deploy

Deployment examples

This directory contains opinionated, copy-pasteable deployment recipes. They are not loaded by the binary and have no runtime coupling to the rest of the codebase. Pick the one that matches your platform and adapt it.

Subdirectory	Use case
compose/	Local dev + small self-hosted deployments. Adds a scraper cron sidecar to docker-compose.yml.
k8s/	Kubernetes via plain manifests: Deployment for the API, CronJob per vendor for scrapes, Service + probes wired to /readyz / /healthz.
github-actions/	Run the scraper on a schedule against a hosted Postgres (e.g. Supabase, Neon, RDS). Zero infrastructure, free-tier friendly.

Design notes

• The scrape subcommand is one-shot: it exits 0 on success, non-zero on failure. All schedulers here rely on that contract.
• A per-vendor Postgres advisory lock (pg_try_advisory_lock('scrape:<vendor>')) makes overlapping runs safe. Two schedulers firing at once will cause one to skip with a warning, not corrupt data.
• Freshness is recorded in the scrape_runs table. To alert on stale data:

  SELECT vendor, MAX(finished_at) AS last_success
  FROM scrape_runs
  WHERE status = 'success'
  GROUP BY vendor;

• Recommended cadences (tradeoff: freshness vs. API politeness):
  • AWS: weekly, bulk price list churns slowly.
  • Azure: daily, retail prices API updates frequently.
  • GCP: daily, catalog has ~daily churn on new SKUs.