Sentry is a developer-first error tracking and performance monitoring platform. Your application reports exceptions and traces to Sentry, which groups them into issues, shows the stack trace with the surrounding source, and tells you which release, browser or device a failure came from — so a production crash becomes a fixable bug report instead of a log line nobody reads.
You create a project, install the Sentry SDK for your language and initialise it with the project’s DSN. Unhandled exceptions, handled errors and — once tracing is enabled — performance transactions then arrive automatically. Sentry groups events into issues by fingerprint, tracks first-seen and last-seen times, and flags a regression when a resolved issue comes back. An issue page gives you the stack trace with source context, the breadcrumbs recorded before the failure, tags, device and release information, and every event in that group.
Two routes. The hosted service at sentry.io is a SaaS with a free Developer plan (limited to one user, 5,000 errors and 5 million spans a month, 50 session replays) and paid plans billed as usage grows. Self-hosting uses the getsentry/self-hosted repository: a Docker Compose stack of Postgres, Redis, Kafka, ClickHouse, the Sentry web services and their workers, which the project describes as feature-complete but aimed at low-volume deployments and proofs of concept. Budget real memory and disk, and expect upgrade work between releases.
Not open source. Sentry is released under the Functional Source License 1.1 with an Apache 2.0 future licence (FSL-1.1-Apache-2.0). The source is public and you may self-host and modify it, but the licence forbids offering it as a competing product or service and is not OSI-approved; each release converts to Apache 2.0 on its second anniversary. Some components and SDKs carry their own licences, and Sentry the company sells the hosted service that funds the project.