Operating Live Trading Nodes with Configuration, Persistence, and Shutdown

Summary

This guide explains how to configure and operate a live trading node, including its core settings, client registration, cache and message bus backings, strategy configuration, and shutdown behavior. It emphasizes running live nodes as standalone scripts or services, keeping only one node per process, and avoiding blocking work on the event loop because delays can lead to stale data, missed fills, and late order submissions. It also explains trader ID tag uniqueness and the implications for generated order, order-list, and position identifiers.

The persistence sections distinguish cache snapshots from actor and strategy state, outline Redis and Postgres backing behavior, and warn that state saving occurs at most once per run rather than as continuous checkpointing. The guide notes that database-backed nodes must use the synchronous run path and that node disposal is needed to flush buffered writes. This is operational documentation rather than evidence of strategy performance; the provided defaults and examples must be matched to the deployment and runtime requirements.

Key ideas

• Run live nodes as standalone processes and use one LiveNode per process.
• Keep callbacks and event-loop work short, offloading heavy computations and synchronous I/O.
• Use unique trader ID tags because generated order and position identifiers include the tag.
• Redis and Postgres support cache persistence, while actor and strategy state persistence requires Redis.
• State saves are not continuous, and disposing the node helps flush buffered writes during shutdown.

Tags

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.