Skip to main content

Self-hosting

Availability

Open core · self-host + all Splyntra Cloud plans

The source-available Splyntra core runs with a single command. It brings up the full pipeline — collector, security and evaluation services, dashboard, and their datastores — so you can send your first trace locally in minutes.

Start the stack​

git clone https://github.com/splyntra/splyntra
cd splyntra
docker compose up

Migrations are applied automatically on first start, so a fresh database is never missing a table.

Services and ports​

docker compose up starts the application services and their backing infrastructure:

ServicePortRole
collector4318OTLP/HTTP ingest and query API
security8001Detection (secrets, PII, moderation, injection)
evaluation8002Scorers and evaluation runs
web3000Dashboard

Backing infrastructure:

ServiceRole
postgres (16)Metadata
clickhouse (24.3)Traces, logs, metrics
valkey (7.2)Cache / queue
nats (2.10, JetStream)Messaging
minioS3-compatible storage for evaluation datasets

First run​

  1. Open the dashboard at http://localhost:3000.
  2. The collector accepts OTLP/HTTP at http://localhost:4318 (paths /v1/traces and /v1/logs).
  3. In development, the fallback ingest key splyntra_dev_key is accepted. It is rejected in production — set ENV/NODE_ENV to anything other than development and it fails closed. Mint a real key from the dashboard for any non-local use.

Send your first trace​

Point an SDK (or your OTLP exporter) at the local collector:

export SPLYNTRA_ENDPOINT="http://localhost:4318"
export SPLYNTRA_API_KEY="splyntra_dev_key"   # dev only

Run an instrumented agent once, then open Traces in the dashboard to see the span tree, its risk score, and cost. The Quickstart walks through instrumenting an agent end to end.

Next steps​