Skip to main content

Advanced

This page contains information that is useful to know but that should not stop somebody from running a node.

Creating Your Own Private Network

The stellar-core GitHub repository holds a testnet.md file which contains a short tutorial demonstrating how to configure and run a short-lived, isolated test network.

Runtime Information: Start and Stop

Stellar-core can be started directly from the command line, or through a supervision system such as init, upstart, or systemd.

Stellar-core can be gracefully exited at any time by delivering SIGINT or pressing CTRL-C. It can be safely, forcibly terminated with SIGTERM or SIGKILL. The latter may leave a stale lock file in the BUCKET_DIR_PATH, and you may need to remove this file before stellar-core will restart. Broadly speaking, all components are designed to recover from abrupt termination.

Stellar-core can also be packaged in a container system such as Docker, so long as BUCKET_DIR_PATH and the database are stored on persistent volumes. For an example, see stellar/quickstart.

In-depth Architecture

The stellar-core GitHub repository also contains the architecture.md file, which describes how stellar-core is structured internally, how it is intended to be deployed, and the collection of servers and services needed to get the full functionality and performance.

Reproducible Performance Testing With Stellar Supercluster

Stellar Supercluster is a tool that enables running simulated networks of stellar-core nodes on Kubernetes clusters. One use case for Supercluster is running reproducible performance tests. The Supercluster GitHub repository contains a few documents to help users run theoretical maximum Transaction Per Second (TPS) tests:

  • doc/getting-started.md details dependencies for building and running Supercluster.
  • doc/eks.md details how to build a supercluster-compatible EKS cluster on AWS.
  • doc/theoretical-max-tps.md explains how to run the theoretical max TPS test on an EKS cluster. It also contains a table of results from SDF's own theoretical max TPS runs, as well as the configurations used to achieve those results.

Other Supercluster Resources

Supercluster is useful beyond reproducible performance testing. Some helpful resources to learn more about Supercluster include: