Engineering articles from a DevOps studio: VPS setup for Docker, Prometheus + Grafana monitoring, self-hosted LLMs (Ollama), CI/CD with GitHub Actions, backups and failover. All tested in production.https://opsmith.ru/enhttps://opsmith.ru/en/blog/vps-docker-setup/https://opsmith.ru/en/blog/vps-docker-setup/Step by step to a server you can be proud of: users and SSH keys via cloud-init, Docker with sane logging, firewall and unattended upgrades. All reproducible, as code.Sun, 12 Jul 2026 00:00:00 GMThttps://opsmith.ru/en/blog/prometheus-grafana-monitoring/https://opsmith.ru/en/blog/prometheus-grafana-monitoring/Metrics, a dashboard and Telegram alerts — no SaaS subscriptions, no extra agents. One compose file, three services, and you learn about problems before your customers do.Sun, 28 Jun 2026 00:00:00 GMThttps://opsmith.ru/en/blog/self-hosted-llm-ollama/https://opsmith.ru/en/blog/self-hosted-llm-ollama/Which model fits your server, what it costs, and when an API is actually cheaper. We do the memory math, pick a quant and serve the model over HTTP.Mon, 15 Jun 2026 00:00:00 GMThttps://opsmith.ru/en/blog/github-actions-cicd/https://opsmith.ru/en/blog/github-actions-cicd/Deploy on push to main, secrets, rollbacks. One workflow file you can copy — no Jenkins, no dedicated CI server.Sat, 30 May 2026 00:00:00 GMThttps://opsmith.ru/en/blog/backups-321-restic/https://opsmith.ru/en/blog/backups-321-restic/restic + S3-compatible storage, restore checks on cron, and an alert if a backup goes silent. A backup nobody has tested is a lottery ticket.Mon, 11 May 2026 00:00:00 GMThttps://opsmith.ru/en/blog/failover-without-kubernetes/https://opsmith.ru/en/blog/failover-without-kubernetes/docker compose + healthchecks + a standby server: 99.9% uptime with no cluster. Kubernetes solves problems of a scale most projects never reach.Sun, 19 Apr 2026 00:00:00 GMT