduling for AI Workloads
Schedule GPUs in Kubernetes — device plugin vs GPU Operator, requesting nvidia.com/gpu, taints and node selectors, time-slicing and MIG for sharing, and why pods stay Pending.
12 minKhimananda OliKhimananda Oli12 min · est. read - ai
Self-Hosted ChatGPT Alternatives on Linux: Open WebUI and text-generation-webui
Set up a private chat interface for local models on Linux — Open WebUI with Ollama, or text-generation-webui with multiple loaders — plus systemd units and safe remote access.
12 minKhimananda OliKhimananda Oli12 min · est. read - ai
Local RAG on Linux: Search Your Own Documents with a Local LLM
Build a fully offline retrieval-augmented generation setup on Linux — embeddings with Ollama, a vector store, sensible chunking, and how to tell when retrieval is failing.
13 minKhimananda OliKhimananda Oli13 min · est. read - ai
GPU-Accelerated Docker Containers on Linux with NVIDIA Container Toolkit
Give Docker and Podman containers GPU access on Linux — install the NVIDIA Container Toolkit, pick the right CUDA image, use Compose GPU reservations, and fix driver errors.
12 minKhimananda OliKhimananda Oli12 min · est. read - ai
Monitoring GPU Usage on Linux: nvidia-smi, nvtop and rocm-smi
Read GPU utilization, VRAM, power and throttling correctly on Linux — nvidia-smi query mode, nvtop, rocm-smi, intel_gpu_top, CSV logging and Prometheus metrics.
11 minKhimananda OliKhimananda Oli11 min · est. read - ai
Running JupyterLab as a systemd Service on a Linux Server
Install JupyterLab in a venv, run it as a hardened systemd service, set a password, put nginx and TLS in front, cap resources, and fix the usual proxy and kernel errors.
12 minKhimananda OliKhimananda Oli12 min · est. read - ai
Intel GPUs and NPUs on Linux: OpenVINO for Local Inference
Get Intel Arc, integrated Xe graphics and Core Ultra NPUs running AI workloads on Linux — driver stack, OpenVINO setup, device selection, and troubleshooting.
11 minKhimananda OliKhimananda Oli11 min · est. read - ai
Running AI Models on a Raspberry Pi: What's Actually Possible
An honest look at local AI on Raspberry Pi — which models run usefully, how to set up Ollama and whisper.cpp on ARM, when to add an accelerator, and thermal and memory limits.
11 minKhimananda OliKhimananda Oli11 min · est. read - ai
LLM Quantization Explained: GGUF, AWQ, GPTQ and Choosing a Format
How quantization shrinks LLMs to fit consumer GPUs — the memory math, what Q4_K_M actually means, how GGUF, GPTQ, AWQ and bitsandbytes differ, and which to pick.
11 minKhimananda OliKhimananda Oli11 min · est. read - ai
vLLM vs llama.cpp vs TGI: Choosing an LLM Server for Linux
How vLLM, llama.cpp, TGI and Ollama differ on hardware requirements, concurrency, quantization support and operations — with a decision guide and setup commands for each.
12 minKhimananda OliKhimananda Oli12 min · est. read - ai
Whisper on Linux: Local Speech-to-Text from the Command Line
Run OpenAI's Whisper offline on Linux with whisper.cpp or faster-whisper — build it, pick a model size, convert audio correctly, batch transcribe, and generate subtitles.
11 minKhimananda OliKhimananda Oli11 min · est. read - ai
Running Stable Diffusion Locally on Linux with ComfyUI or A1111
Install ComfyUI and AUTOMATIC1111 on Linux, size VRAM correctly, manage models safely, run them as systemd services, and fix the usual CUDA and VRAM errors.
12 minKhimananda OliKhimananda Oli12 min · est. read - ai
AMD ROCm on Linux: Getting Radeon GPUs Working for AI
Install ROCm on Linux, fix device permissions, run PyTorch and llama.cpp on Radeon, use HSA_OVERRIDE_GFX_VERSION for unsupported cards, and debug the common failures.
12 minKhimananda OliKhimananda Oli12 min · est. read - ai
Install NVIDIA Drivers and CUDA on Linux Without Breaking Your System
Driver vs toolkit vs container runtime, distro-native installation, Secure Boot and DKMS, GPU containers, and fixes for every common NVIDIA error on Linux.
12 minKhimananda OliKhimananda Oli12 min · est. read - ai
Best Linux Distro for AI and Machine Learning Workstations
Ubuntu, Fedora, Arch, Debian or RHEL for AI work? A practical comparison focused on GPU driver packaging, kernel cadence, vendor support, and what actually breaks.
11 minKhimananda OliKhimananda Oli11 min · est. read - ai
Building llama.cpp on Linux: CUDA, ROCm, Vulkan, and CPU Backends
Compile llama.cpp from source on Linux with the right backend — CUDA, ROCm/HIP, Vulkan, or optimized CPU — then run, serve, quantize, and benchmark GGUF models.
13 minKhimananda OliKhimananda Oli13 min · est. read - ai
How to Run Local LLMs on Linux with Ollama: Complete Setup Guide
Install Ollama on Linux, run your first local LLM, enable GPU acceleration, tune the systemd service, and use the API — a practical guide for admins and developers.
12 minKhimananda OliKhimananda Oli12 min · est. read - linux
Linux Tools Cheat Sheet: The Only Reference You Need
A concise, printable Linux command cheat sheet covering file ops, networking, permissions, processes, and more. Print it or bookmark it.
8 minKhimananda OliKhimananda Oli8 min · est. read - linux
Linux File Permissions Explained: A Complete Guide
Understand how Linux file permissions work — from the basics of read, write, and execute to advanced topics like setuid, setgid, sticky bits, and ACLs.
10 minKhimananda OliKhimananda Oli10 min · est. read - linux
Cron Jobs: The Complete Guide to Scheduling Tasks in Linux
Learn how to schedule automated tasks with cron — syntax, examples, common pitfalls, and best practices for production crontabs.
9 minKhimananda OliKhimananda Oli9 min · est. read - docker
Docker for Linux Admins: From Basics to Production
A practical guide to Docker for Linux administrators — containers, images, Dockerfiles, networking, volumes, and production deployment tips.
12 minKhimananda OliKhimananda Oli12 min · est. read - security
SSH Security Best Practices for Production Servers
Harden your SSH configuration with key-based authentication, port changes, fail2ban, jump hosts, and modern security techniques.
10 minKhimananda OliKhimananda Oli10 min · est. read - security
Linux Server Hardening Checklist: 20 Essential Steps
A practical checklist for securing Linux servers — from initial setup to firewall rules, kernel tuning, audit logging, and intrusion detection.
11 minKhimananda OliKhimananda Oli11 min · est. read