Get Started

1. Pull the image

The image is published to Docker Hub on every release.

docker pull crashguardinc/crashguard.io:latest

2. Run it

The container exposes two ports: 8080 for the Admin App (nginx) and 5050 for the Engine API. Map them to whatever host ports you'd like:

docker run -d \
  --name crashguard \
  -p 8080:8080 \
  -p 8050:5050 \
  -v crashguard-data:/data \
  crashguardinc/crashguard.io:latest

Once it's running, open http://localhost:8080 for the Admin App. Point your services at http://localhost:8050 to create, checkpoint, and resolve canaries.

3. Point your client SDK at the Engine

The Crashguard.Client SDK doesn't assume a URL — you construct the underlying RestClient yourself and pass it in, pointed at whatever host port you mapped to the container's 5050:

var restClient = new RestClient("http://localhost:8050");
var client = new CrashguardClient(restClient);

This is a different URL than the Admin App's Admin portal URL setting in step 5 below — the SDK runs inside your own services and talks to the Engine directly, so it needs the Engine's published host port (8050 in the examples above), not the Admin App's port or address.

4. Persisting data

CrashGuard stores everything in a SQLite database at /data/crashguard.db inside the container. That path is declared as a Docker volume, so your data survives container restarts and recreation even if you don't specify a volume yourself — Docker will create an anonymous one for you.

Mounting your own data directory is optional, but recommended — it makes the database easy to find, back up, or move to a new host:

docker run -d \
  --name crashguard \
  -p 8080:8080 \
  -p 8050:5050 \
  -v /home/me/crashguard-data:/data \
  crashguardinc/crashguard.io:latest

Either way, don't run with --rm if you rely on the anonymous volume — removing the container without a named or bind-mounted volume can leave you without an easy way to locate your data afterward.

5. docker-compose, if you prefer

services:
  crashguard:
    image: crashguardinc/crashguard.io:latest
    ports:
      - "8080:8080"
      - "8050:5050"
    volumes:
      - crashguard-data:/data

volumes:
  crashguard-data:

Run it with docker compose up -d.

6. Set the admin portal URL

Open the Admin App and go to Settings, then set Admin portal URL to whatever address you actually use to reach it in your browser — in the examples above, that's http://localhost:8080.

This isn't optional cosmetics — alerts (Slack, email, etc.) use it to link directly to the canary that triggered them. Without it, alerts have no way to point you anywhere useful.

The Engine can't figure this out on its own. It runs in the same container as the Admin App, behind an internal reverse proxy, so it only ever sees its own internal ports — it has no visibility into whatever host port you mapped with -p when you started the container. If you remap ports (as in the examples above, where the internal nginx port 8080 is exposed as 8080 on the host), this field is the only place that mapping is recorded — leave it blank and alerts simply omit the link rather than guess wrong.