Feature Request: Durable Workflow Persistence for Reliability

[chenwanqq]

Hi Maintainers,

I have been exploring the graphbit core codebase, specifically focusing on WorkflowExecutor and WorkflowContext in core/src/workflow.rs. I noticed that currently, the workflow execution state appears to be maintained primarily in-memory. This suggests that if the host process crashes or restarts, any active long-running workflows would be lost and cannot be recovered.

I believe adding Durable Workflow Persistence would be a significant enhancement for production reliability.
Is there an existing mechanism for persistence that I might have missed during my review? Is there currently any active work or planned roadmap items regarding workflow persistence? If this is a desirable feature and open for contribution, I am very interested in contributing this feature.

Thanks!

[JunaidHossain04]

Hi @chenwanqq,

Thank you for the thoughtful review of the GraphBit core and for raising this important point. You are correct that, at present, workflow execution state is primarily maintained in memory, and there is no built-in durable persistence or recovery mechanism for in-flight workflows.

We agree that durable workflow persistence is highly desirable for production use cases. While there is no finalized implementation yet, this area is under our consideration just like many others. We welcome contributions and are happy to discuss design ideas and review proposals for adding persistence and recovery support

If you’re interested in contributing, please feel free to checkout contributing guideline.

Thanks again for your interest and for taking the initiative to improve the framework.

Best regards,
GraphBit Team