Skip to content

deployment.md

Latest commit

 

History

History
75 lines (51 loc) · 7.17 KB

deployment.md

File metadata and controls

75 lines (51 loc) · 7.17 KB
Raw

Deployment Guide

This guide captures the production considerations for running SourceMonitor inside a host Rails application. Pair it with your platform playbook (Heroku, Render, Kubernetes, etc.) for environment-specific instructions.

Build & Release Pipeline

Ruby version management in production: Use rbenv, asdf, or the Ruby version baked into your container image, depending on your deployment platform. The commands below are shown without version manager prefixes—adjust for your environment (e.g., rbenv exec bundle install, asdf exec bundle install, or bare bundle install in Docker).

  1. Install dependencies – use bundle install and npm install during build steps.
  2. Copy and run migrations – always run bin/rails railties:install:migrations FROM=source_monitor before bin/rails db:migrate so new engine tables ship with each release.
  3. Precompile assets – bin/rails assets:precompile pulls in SourceMonitor's bundled CSS/JS outputs and Stimulus controllers. Fail the build if source_monitor:assets:verify raises.
  4. Run quality gates – bin/rubocop, bin/brakeman --no-pager, bin/lint-assets, and bin/test-coverage mirror the repository CI setup.

Process Model

SourceMonitor assumes the standard Rails 8 process split:

  • Web – your application server (Puma) serving the mounted engine and Action Cable. When using Solid Cable, no separate Redis process is required.
  • Worker – at least one Solid Queue worker (bin/rails solid_queue:start). Scale horizontally to match feed volume and retention pruning needs. The engine uses three queues: source_monitor_fetch (time-sensitive feed polling), source_monitor_scrape (content extraction), and source_monitor_maintenance (health checks, cleanup, favicon, images, OPML import). Use queue selectors if you dedicate workers to specific queues.
  • Scheduler/Recurring – optional process invoking bin/jobs --recurring_schedule_file=config/recurring.yml so the bundled recurring tasks enqueue fetch/scrape/cleanup jobs. Disable with SOLID_QUEUE_SKIP_RECURRING=true when another scheduler handles cron-style jobs.

Database & Storage

  • Primary Database – hosts sources, items, logs, and Solid Cable messages. Ensure autovacuum keeps up; retention pruning helps bound growth.
  • Solid Queue Database (optional) – create a dedicated connection via rails solid_queue:install if queue load warrants isolation. Update the generated config before running migrations.
  • Backups – include Solid Queue tables in backups if they share the primary database; they store job state.

Observability & Alerting

  • Subscribe to ActiveSupport notifications (source_monitor.fetch.finish, source_monitor.scheduler.run, source_monitor.dashboard.*) to emit logs or metrics into your monitoring stack.
  • Scrape SourceMonitor::Metrics.snapshot periodically (e.g., via a health check controller) to track counters and gauges in Prometheus or StatsD.
  • Mission Control integration becomes useful once queues exceed a few hundred jobs; enable it when your platform already hosts the Mission Control UI.

Security & Authentication

  • SourceMonitor is fail-closed by default: without a configured handler every engine route returns 403 Forbidden. Lock down the routes with authentication hooks (config.authentication.authenticate_with / authorize_with). Only set config.authentication.open_access = true for non-production demos where public access is intentional.
  • Configure HTTPS for Action Cable if you expose Solid Cable over the public internet.
  • Store API keys for authenticated feeds in encrypted credentials and inject them via per-source custom headers.

Scaling Guidelines

  • Increase config.fetch_queue_concurrency and the number of Solid Queue workers as source volume grows.
  • Adjust config.fetching multipliers to smooth out noisy feeds; raising failure_increase_factor slows retries for consistently failing sources.
  • Tune config.fetching.scheduler_batch_size (default 25) to control how many sources are picked up per scheduler run. On larger servers, increase this to 50-100.
  • The config.fetching.stale_timeout_minutes (default 5) controls how quickly stuck "fetching" sources are recovered. Lower values mean faster recovery but more aggressive reconciliation.
  • After deploys or queue stalls where many sources become overdue simultaneously, run bin/rails source_monitor:maintenance:stagger_fetch_times to distribute them across a time window and prevent thundering herd.
  • The maintenance queue (concurrency 1 by default) handles non-time-sensitive work. Scale independently of fetch/scrape via config.maintenance_queue_concurrency or SOURCE_MONITOR_MAINTENANCE_CONCURRENCY env var.
  • Use config.retention to cap database growth; nightly cleanup jobs can run on separate workers if pruning becomes heavy.

Rolling Upgrades

  1. Merge and deploy code.
  2. Run migrations before restarting workers so new tables/columns exist before jobs enqueue.
  3. Restart Solid Queue workers after deploy so they load updated configuration and code.
  4. Verify the dashboard loads, queue counts appear, and Mission Control links resolve when enabled.

Disaster Recovery Checklist

  • Restore the database, then replay any external feed data if necessary (SourceMonitor deduplicates using GUID/fingerprint).
  • Resume Solid Queue workers and monitor fetch_failed_total gauge for anomalies.
  • Rebuild assets if you deploy to ephemeral filesystems.

Keep this guide alongside your platform runbooks so teams can confidently deploy and operate SourceMonitor in any environment.

Container Reference Stack

The repository ships a reusable Docker stack under examples/docker that mirrors the recommended process model. It builds a Ruby 4.0 image with Node, mounts your generated example via APP_PATH, and launches three services (web, worker, scheduler) alongside Postgres and Redis. Use it to trial production settings locally or as a baseline for ECS/Kubernetes manifests.

Setup Workflow Rollout Checklist

  1. Adopt the CLI – add bin/source_monitor install --yes (or an equivalent rake task wrapper) to your internal onboarding docs for both greenfield apps and existing hosts. Pair it with bin/source_monitor verify so operators can re-check queue/Action Cable health after deployments.
  2. CI Gate – extend your deploy pipeline to invoke bin/source_monitor verify right after migrations. Fail the build when the command exits non-zero and surface the JSON blob produced by the printer for diagnostics. This step replaces ad-hoc “did you start Solid Queue?” questions before merging.
  3. Telemetry Opt-In – set SOURCE_MONITOR_SETUP_TELEMETRY=true in staging/pre-prod so setup runs append logs to log/source_monitor_setup.log. Ship the file as a build artifact for debugging failed installs.
  4. Rollback Ready – reference docs/setup.md#rollback-steps in your release checklist so anyone piloting the workflow knows how to revert gem/routes/initializer changes if the experiment needs to pause.
  5. Worker Coverage – ensure at least one Solid Queue worker is running (or emits a heartbeat) before moving the PR out of draft; otherwise verification exits with a warning and CI will fail.