Introducing Pelagia: Open Source Lifecycle Management for Ceph on Kubernetes

Today, in line with our commitment to open source, Mirantis is releasing Pelagia—another foundational building block for modern cloud infrastructure that we are contributing to the community. Pelagia is a Kubernetes controller for lifecycle management of Ceph Software Defined Storage on Kubernetes, it is named after a jellyfish, continuing the marine theme common to projects in the Ceph ecosystem. The Pelagia controller reconciles the desired state of a Ceph cluster into concrete operations and exposes a concise set of high-level CRDs/APIs that offer a unified control surface for the cluster. These APIs simplify common management routines, making large, highly loaded storage clusters easier to operate.

Why Pelagia and How It Complements Rook

Managing Ceph via Rook alone can be heavy: there are many CRDs to coordinate and no built-in automated OSD decommissioning, which is especially painful on bare metal. We created Pelagia to close these gaps without replacing Rook but complementing it.

With Pelagia, cloud operators can interact with a small, opinionated set of higher-level CRDs/APIs instead of juggling dozens of low-level objects. This simplifies day-2 operations, improves operator’s transparency into the cluster status, and fits cleanly into GitOps workflows. Pelagia adds a higher-level controller that reconciles Pelagia’s intent-driven CRDs into Rook objects and orchestrates cross-component workflows (for instance, automated OSD decommission with disk wipe), thus complementing, not replacing, Rook’s controllers. 

Rook remains the vendor-neutral low-level Ceph orchestrator. Pelagia builds on top of Rook, adding lifecycle automation and the streamlined operator experience which is critical for large, bare-metal, and high-traffic environments. We’re open-sourcing Pelagia so the battle-proven Ceph lifecycle management used in Mirantis products becomes available to everyone.

How Pelagia and Rockoon Emerged from Mirantis OpenStack for Kubernetes

Pelagia was originally born as a key component of Mirantis OpenStack for Kubernetes (MOSK), where for more than 5 years it was used to deploy and manage large Ceph deployments (many petabytes of capacity), in production environments where reliability was paramount.

Going forward, MOSK will continue to rely on Pelagia—nothing is being kept proprietary—and the transition for existing MOSK customers will be seamless. We will keep developing Pelagia in the open, guided by the needs of our users and the community.

Rockoon is another major open source component of MOSK, published in January ‘25, a Kubernetes controller that serves as the operational “brain” for OpenStack on Kubernetes. Similar to how Pelagia approaches Ceph, Rockoon treats an OpenStack deployment as a unified system and provides a stable, high-level abstraction over OpenStack-Helm to streamline day-2 operations and ensure cluster reliability. 

The Rockoon and Pelagia projects integrate cleanly with each other, making it possible for anyone to stand up a fully open source, production-grade OpenStack cluster with a software defined storage backend, by simply deploying the respective operators in the same Kubernetes underlay.

Pelagia Standardizes Ceph for k0rdent

k0rdent is Mirantis’ innovative approach to assembling Kubernetes-native clouds from open, composable building blocks, delivered as a curated catalog for any conformant Kubernetes and a platform layer that brings VM-centric workloads alongside containers.

Pelagia as the storage lifecycle layer makes this approach practical at scale. By wrapping Rook with a focused set of high-level CRDs/APIs, Pelagia standardizes how Ceph is operated across clusters, teams, and environments. This reduces toil, clarifies health and status, and fits neatly into GitOps pipelines.

For Mirantisk0rdent Virtualization, which relies on KubeVirt to run VMs natively in Kubernetes, Pelagia provides predictable, repeatable Ceph lifecycle management for the PVCs/DataVolumes behind those VMs, enabling safer capacity changes and smoother day-2 operations.

Roadmap for k0rdent + Pelagia

• Add Pelagia to the k0rdent catalog as a supported component with opinionated defaults, tested integrations, and upgrade hooks implemented.

• Offer a guided migration for existing Ceph setups to Pelagia and use it as the primary storage layer for Mirantis k0rdent Virtualization.

Get Started with Pelagia & Join the Community

Simply install the Pelagia controlleron any Kubernetes cluster—see the Quick Start for steps.

For questions or feedback, join the community on GitHub Discussions or the #pelagia channel in CNCF Slack space. We welcome issue reports, feature proposals, documentation improvements, and real-world operational insights.

Mirantis Pelagia experts will also be onsite at OpenInfra Summit Europe 2025 in Paris-Saclay on October 17-19. Connect with us at the Mirantis table or check out our presentation on open storage. Learn more.

Conclusion

Pelagia distills years of operating Ceph at scale into an open, Kubernetes-native controller with a streamlined set of high-level CRDs/APIs for lifecycle management. It complements Rook, powers MOSK and integrates with Rockoon, and gives k0rdent a clean, open path for repeatable, supportable storage. Try it, deploy it, and help us shape the future of cloud-native storage ops!