Rustinel writes one JSON alert per line to alerts.json.<date> in the alert directory, where <date> is the UTC date and the file rotates at midnight UTC.
Forward them with any shipper that tails NDJSON files, or have Rustinel POST each alert to a webhook.
The fields are described in Alert format.
The repository has runnable demos for Elastic and Splunk. Generate a test alert first with the Quickstart.
examples/siem/elastic starts Elasticsearch, Kibana, and Filebeat:
cd examples/siem/elastic
docker compose up -d elasticsearch kibana
RUSTINEL_ALERTS_DIR=/path/to/rustinel/logs docker compose up filebeatOpen Kibana at http://localhost:5601, create a data view for rustinel-alerts-*, and search event.kind : "alert".
The Filebeat input it uses:
filebeat.inputs:
- type: filestream
paths:
- /rustinel-logs/alerts.json.*
parsers:
- ndjson:
target: ""
add_error_key: trueexamples/siem/splunk starts Splunk with the HTTP Event Collector enabled and sends each alert as one event:
cd examples/siem/splunk
docker compose up -d
sudo python3 send-alerts.py /path/to/rustinel/logs/alerts.json.$(date -u +%Y-%m-%d)Open http://localhost:8000 (user admin, password ChangeMe123!) and search:
index=main source=rustinel sourcetype=_json event.kind=alert
The demo HEC token is rustinel-demo-token.
In production, create your own index and token and keep the token in a secret manager.
Add one [[alerts.webhook]] table per endpoint, then restart Rustinel:
[[alerts.webhook]]
url = "https://collector.example/rustinel"
secret = "<shared secret>"Each alert is sent as one JSON POST, signed with secret, and retried on failure.
The alert file is always written first, so a failing endpoint never loses an alert from it.
Options are in Webhook destinations, and the request format and a receiver example in Webhooks.
- Keep the alert directory on persistent storage.
- Ship the operational log (
rustinel.log.<date>) to a separate index from alerts. - Repeated identical alerts are collapsed, see Deduplication.
Sum
event.countto get true volumes. - On Linux and macOS the log directory is readable by root only, so the shipper must run as root.