[ English | русский | Indonesia ]

All-in-One (AIO) Ceph Deployment

OpenStack-Ansible provides a dedicated AIO Ceph scenario that deploys OpenStack together with a minimal Ceph cluster on a single host.

Important

The AIO Ceph scenario is intended for testing and development only. It must not be used in production environments. For production use, deploy a dedicated multi-node Ceph cluster.

This document describes how to enable the AIO Ceph scenario using the standard AIO quickstart procedure.

For the regular AIO workflow, refer to OpenStack-Ansible AIO Quickstart Guide

Overview

The AIO Ceph scenario extends the default All-in-One deployment by adding a minimal Ceph cluster to the same host. During bootstrap, Ceph services are installed and configured automatically.

Ceph OSDs are backed by sparse files created locally on the AIO host. This allows Ceph to operate without dedicated block devices, which makes the scenario suitable for labs and CI environments.

Enabling the AIO Ceph Scenario

Prepare the host as described in the standard AIO quickstart guide.

Before running the AIO bootstrap script, set the scenario variable to enable the Ceph-enabled layout:

export SCENARIO='aio_ceph'

Then start the bootstrap process:

scripts/bootstrap-aio.sh

Bootstrap Behavior

During the bootstrap process, OpenStack-Ansible prepares the host, creates the required containers, and deploys a minimal Ceph cluster.

As part of this process, three sparse files are automatically created. These files are used as Ceph OSD backing devices and together form the Ceph cluster on the single AIO node.

Because the OSDs are file-backed and all services run on a single host, this configuration is strictly for functional testing.

Ceph Services and Object Storage

The deployment includes Ceph MON, OSD, and MGR services, as well as Ceph RADOS Gateway (RGW).

After bootstrap completes, RGW is configured to provide a Swift-compatible object storage endpoint. This allows Ceph RADOS Gateway to act as a replacement for Swift and to be used as the object storage backend in the AIO environment.

Once configured, Object Storage will also be available to users through both the Skyline and Horizon dashboards.

For more details about using Ceph as a Swift replacement, see:

Using Ceph for Swift-compatible Object Storage

Verification

After running the OpenStack-Ansible playbooks, you can verify the Ceph cluster status. Since Ceph services run inside LXC containers, the command should be executed on the Ceph monitor container.

For example, you can use an Ansible ad-hoc command:

ansible -m command -a "ceph -s" ceph_mon[0]

A healthy cluster should report three OSDs up and the monitor in quorum.

For production deployments, use dedicated Ceph nodes with proper storage devices, redundancy, and failure domain design, see Ceph production example.

Enabling Ceph dashboard

The ceph-mgr role installs ceph-mgr-dashboard and enables the dashboard module by default, but does not finish configuring it. The steps below complete the setup.

Run the ceph commands on a monitor host (aio1 on an all-in-one or ceph-mon container) as a user with the cluster admin keyring.

  1. Configure the module. The simplest option is plain HTTP, which leaves dashboard traffic (including the admin login) unencrypted and is only suitable on a trusted network:

    # ceph config set mgr mgr/dashboard/ssl false

    Operators who require encryption can instead keep SSL enabled and load a certificate (self-signed or their own):

    # ceph config set mgr mgr/dashboard/ssl true
# ceph config-key set mgr/dashboard/crt -i dashboard.crt
# ceph config-key set mgr/dashboard/key -i dashboard.key

  2. Restart the module to apply the change:

    # ceph mgr module disable dashboard
# ceph mgr module enable dashboard

  3. Create a dashboard user. Recent releases require the password from a file via -i. Use administrator for full access or read-only for a view-only account:

    # echo 'SecretPassword' > /tmp/dash_pw
# ceph dashboard ac-user-create admin -i /tmp/dash_pw administrator
# shred -u /tmp/dash_pw

  4. Expose the dashboard through the load balancer by adding an HAProxy service in /etc/openstack_deploy/user_variables.yml (match haproxy_port to the port set above):

    haproxy_extra_services:
  - haproxy_service_name: ceph-dashboard
    haproxy_backend_nodes:
      - name: "{{ groups[mon_group_name][0] }}"
        ip_addr: "{{ hostvars[groups[mon_group_name][0]]['ansible_facts']['br_storage']['ipv4']['address'] }}"
    haproxy_port: 8443
    haproxy_balance_type: http

  5. Apply it with:

    # openstack-ansible openstack.osa.haproxy -t haproxy-config

The dashboard is then reachable on the external load balancer address, for example http://<external_lb_vip_address>:8443 (use https if SSL was left enabled). Log in with the account created above.