Notice

This document is for a development version of Ceph.

Installing Ceph

Choose a deployment method, plan the hardware and operating system, then set up the cluster.

Method

When to use it

Where to go

cephadm (recommended)

Installing and managing a cluster on Linux hosts that have systemd, Python 3, and Podman or Docker. Cephadm supports only Octopus and newer releases.

Using Cephadm to Deploy a New Ceph Cluster

Rook (recommended for Kubernetes)

Running Ceph in Kubernetes, with storage resources and provisioning managed through Kubernetes APIs, or connecting a Kubernetes cluster to an existing (external) Ceph cluster

rook.io

ceph-salt

Installing Ceph with Salt and cephadm

github.com/ceph/ceph-salt

Juju

Installing Ceph with Juju

charmhub.io/ceph-mon

Puppet

Installing Ceph with Puppet

github.com/openstack/puppet-ceph

OpenNebula HCI clusters

Deploying Ceph as the storage of OpenNebula 6.10 HCI clusters, on bare-metal servers in AWS or on premises (a legacy component that OpenNebula 7 no longer includes)

docs.opennebula.io (6.10)

Manual installation

Developing your own deployment scripts, or installing on hosts that cannot run cephadm, such as FreeBSD hosts

Installing Ceph Manually

Cephadm is fully integrated with the orchestrator API, so the CLI and dashboard features that manage cluster deployment are fully supported. Rook supports the orchestrator API but implements only some of its commands: for example, ceph orch cannot create or remove OSDs in a Rook cluster. Use the Rook operator for that.

Plan

  • Hardware Recommendations: how to choose CPUs, memory, storage devices, and networks for a cluster.

  • OS Recommendations: the distributions, kernels, and container hosts that each Ceph release is built and tested on.

  • Compatibility and Stability: the Podman versions that work with each Ceph release, and the cephadm features that are still under development.

Set Up

Look Up

Next Steps

  • Upgrading Ceph: upgrade a cluster that cephadm manages to a new release.

  • Cluster Operations: check the cluster’s health, monitor it, manage pools and data placement, and add or replace hardware.

Additional Resources

  • Cephadm: managing hosts and services, and the other tasks that cephadm performs after deployment.

  • Troubleshooting cephadm: what to check when a cephadm command fails or a service stops running.

  • Start Here: what Ceph is, and a single-host test cluster to learn on.

Brought to you by the Ceph Foundation

The Ceph Documentation is a community resource funded and hosted by the non-profit Ceph Foundation. If you would like to support this and our other efforts, please consider joining now.