Stabilize Documentation
Stabilize is a lightweight, high-performance Python workflow engine designed for massively parallel execution and “enterprise-grade” reliability. It features a DAG-based orchestration model, robust concurrency control, and pluggable architecture.
Key Features
Reliable Execution: Built on the “Transactional Outbox” pattern with atomic state transitions.
Concurrency Control: Optimistic locking with phase-aware updates prevents “lost updates” in high-throughput environments.
Massive Parallelism: Supports dynamic forking, parallel branches, and joins.
Pluggable Backends: SQLite for development, PostgreSQL for production.
Resilience: Built-in retries, timeouts, circuit breakers, bulkheads, and bloom filter deduplication.
Structured Errors: Semantic error codes with chain traversal and automatic classification.
Configuration Versioning: Immutable configs with fingerprinting for reproducibility.
Event Sourcing: Full audit trail, event replay, projections, and time-travel debugging.
43 Workflow Patterns: Complete implementation of all WCP control-flow patterns (van der Aalst et al.) including OR-split/join, discriminator, N-of-M, deferred choice, milestones, mutex, signals, multi-instance, and structured loops.
Observability: Detailed structured logging and audit trails.
User Guide
API Reference
- Models
- Handlers
- Persistence
- Tasks
- Queue
MessageHandlerQueueProcessorQueueProcessorConfigSynchronousQueueProcessorMessageWorkflowLevelStartWorkflowCompleteWorkflowCancelWorkflowStartWaitingWorkflowsStageLevelStartStageCompleteStageSkipStageCancelStageRestartStageResumeStageContinueParentStageJumpToStageSignalStageCancelRegionAddMultiInstanceTaskLevelStartTaskRunTaskCompleteTaskPauseTaskInvalidWorkflowIdInvalidStageIdInvalidTaskIdInvalidTaskTypeget_message_type_name()create_message_from_dict()
- Events API