Deploy

Creating a deployment produces a forwarder bundle and runs it on a target you control. The bundle is the same artifact whether you use the managed Fly path or download it and run it yourself.

Bundle contents

logtura-forwarder/
├── vector.yaml          # Generated Vector topology
├── Dockerfile           # Vector + driver helpers
├── bin/
│   └── logtura-cf-tail  # Driver helper binaries (only those the topology needs)
└── README.txt           # The exact command the deploy ran

The Vector config has three sections:

• sources — one entry per source kind, configured against the connection credentials.
• transforms — VRL transforms generated from the monitor pipeline.
• sinks — one entry per destination, plus the metrics sink that posts back to the control plane.

Example of a generated section for a Cloudflare worker tail driving a Slack destination:

sources:
  cf_workers_dirtsignal:
    type: exec
    command: ["/usr/local/bin/logtura-cf-tail", "--script", "dirtsignal"]
    streaming: true
    decoding: { codec: json }

sinks:
  slack_alerts:
    type: http
    inputs: [transform_errors_only]
    uri: https://hooks.slack.com/services/T.../B.../...
    encoding: { codec: json }

Managed Fly deploys

If you connect Fly as a deploy target, the control plane:

1. Renders the bundle.
2. Creates or updates a Fly app under your Fly organization, named logtura-fwd-<deploymentId>.
3. Builds the image inside Fly Machines using the Dockerfile from the bundle.
4. Sets the machine env vars: provider credentials, control-plane heartbeat URL, deployment ID.
5. Polls the Machines API and surfaces status in the deploy job.

The Fly app lives in your org and is billed to your Fly account. You can fly logs -a logtura-fwd-... or fly ssh console into it. The control plane does not need access to the app after it is created; it polls the heartbeat endpoint to confirm it is alive.

Self-deploy

To run the forwarder yourself, download the bundle from the deployment page and run it in your own environment:

tar xf logtura-forwarder.tgz
cd logtura-forwarder
docker build -t logtura-forwarder .
docker run --rm \
  -e CLOUDFLARE_ACCOUNT_ID=acct \
  -e CLOUDFLARE_API_TOKEN=token \
  -e SLACK_WEBHOOK_URL=https://hooks.slack.com/... \
  logtura-forwarder

The forwarder will tail the configured sources and post events. If you set LOGTURA_HEARTBEAT_URL, it will also post liveness to the control plane so the deployment status reflects reality.

Updates

When the source list, destination config, or driver package version changes, the deployment is marked bundleOutdated: true. The Deployments page shows a badge with the count. Click Redeploy to render a fresh bundle and update the running instance.