mancube/dagron-engine repository overview

⁠dagron engine (mancube/dagron-engine)

The dagron workflow/DAG engine — runs DAGs of tasks with retries and dependencies, on a local, Docker, or Kubernetes executor, from one Rust binary.

dagron is a lightweight workflow scheduler in Rust. The engine parses a YAML DAG, reconciles pending → ready → running → done, claims work with optimistic CAS / FOR UPDATE SKIP LOCKED, dispatches it to a bounded worker pool, and decrements downstream dependencies — the control-plane half of the dagron stack.

• Image: mancube/dagron-engine — Rust binary on distroless/cc (glibc), runs as nonroot, no shell.
• Arch: linux/amd64, linux/arm64
• Binary inside: /usr/local/bin/dagron (entrypoint)
• Datastore: Postgres (DATABASE_URL)
• Website: dagron.dev · Source / full docs: github.com/lucheeseng827/dagron · Apache-2.0

⁠Tags

Pin in production: mancube/dagron-engine:0.11.1.

⁠Run

docker run \
  -e DATABASE_URL=postgres://dagron:dagron@postgres:5432/workflow \
  -e API_ADDR=0.0.0.0:8787 \
  -e EXECUTOR=local \
  mancube/dagron-engine:0.11.1

⁠Configuration (env)

The engine is built with FEATURES=postgres,ops,kubernetes: Postgres, the ops API and the Kubernetes executor. From 0.11.0 it adds otel,archive-s3,archive-gcs,archive-azure: OpenTelemetry (a trace context for each task, and OTLP span export once OTEL_EXPORTER_OTLP_ENDPOINT names a collector), and S3, Google Cloud Storage and Azure Blob for archived runs (GC_ARCHIVE_URL). Tags through 0.10.x do not have them: there GC_ARCHIVE_URL with a bucket is a startup error. An image built with them says so as it starts, on its configuration resolved line (features=…); one that prints no features= there is an earlier one. It needs a Postgres datastore — run it with dagron-api (auth, REST API and the console) for the full stack, or deploy everything with the Helm chart (oci://registry-1.docker.io/mancube/dagron).