dbos-typescript

To build fault-tolerant TypeScript workflows, you must structure them deterministically and move non-deterministic logic into checkpointed steps. This ensures external operations recover correctly and avoid lost progress during failures.

Configuring queues for durable TypeScript workflows involves setting concurrency controls, partitioning work per entity, and applying rate limits for external APIs. This approach safely manages scale and prevents overloading downstream services.

External API calls in a durable workflow must be encapsulated within steps to ensure deterministic execution. Step checkpointing stores the results of these external operations, preventing lost progress and duplicate calls during recovery.

External clients can enqueue work into a partitioned queue using specific client enqueue patterns. This allows external systems to submit tasks directly to durable workflows while respecting concurrency and per-entity partitioning rules.

Events, messages, and streams provide real-time communication in TypeScript workflows by enabling progress notifications and streaming outputs. These patterns allow workflows to exchange data efficiently without blocking deterministic execution.

Patching and versioning durable workflows requires applying specific operational practices to ensure safe upgrades. This process maintains deterministic execution and workflow continuity when modifying steps or introducing new logic during runtime.