airbroke

Features

Airbroke is currently in active development! Using it in production environments is at your own discretion. We appreciate your feedback and support as we work towards a stable release.

• :floppy_disk: Based on PostgreSQL
• :globe_with_meridians: Airbrake(tm)/Sentry(tm)-compatible HTTP collector endpoints
• :computer: Modern, React-based frontend for error management
• :rocket: Designed with simplicity at its core
• :wrench: Maintains small database footprint even under heavy data ingestion
• :robot: Ask AI about issues
• :clipboard: Provide cURL command to reproduce HTTP exceptions
• :arrow_forward: Replay HTTP exceptions
• :key: Supports multiple OAuth providers for secure user authentication
• :bar_chart: Occurrence charts
• :bookmark: Save and manage bookmarks for important occurrences
• :electric_plug: MCP API for LLM/agent integrations

System Requirements

• Node.js 22/24+ compatible environment
• Minimum of 300MB RAM
• At least 1000 millicores, equivalent to 1 CPU core
• PostgreSQL 15+ database
• 8+ free database connections slots per instance

Deployment Options

Airbroke provides flexibility in deployment options. You can either deploy it from the built source code or use a multiarch Docker image. For Kubernetes deployments, a Helm chart is provided. As Airbroke is a Next.js 16 application, it can be deployed wherever a Node.js server is supported. This includes managed environments such as Vercel, Netlify, and Heroku.

Build from source

For a production build, you can run:

cp .env.dist .env

Then edit the .env file to set your own values.

yarn install
yarn build

This will generate a .next build output. Because this repo uses output: "standalone", the runnable server output is in .next/standalone (see the Dockerfile for a working copy strategy / asset layout).

You can also run yarn start to test the production build locally on port 3000.

Docker

We publish images for both amd64 and arm64 architectures on ghcr.io but in case you want to build your own image you can do so.

You can build the Docker image with:

docker build --no-cache -t icoretech/airbroke:latest .

You can then run the image locally with:

docker run -p 3000:3000 icoretech/airbroke:latest

Vercel

While testing on Vercel has not been conducted, Airbroke should be fully compatible.

It's important to keep the following points in mind:

• For optimal performance, ensure your database is located in the same region.
• The endpoints under /api/* will be converted into serverless functions, which may introduce potential cold boot time.
• If you connect directly to Postgres, use a pooler (or Prisma Accelerate / Data Proxy) to avoid exhausting DB connections.
• Migrations must be executed as part of deployment. The Deploy Button above sets a build-command that runs DB migrations only for VERCEL_ENV=production before building (adjust this in your Vercel project settings if you need a different workflow).
• The Deploy Button defaults AIRBROKE_CORS_ORIGINS to * for quick starts. After your first deploy, set it to your deployed origin (for example https://<project>.vercel.app) and any custom domains.

Detailed instructions for this process can also be found in the Prisma deployment guide for Vercel.

Render.com

The Render Deploy Button uses render.yaml (Render Blueprint). On the free tier, the template runs yarn db:migrate at boot (safe no-op if already applied). Airbroke allows all origins if AIRBROKE_CORS_ORIGINS is unset — once Render assigns your public URL, lock it down in the Render dashboard for production (this controls CORS for /api/v3/notices and /api/sentry/*/envelope).

Railway

Railway supports one-click deployments via Templates. Because templates are created and published from within Railway, you’ll need to create a template once and then wire up the button URL.

Steps:

• Create a Railway project from this repo (GitHub deploy).
• Add a PostgreSQL database in Railway.
• Set required environment variables (Railway can suggest vars from .env.dist).
• Publish the project as a Template and replace REPLACE_ME above with the template ID.

Netlify

The Deploy to Netlify button is configured via netlify.toml. The build command runs DB migrations only for production deploys.

Because you don't know your final *.netlify.app URL before the first deploy, you can start with AIRBROKE_CORS_ORIGINS=* and then tighten it to your deployed origin (for example https://<site>.netlify.app) and any custom domains.

Heroku

The Deploy to Heroku button uses app.json to provision add-ons and prompt for required environment variables.

Because you don't know your final *.herokuapp.com URL before the app is created, you can start with AIRBROKE_CORS_ORIGINS=* and then tighten it to your deployed origin (for example https://<app>.herokuapp.com) and any custom domains.

Helm

You can deploy Airbroke to Kubernetes using the dedicated Helm chart.

The Helm chart includes a values.yaml file with some default values that you can override with your own. It also includes a pgBouncer chart as optional dependency.

When using Helm we recommend using a GitOps approach to deploy your application(s), such as Flux.

Please find more information about the Helm chart in the dedicated repository: icoretech/charts

Docker Compose

For users who prefer Docker Compose for managing multi-container Docker applications, a docker-compose.yml file is provided at the root of the repository.

To get started, make sure you have Docker and Docker Compose installed on your system. Then, you can start the application using the following command in the terminal:

make build
make run

You can override the default values in the docker-compose.yml file by creating a docker-compose.override.yml file in the same directory. This file is ignored by Git and will not be committed to the repository.

Setup

Please view all the available configuration variables in the .env.dist file.

Airbroke requires DATABASE_URL at runtime. DIRECT_URL is required when you need a direct database connection for migrations (for example, when DATABASE_URL goes through PgBouncer or a data proxy). Some examples:

# Example Connection to Prisma Data Proxy
DATABASE_URL="prisma://__HOST__/?api_key=__KEY__"
# Example Connection to PostgreSQL
DATABASE_URL="postgresql://__USER__:__PASSWORD__@__HOST__:__PORT__/__DATABASE__?connection_limit=20&pool_timeout=10&application_name=airbroke"
# Direct connection to the database, used for migrations
DIRECT_URL="postgresql://__USER__:__PASSWORD__@__HOST__:__PORT__/__DATABASE__"

The optimal connection pool size without pgBouncer (connection_limit) can be calculated using the following formula:

connection_limit = (num_physical_cpus * 2 + 1) ÷ number_of_application_instances

For a system with 8 CPU cores and 3 application instances, the calculation would proceed as follows:

connection_limit = (8 * 2 + 1) ÷ 3
connection_limit = (16 + 1) ÷ 3
connection_limit = 17 ÷ 3
connection_limit ≈ 5.67

Since connection_limit must be an integer, it should be rounded down to the nearest whole number. In this scenario, each of the 3 application instances should have a connection_limit of 5.

This limit can be set in your connection strings.

After deployment, you should be able to