Sentry: Error Tracking and Performance Monitoring
/sentry
Sentry error tracking covers the full debugging workflow, from exception capture to release correlation, in one place. Self-hosting eliminates per-seat pricing and keeps data on your infrastructure, but you take on Docker or Kubernetes maintenance. Worth reaching for when your stack spans multiple services and languages, and you need to correlate errors across them. If zero ops is the priority, the managed cloud service exists; if you're already happy with a simple logger, this adds more surface area than you'll use.
What is Sentry?
Sentry is a developer-first debugging platform for error tracking and performance monitoring. It works by instrumenting your application with a language-specific SDK that captures exceptions, crashes, and transaction timings, then sends them to a central instance. The platform groups similar events, shows full stack traces and user breadcrumbs, and ties errors to specific releases so you can immediately see which deploy introduced a regression. You can run it yourself via Docker or Kubernetes, or use the managed hosted option.
Core features: error tracking, performance, and releases
- โException tracking with full stack traces, local variables, and user breadcrumbs
- โPerformance monitoring (APM) for slow transactions, database queries, and network requests
- โRelease detection to correlate errors with specific deploys or commits
- โBreadcrumb trails recording user actions and application events leading up to a crash
- โSource map support for readable stack traces in minified JavaScript
- โMulti-language SDK coverage across Python, JavaScript, Ruby, Go, Java, Rust, and 15+ more
- โConfigurable alerting via Slack, email, PagerDuty, or custom webhooks
- โSearch and filter events by environment, release, user, or custom tags
What you can monitor with Sentry
- โขMonitoring web application errors across Django, Express, or Rails backends
- โขCrash reporting for mobile apps built with React Native, Flutter, or Swift
- โขBackend microservice observability across polyglot environments
- โขFrontend JavaScript error tracking with source map decoding
- โขCorrelating production errors with specific releases or feature flags
- โขPerformance profiling to identify slow endpoints or database bottlenecks
Getting started with Sentry
The self-hosted version runs on Docker Compose or Kubernetes. The README links to deployment documentation rather than enumerating setup steps. For most languages, the SDK is installed as a package, your DSN is configured, and the client is initialized in your application entry point. JavaScript projects add a snippet to the HTML head. The exact SDK initialization varies by language.
Strengths
- โCaptures detailed debugging context (stack traces, breadcrumbs, user data) that plain logs miss
- โSupports over 20 programming languages and runtimes through official SDKs
- โSelf-hosting option gives complete data sovereignty with no per-seat licensing
- โRelease tracking correlates errors directly to deploys, narrowing down regressions fast
- โBreadcrumb system reconstructs the sequence of events before a crash
- โPerformance monitoring surfaces slow queries and transactions alongside errors
Known limitations and trade-offs
- โณSelf-hosting requires Docker Compose or Kubernetes knowledge and ongoing maintenance
- โณInitial setup is more involved than adding a simple logger to a project
- โณThe open-source version does not include some features available in the managed cloud tier
- โณStorage, backups, and scaling become your responsibility at scale
- โณLicense terms for self-hosting are not explicitly documented in the README
How Sentry compares to alternatives
Frequently asked questions about Sentry
Yes, the core platform is open source and can be self-hosted at no cost. You manage your own infrastructure, so while there is no per-seat fee, running the servers, databases, and handling scaling is your responsibility.
Sentry official SDKs cover over 20 languages: JavaScript, Python, Ruby, PHP, Go, Rust, Java, Kotlin, Swift, Objective-C, C#, C/C++, Dart, Perl, Clojure, Elixir, and Unity, Unreal Engine, Godot, and PowerShell.
Datadog is broader, covering infrastructure metrics, logs, and APM in one platform. Sentry is deeper on the developer debugging workflow: exception grouping, breadcrumbs, and release tracking are more granular. Teams often use both, with Sentry owning error tracking and Datadog owning infrastructure visibility.
Yes, self-hosting is a first-class option. The README references deployment documentation for Docker and Kubernetes. You control the environment, database, and retention policies entirely.
Each language has an SDK you install in your application. The SDK patches language-level exception handlers and, for performance monitoring, instruments framework hooks and HTTP calls. Events are sent over HTTPS to your Sentry instance using a per-project DSN URL.
The managed service offers a free tier with monthly event limits, though the exact caps are defined by Sentry's pricing page rather than the README. The self-hosted version has no event limits, but you operate the infrastructure that stores and processes them.
The problem it solves
Production errors arrive as raw exceptions or logs with no context for reproduction. Stack traces alone don't explain what the user was doing, which deployment introduced the bug, or whether a slow database query is correlated with the spike in 500s. Teams either fly blind or spend time replaying logs that were never designed to answer those questions.
Who should try it โ and who should skip
Development teams running multiple services or languages benefit most. If you need to correlate an error in your Python API with a slow database query and a specific release, Sentry ties those together. A solo developer or small team with a single application may find the setup overhead disproportionate to the benefit. Organizations with strict data residency or privacy requirements who want to own their debugging data rather than send it to a third party will appreciate the self-hosted option.
Related repositories
Is sentry worth your time?
ChatGPT, Claude and Perplexity can all read this page. Ask one of them what it makes of sentry.