How do I self-host Sentry, and what are the operational requirements vs using sentry.io?

Quick Answer: You can self-host Sentry using our open source stack (historically called “Self-Hosted Sentry”), but you’re trading SaaS convenience for running a full observability-style backend yourself: container orchestration, storage, upgrades, monitoring, and security. Using sentry.io offloads that operational overhead to us and gives you the latest features, higher scale, and managed security/compliance out of the box.

The Quick Overview

What It Is: Self-hosted Sentry is the on-premise / self-managed distribution of the Sentry application monitoring platform. You run the services (web, worker, ClickHouse, Kafka, Redis, Postgres, etc.) on your own infrastructure instead of using sentry.io.
Who It Is For: Teams with strong ops/SRE capacity, strict data residency rules beyond US/EU options, or very custom network constraints who are comfortable owning the reliability and upgrades of a fairly complex distributed system.
Core Problem Solved: Gives you full control over where Sentry runs and where data is stored—at the cost of taking on everything we do for you in the cloud: scaling, maintenance, security posture, and feature rollout.

How It Works

At a high level, self-hosting Sentry means you take the same core services that power sentry.io and deploy them into your own environment (bare metal, VMs, or Kubernetes). Our maintained self-hosted repo ships Docker images and orchestration manifests to bootstrap the stack; from there, you’re responsible for keeping it healthy.

With sentry.io, you send events via Sentry SDKs to our hosted, multi-tenant infrastructure on Google Cloud Platform. We operate and secure the stack, maintain data pipelines, and roll out new capabilities like Seer, Profiling, and Insights as they ship.

Install & Bootstrap:
- Self-hosted: Clone the self-hosted repository, configure environment variables, attach to your own Postgres, Redis, object storage, and run the provided Docker/docker-compose or Helm setup. Run migrations and create your first user.
- sentry.io: Create an account, create a project, install the SDK in your app, and point it at your DSN. No infra required.
Operate & Scale:
- Self-hosted: Monitor CPU, memory, and disk on core services; scale workers, web, and queues; tune retention windows; manage backups; and upgrade the stack on your own cadence.
- sentry.io: We handle scaling for 790B+ events per month across 150K+ customers—backed by GCP, TLS, AES‑256 encryption at rest and in transit, and annual third‑party penetration testing.
Maintain & Upgrade:
- Self-hosted: Pull new releases, test them in a staging environment, run migrations, and manage breaking changes. Expect some features to arrive later than on sentry.io or not at all.
- sentry.io: Features roll out automatically. You choose plans and quotas (errors, transactions, replays, attachments, monitors) and optionally add pay‑as‑you‑go and Seer (priced per active contributor).

Features & Benefits Breakdown

Core Feature	What It Does	Primary Benefit
Full Data Control (Self-Hosted)	Run Sentry on your own infrastructure, connect to your own Postgres, object storage, and network controls.	Meet custom data governance or air‑gapped requirements where SaaS isn’t an option.
Managed Scale & Reliability (sentry.io)	Hosted on Google Cloud Platform with multi‑tenant architecture, regional choice (US or Germany), and battle‑tested scaling.	Offload capacity planning, incident response, and infrastructure reliability to a team that runs Sentry at global scale.
Feature Velocity & Ecosystem (sentry.io)	Access to the latest Sentry capabilities (Session Replay, Seer, Profiling, Insights, SAML/SCIM, enterprise governance) as they ship.	Debug faster with the full platform—without waiting on or managing upgrades.

Note: Self-hosted gives you core Sentry workflows (Error Monitoring, Tracing, basic Dashboards, many integrations). sentry.io gives you “the whole thing,” including newer features and enterprise controls.

Ideal Use Cases

Best for self-hosted Sentry:
Because it lets you run everything behind your own firewalls when:
- You have strict on-prem or private cloud policies that SaaS can’t meet.
- Your security/legal teams require direct control over infrastructure, OS hardening, and network paths.
- You have an experienced ops/SRE team and existing monitoring around Kafka, ClickHouse, Redis, and Postgres.
Best for sentry.io:
Because it minimizes operational drag when:
- You want to instrument code quickly and get to debugging, not running observability infrastructure.
- You need SOC 2 Type II, ISO 27001, HIPAA attestation, and well‑documented security controls (TLS, AES‑256, annual pentests) without building that program yourself.
- You care about newer capabilities—Session Replay, Logs, Profiling, Insights, Seer, SAML/SCIM—shipping without your team doing upgrades.

Limitations & Considerations

Self-hosted is a non-trivial distributed system:
- You’re running a stack that looks a lot like a modern data platform: web, workers, cron, Postgres, Redis, Kafka, ClickHouse, symbolicator, relay, and more.
- You must provision and monitor:
  - Persistent disk with enough IOPS and headroom for event and index growth.
  - Object storage (for source maps, attachments, Session Replay assets if enabled).
  - Queues, caches, and long‑running workers.
- Workaround: treat Sentry like any other critical production service—run staging, use IaC (Terraform, Helm/Kustomize), and monitor it with your existing tooling.
Features and support differ from sentry.io:
- Not all Sentry cloud features are guaranteed to be available or first-class in self-hosted; cloud usually gets them first.
- Enterprise features (like SAML + SCIM on Business+, data residency selection in US/Germany, enterprise support) are designed around sentry.io.
- Workaround: if you need full platform + enterprise governance, lean toward sentry.io and use its existing security posture rather than rebuilding it.

Pricing & Plans

For self-hosted Sentry, the “price” isn’t a subscription—it’s the infrastructure and people you dedicate:

Compute for web, workers, cron, and supporting services.
Storage for Postgres, ClickHouse, and object storage (source maps, minidumps, Session Replay).
Ops/SRE time for upgrades, monitoring, backup/restore, and incident response.

You’re responsible for budgeting cloud/metal resources and the operational load.

For sentry.io, pricing is straightforward and usage-based:

You can get started for free.
Pricing depends on the number of monthly events (errors), transactions (for tracing), and attachments you send Sentry. You can also add quotas for Replays and Monitors.
You can:
- Set hard or flexible quotas and add pay‑as‑you‑go budget for overages.
- Reserve volume ahead of time for discounts (“pay ahead, save money… when you use more, you pay less”).
- Purchase Seer as an add-on priced per active contributor.

Common patterns:

Developer / Team plans: Best for product teams that want to instrument a few services or apps, get code‑level issues, and use core Dashboards (10 on Developer, 20 on Team).
Business / Enterprise plans: Best for organizations needing SAML + SCIM, advanced governance, organization audit logs, higher limits, and enterprise support (technical account manager, dedicated support on Enterprise).

For specifics, quotas, and current pricing details, see the pricing page on sentry.io.

Frequently Asked Questions

Can I run self-hosted Sentry in production at scale?

Short Answer: Yes, but you’re taking on a production-grade data system—plan for serious operations, not a side project.

Details:
Self-hosted Sentry is used in production by many teams, but it’s closer to running your own observability backend than hosting a small web app. You need to:

Size Postgres and ClickHouse for your expected event and transaction volume.
Run Kafka, Redis, and workers with enough capacity to handle spikes.
Implement backups and test restore procedures.
Monitor Sentry itself: queue depth, worker throughput, error rates, disk usage, and retention.

If your goal is “set it and forget it,” self-host probably isn’t the right fit. If your goal is “we already run Kafka/ClickHouse at scale and want to control Sentry’s deployment,” then it can be a good trade.

How does security differ between self-hosted Sentry and sentry.io?

Short Answer: Self-hosted security is entirely your responsibility; sentry.io gives you a documented, audited security posture out of the box.

Details:
With self-hosted Sentry, you own:

Infrastructure security: host hardening, patching, firewall rules, access control.
Network security: TLS termination, ingress/egress policies, VPN or zero‑trust access.
Data protection: disk encryption, backups, and retention policies.
Compliance: any audits or attestations (SOC, ISO, HIPAA) are on your organization.

With sentry.io, you inherit our security program:

Hosted on Google Cloud Platform with multi-tenant architecture.
Data encrypted using AES-256 at rest and TLS in transit.
Annual third‑party penetration testing.
Data residency choice: United States or Germany.
SOC 2 Type II, ISO 27001, and HIPAA attestation.
Enterprise governance options: SAML-based SSO, SCIM for automated provisioning (Business+), organization audit logs, and Enterprise support.

If your security posture is “we must keep everything on-prem,” self-hosted is your path—but you’re rebuilding the layers we already maintain in the cloud.

Summary

Self-hosting Sentry gives you maximum control over where your monitoring stack runs and how it’s operated, at the cost of owning a complex, stateful system end‑to‑end. You’ll need strong ops practices, capacity planning, and a clear upgrade process.

Using sentry.io shifts that work to us: you focus on instrumenting your applications with the Sentry SDK, tracing through services to identify poor-performing code, and using features like Error Monitoring, Tracing, Session Replay, Logs, Profiling, and Seer to debug faster. Our team manages infrastructure, scaling, security, and compliance so your team spends less time maintaining tools and more time fixing issues.

If you’re on the fence, a practical pattern is: start on sentry.io to understand your event volume and workflows, then reassess whether the operational cost of self-hosting makes sense for your organization.

Next Step

Get Started