Use Rook to Orchestrate Distributed Open Source Storage

Rook provides cloud-native storage orchestration for the Ceph distributed open source storage system. Ceph runs on Kubernetes and is used to provide object, block, and file interfaces from a single cluster. Together, these provide a method to automate and manage large blocks of storage, typical of immense data storage centers. This guide demonstrates how to install and use Rook to orchestrate open source storage on a Kubernetes cluster.

What Is Rook and How Does It Work?

Rook automates the deployment and management of Ceph to create self-managing, self-scaling, and self-healing storage services. This combination can make large storage management easier than managing storage manually.

• CRUSH: Ceph provides file system, object, and block storage on a single cluster that is controlled using the Controlled Replication Under Scalable Hashing (CRUSH) algorithm. CRUSH ensures that the load placed on storage locations such as rows, racks, chassis, hosts, and devices remains consistent. A CRUSH map defines rules that specify how CRUSH stores and retrieves data.

• OSDs: Object Storage Daemons (OSDs) each manage a single device. Rook simplifies device management and performs tasks such as verifying OSD health.

• CRDs: Rook can also create and customize storage clusters through Custom Resource Definitions (CRDs) . There are four different modes in which to create a cluster:

  • Host Storage Cluster consumes storage from host paths and raw devices.
  • Persistent Volume Claim (PVC) Storage Cluster dynamically provisions storage underneath Rook by specifying the storage class for Rook to consume storage via PVCs.
  • Stretched Storage Cluster distributes Ceph Monitors (MONs) across three zones, while storage (i.e. OSDs) is only configured in two zones.
  • External Ceph Cluster connects your Kubernetes applications to an external Ceph cluster.

Before You Begin

1. Before you can work with Ceph and Rook, you need a Kubernetes cluster running Kubernetes version 1.28 or later with kubectl configured to communicate with your cluster. The recommended minimum Kubernetes setup for Rook includes three nodes with 4 GB memory and 2 CPUs each.

   To create a cluster and configure kubectl, follow the instructions in our Linode Kubernetes Engine - Get Started guide.

2. The Ceph storage cluster configured by Rook requires one of the following local storage options as a prerequisite :

   • Raw Devices
   • Raw Partitions
   • LVM Logical Volumes
   • Encrypted Devices
   • Multipath Devices
   • Persistent Volumes

   This guide utilizes Linode Block Storage Volumes attached to individual cluster nodes to demonstrate initial Rook and Ceph configuration. For production environments, we recommend utilizing persistent volumes or another local storage option.

Creating and Attaching Volumes

Once your Kubernetes cluster is set up, use the steps below to create and attach a volume for each node. This is to satisfy one of the local storage prerequisites required by Rook for Ceph configuration.

1. Open the Cloud Manager and select Linodes from the left menu.

2. Select one of the Kubernetes nodes in your cluster. The node should have a name such as lke116411-172761-649b48bf69f8.

3. Open the Storage tab.

4. Click Create Volume.

5. In the Label field, enter a name for the volume such as rook-volume-1.

6. Select a volume size of at least 40 GB.

7. Click Create Volume. After several minutes, the volume should show up as Active:

8. Repeat Steps 1 through 7 for remaining nodes in your cluster.

Rook and Ceph Installation

Once the local storage requirement has been met for the cluster, the next step is to install Rook and Ceph.

1. Use the following commands to install Rook and Ceph on the cluster:

   git clone --single-branch --branch v1.14.8 https://github.com/rook/rook.git
   cd rook/deploy/examples
   kubectl create -f crds.yaml -f common.yaml -f operator.yaml
   kubectl create -f cluster.yaml

   Once complete, you should see:

   ...
   cephcluster.ceph.rook.io/rook-ceph created

2. Verify the status of the Rook-Ceph cluster:

   kubectl -n rook-ceph get pod

   Sample output:

   NAME                                                              READY   STATUS      RESTARTS        AGE
   csi-cephfsplugin-2rqlv                                            2/2     Running     0               2m57s
   csi-cephfsplugin-4dl9l                                            2/2     Running     0               2m57s
   csi-cephfsplugin-provisioner-868bf46b56-2kt4d                     5/5     Running     0               2m58s
   csi-cephfsplugin-provisioner-868bf46b56-sx7fj                     5/5     Running     1 (2m21s ago)   2m58s
   csi-cephfsplugin-wjmmw                                            2/2     Running     0               2m57s
   csi-rbdplugin-4bq8k                                               2/2     Running     0               2m57s
   csi-rbdplugin-58wvf                                               2/2     Running     0               2m57s
   csi-rbdplugin-gdvjk                                               2/2     Running     0               2m57s
   csi-rbdplugin-provisioner-d9b9d694c-bl94s                         5/5     Running     2 (2m ago)      2m58s
   csi-rbdplugin-provisioner-d9b9d694c-vmhw2                         5/5     Running     0               2m58s
   rook-ceph-crashcollector-lke199763-288009-0c458f4a0000-7fcwlc5s   1/1     Running     0               81s
   rook-ceph-crashcollector-lke199763-288009-1f9ed47e0000-85b788df   1/1     Running     0               84s
   rook-ceph-crashcollector-lke199763-288009-28c3e5450000-978tzmcx   1/1     Running     0               69s
   rook-ceph-exporter-lke199763-288009-0c458f4a0000-7ffb6fbc5kmlbn   1/1     Running     0               81s
   rook-ceph-exporter-lke199763-288009-1f9ed47e0000-6dc9d57cbwj8m6   1/1     Running     0               84s
   rook-ceph-exporter-lke199763-288009-28c3e5450000-78cd58d665667r   1/1     Running     0               69s
   rook-ceph-mgr-a-b75b4b47-xxs4p                                    3/3     Running     0               85s
   rook-ceph-mgr-b-d77c5f556-nxw4w                                   3/3     Running     0               83s
   rook-ceph-mon-a-6f5865b7f8-577jw                                  2/2     Running     0               2m37s
   rook-ceph-mon-b-75ff6d4875-qq9nb                                  2/2     Running     0               106s
   rook-ceph-mon-c-6c7d4864b4-blnr8                                  2/2     Running     0               96s
   rook-ceph-operator-7d8898f668-nfp8w                               1/1     Running     0               3m45s
   rook-ceph-osd-prepare-lke199763-288009-0c458f4a0000-l665b         0/1     Completed   0               28s
   rook-ceph-osd-prepare-lke199763-288009-1f9ed47e0000-qlv2s         0/1     Completed   0               25s
   rook-ceph-osd-prepare-lke199763-288009-28c3e5450000-lm8ds         0/1     Completed   0               22s

Installing and Using the Rook-Ceph Toolbox

Once you have Rook and Ceph installed and configured, you can install and use the Ceph Toolbox :

1. Create the toolbox deployment:

   kubectl create -f toolbox.yaml

   deployment.apps/rook-ceph-tools created

2. Check the deployment status:

   kubectl -n rook-ceph rollout status deploy/rook-ceph-tools

   deployment "rook-ceph-tools" successfully rolled out

3. Access the toolbox pod:

   kubectl -n rook-ceph exec -it deploy/rook-ceph-tools -- bash

4. Check the Ceph status:

   ceph status

   At this point, your Rook-Ceph cluster should be in the HEALTH_OK state:

     cluster:
       id:     21676ecd-7f25-466d-9b90-a4ff13d2c0b5
       health: HEALTH_OK
   
     services:
       mon: 3 daemons, quorum a,b,c (age 103m)
       mgr: a(active, since 101m), standbys: b
       osd: 3 osds: 3 up (since 101m), 3 in (since 102m)
   
     data:
       pools:   1 pools, 1 pgs
       objects: 2 objects, 577 KiB
       usage:   26 MiB used, 120 GiB / 120 GiB avail
       pgs:     1 active+clean

   Cluster HEALTH Statuses

   Alternatively, the HEALTH_WARN state indicates that the cluster has no storage objects. While this doesn’t mean it won’t work, all I/O stops if it goes into a HEALTH_ERROR state. Consult the Ceph common issues troubleshooting guide to run diagnostics.

5. Check the Object Storage Daemon (OSD) status for Ceph’s distributed file system:

   ceph osd status

   ID  HOST                            USED  AVAIL  WR OPS  WR DATA  RD OPS  RD DATA  STATE
    0  lke195367-280905-1caa0a3f0000  9004k  39.9G      0        0       0        0   exists,up
    1  lke195367-280905-1d23f6860000  9004k  39.9G      0        0       0        0   exists,up
    2  lke195367-280905-55236f3a0000  8940k  39.9G      0        0       0        0   exists,up

6. Check the Ceph disk usage:

   ceph df

   --- RAW STORAGE ---
   CLASS     SIZE    AVAIL    USED  RAW USED  %RAW USED
   ssd    120 GiB  120 GiB  26 MiB    26 MiB       0.02
   TOTAL  120 GiB  120 GiB  26 MiB    26 MiB       0.02
   
   --- POOLS ---
   POOL  ID  PGS   STORED  OBJECTS     USED  %USED  MAX AVAIL
   .mgr   1    1  577 KiB        2  1.7 MiB      0     38 GiB

7. You can also use RADOS , a utility used to interact with Ceph object storage clusters, to check disk usage:

   rados df

   POOL_NAME     USED  OBJECTS  CLONES  COPIES  MISSING_ON_PRIMARY  UNFOUND  DEGRADED  RD_OPS       RD  WR_OPS       WR  USED COMPR  UNDER COMPR
   .mgr       1.7 MiB        2       0       6                   0        0         0     222  315 KiB     201  2.9 MiB         0 B          0 B
   
   total_objects    2
   total_used       26 MiB
   total_avail      120 GiB
   total_space      120 GiB

8. When complete, exit the toolbox pod, and return to the regular terminal prompt:

   exit

9. If you wish to remove the Rook-Ceph Toolbox, use the following command:

   kubectl -n rook-ceph delete deploy/rook-ceph-tools

   deployment.apps "rook-ceph-tools" deleted

Testing the Rook-Ceph Cluster

With the Rook-Ceph cluster set up, the links below offer some options for testing your storage:

• Block Storage
• Shared File System
• Object Stores

The Block Storage walkthrough is a recommended starting point for testing. This walkthrough creates a storage class then starts mysql and wordpress in your Kubernetes cluster, allowing access to WordPress from a browser.