Rails 8.1 + SolidQueueで自ホストエラーレポストグラムの設定手順：完全ガイド

如何にrails_error_dashboardを使用してSolidQueueにより、コンテンソースとバックグランドジョブからのエラーを検出するのにキャプチャート、選択的なデータベース隔離。Rails 8.1はSolidQueueを使用するデフォルトのバックグランドジョブバックエンドとして含まれています。それでは、バックグラウンドジョブが失敗したら何が起こりますか？したくはありません。SolidQueueは失敗した実行を追跡しますが、ダッシュボードやシブルの自動スコアリング、要約プロット、メーリングリストなどを含むエラーのキャプチャも必要ですか？これは現時点ではオールアンダーではない。rails_error_dashboardを使用してこれらの機能を解決しました。完全にあなたのRailsプロセス内で動作します。外部ツールや月額賃料なし、サーバーを超えるデータなしです。この投稿には、Rails 8.1アプリケーションでsolidqueueを用いたrails_error_dashboardのインストール、オプションディスタブレークからデータベース隔離のための設定について説明します。このガイドで最後にあなたは自動的なキャプチャが：コンテナとSolidQueueのおかげで両方のエラーダッシュボードにログとして蓄積する/エラードッバックダッシュボード、アナリティクス、セーフガードスコアリングとワークフロー管理に至ります。オプション：アプリデータベースから一貫したエラーを保存その他のデータから。固有のジョブの間でSOLID_QUEUEの再転送メカニズムがまだ動作していることを確認しました（エラーはキャプチャされ、消えるものではありません）。 Ruby 3.2+ (テストにRuby 4.0.1) Rails 7.0+ （このガイドはRails 8.1.2を使用しています）SolidQueue（Railsの8.1がデフォルトで提供します） # gemfile gem

How to set up rails_error_dashboard with SolidQueue — capturing errors from controllers AND background jobs, with optional database isolation. Rails 8.1 ships with SolidQueue as the default background job backend. No more Redis. No more Sidekiq licensing questions. Just your database. But here's a question nobody talks about: what happens when your background jobs fail? Sure, SolidQueue tracks failed executions. But do you get a dashboard? Severity classification? Trend charts? Notifications? Priority scoring? Not out of the box. I built rails_error_dashboard to solve exactly this — a self-hosted, open-source error tracking gem that works entirely inside your Rails process. No external services, no monthly fees, no data leaving your servers. In this post, I'll walk you through setting it up with a Rails 8.1 app using SolidQueue, including the optional separate database configuration for production-grade isolation. By the end of this guide, you'll have: Automatic error capture from both controllers and SolidQueue jobs A full error dashboard at /error_dashboard with analytics, severity scoring, and workflow management Optional: errors stored in a separate database from your app data Verified that SolidQueue's retry mechanism still works (errors are captured, not swallowed) Ruby 3.2+ (tested with Ruby 4.0.1) Rails 7.0+ (this guide uses Rails 8.1.2) SolidQueue (ships with Rails 8.1 by default) # Gemfile gem "rails_error_dashboard" bundle install rails generate rails_error_dashboard:install rails db:migrate That's it. Seriously. The installer: Creates the initializer at config/initializers/rails_error_dashboard.rb Copies 19 migrations for error logs, occurrences, baselines, comments, and more Mounts the engine at /error_dashboard in your routes Rails 8.1 only configures SolidQueue as the queue adapter in production by default. For development testing, add it explicitly: # config/environments/development.rb config.active_job.queue_adapter = :solid_queue config.solid_queue.connects_to = { database: { writing: :queue } } You also need the multi-database setup for the queue database in development: # config/database.yml development: primary: <<: *default database: storage/development.sqlite3 queue: <<: *default database: storage/development_queue.sqlite3 migrations_paths: db/queue_migrate Then load the queue schema: rails db:schema:load:queue The simplest way to run both Puma and SolidQueue together is the Puma plugin: SOLID_QUEUE_IN_PUMA=1 rails server This runs the SolidQueue supervisor (dispatcher + worker) inside the Puma process — perfect for development and single-server deployments. Alternatively, run them separately: # Terminal 1 rails server # Terminal 2 bin/jobs Visit http://localhost:3000/error_dashboard and log in with the default credentials (gandalf / youshallnotpass). Change these in your initializer before deploying to production. To test error capture, create a simple failing job: # app/jobs/test_error_job.rb class TestErrorJob < ApplicationJob queue_as :default def perform raise RuntimeError, "Test error from SolidQueue" end end Enqueue it: rails runner "TestErrorJob.perform_later" Within seconds, the error should appear in your dashboard with: Error type and message Full backtrace Severity classification (auto-detected) Occurrence count Priority score The gem uses a dual-layer approach to ensure no error goes uncaptured: Layer 1: Rails Error Subscriber Rails 7+ introduced Rails.error.subscribe — a built-in error reporting API. The gem subscribes to this, which means it automatically captures errors from: Controllers (via rescue_from and error handling) ActiveJob (SolidQueue, Sidekiq, etc.) Anywhere you use Rails.error.handle or Rails.error.record # This happens automatically — you don't need to do anything Rails.error.subscribe(RailsErrorDashboard::ErrorReporter.new) Layer 2: Rack Middleware A lightweight Rack middleware sits at the top of the middleware stack as a final safety net. If an exception escapes all the way up, it captures it and then re-raises it so Rails can still render the appropriate error page. # Simplified version of what the middleware does def call(env) @app.call(env) rescue => error Rails.error.report(error, handled: false, source: "middleware") raise # Always re-raise! end This is the most important design decision in the entire gem. When we capture an error from a background job, we never swallow the exception. We report it and re-raise it. Why? Because of Sentry issue #1173. Sentry's Sidekiq middleware once swallowed exceptions, which meant Sidekiq thought failed jobs had succeeded — retries never happened. Data was silently lost. We verified this explicitly: after our gem captures a SolidQueue job error, the error still appears in solid_queue_failed_executions. SolidQueue's retry mechanism is fully intact. For production apps, you might want error data in a separate database. Benefits: Performance isolation — high-frequency error writes don't contend with your app's queries Independent scaling — put your error DB on a different server Different retention — aggressive cleanup of old errors without touching app data Security — separate access controls and backup schedules # config/database.yml development: primary: <<: *default database: storage/development.sqlite3 queue: <<: *default database: storage/development_queue.sqlite3 migrations_paths: db/queue_migrate error_dashboard: <<: *default database: storage/development_error_dashboard.sqlite3 migrations_paths: db/error_dashboard_migrate production: primary: <<: *default database: storage/production.sqlite3 queue: <<: *default database: storage/production_queue.sqlite3 migrations_paths: db/queue_migrate error_dashboard: <<: *default database: storage/production_error_dashboard.sqlite3 migrations_paths: db/error_dashboard_migrate For PostgreSQL in production (recommended): production: primary: adapter: postgresql database: myapp_production # ... your existing config queue: adapter: postgresql database: myapp_queue_production # ... error_dashboard: adapter: postgresql database: myapp_errors_production pool: <%= ENV.fetch("RAILS_MAX_THREADS", 5) %> username: <%= ENV['ERROR_DB_USER'] %> password: <%= ENV['ERROR_DB_PASSWORD'] %> host: <%= ENV['ERROR_DB_HOST'] %> migrations_paths: db/error_dashboard_migrate rails generate rails_error_dashboard:install \ --separate-database \ --database error_dashboard Or if you've already installed, update the initializer: # config/initializers/rails_error_dashboard.rb RailsErrorDashboard.configure do |config| config.use_separate_database = true config.database = :error_dashboard # Must match your database.yml key # ... end The installer places migrations in db/migrate/ by default. For a separate database, move them to match your migrations_paths: mkdir -p db/error_dashboard_migrate mv db/migrate/*rails_error_dashboard* db/error_dashboard_migrate/ rails db:create # Creates all databases rails db:migrate # Runs migrations against the correct databases You can verify the databases are truly isolated: # Check error_dashboard DB has the gem's tables sqlite3 storage/development_error_dashboard.sqlite3 ".tables" # => rails_error_dashboard_error_logs, rails_error_dashboard_applications, ... # Check primary DB has NO error tables sqlite3 storage/development.sqlite3 ".tables" # => ar_internal_metadata, schema_migrations (clean!) # Check queue DB has only SolidQueue tables sqlite3 storage/development_queue.sqlite3 ".tables" # => solid_queue_jobs, solid_queue_failed_executions, ... Three databases, three concerns, zero cross-contamination. Once installed, the dashboard at /error_dashboard gives you: Error list with search, filtering by type/severity/status/timeframe, and batch operations Error details with full backtrace, occurrence timeline, and context Analytics with 7-day trend charts, severity breakdown, and spike detection Workflow — assign errors, set priority, add comments, snooze, resolve Multi-channel notifications — Slack, Email, Discord, PagerDuty, webhooks (all optional) Platform detection — automatically tags errors as iOS/Android/Web/API Dark mode — because we're not savages Before deploying: Change the default credentials in your initializer Set retention_days to prevent unbounded growth (e.g., config.retention_days = 90) Consider async logging for high-throughput apps: config.async_logging = true config.async_adapter = :solid_queue # Use your existing SolidQueue! Configure notifications — at minimum, set up Slack or email for critical errors Consider error sampling if you're generating thousands of errors per day: config.sampling_rate = 0.1 # Log 10% of non-critical errors Rails 8.1 only configures the queue database in production. If you see Could not find table 'solid_queue_processes' in development, you need to: Add the queue database to your development config in database.yml Add config.solid_queue.connects_to = { database: { writing: :queue } } to your development environment Run rails db:schema:load:queue Migrations Running Against the Wrong Database When using separate databases, the gem's migrations must be in the directory matching your migrations_paths. If you see error tables in your primary database, move the migration files: mv db/migrate/*rails_error_dashboard* db/error_dashboard_migrate/ If you're testing the enqueue endpoint with curl and getting 422 errors, that's Rails' CSRF protection working correctly. Use rails runner or the Rails console instead: rails runner "TestErrorJob.perform_later" The gem is open source and free forever: GitHub: AnjanJ/rails_error_dashboard RubyGems: gem install rails_error_dashboard Live Demo: rails-error-dashboard.anjan.dev (gandalf / youshallnotpass) If you're running Rails 8.1 with SolidQueue, this is the easiest way to get production-grade error tracking without paying for a SaaS product. Five minutes of setup, zero monthly fees, and your data never leaves your servers. Have questions or feedback? Open an issue on GitHub or find me on Twitter.