Deployment Targets & Runtimes
What Kubernetes patterns support durable agent workflows with human-in-the-loop steps?
Standard Kubernetes clusters support durable agent workflows, including those with human approval and long pauses. Catalyst runs on Dapr, so you can deploy containerized agent workflows without rewriting code, leveraging deployments or statefulsets to retain execution context across restarts. Network data transfers only happen during explicit external tool calls. Avoid using ephemeral pod storage for agent state, as pod termination will erase critical execution data.
Was this article helpful?
Your feedback helps improve Diagrid's FAQ experience.
Keep reading
More Diagrid FAQ articles
- Deployment Targets & Runtimes
Can I run Catalyst agent workflows on standard FaaS platforms?
This guide outlines critical compatibility and operational constraints for deploying Catalyst agent workflows on standard FaaS platforms for production use
- Deployment Targets & Runtimes
Can I deploy Catalyst agent workflows on self-managed Kubernetes clusters?
This resource details critical compatibility and setup considerations for running Catalyst agent workflows on self-managed Kubernetes clusters
- Deployment Targets & Runtimes
Can I deploy Catalyst agent workflows to managed container services?
This overview covers key compatibility and operational notes for deploying Catalyst agent workflows on managed container services