Chapter 1. Deploying Red Hat Ceph Storage in Containers

1. As the Ansible user, create the ceph-ansible-keys directory where Ansible stores temporary values generated by the ceph-ansible playbook.

   [user@admin ~]$ mkdir ~/ceph-ansible-keys

2. As root, create a symbolic link to the /usr/share/ceph-ansible/group_vars directory in the /etc/ansible/ directory:

   [root@admin ~]# ln -s /usr/share/ceph-ansible/group_vars /etc/ansible/group_vars

3. Navigate to the /usr/share/ceph-ansible/ directory:

   [root@admin ~]$ cd /usr/share/ceph-ansible

4. Create new copies of the yml.sample files:

   [root@admin ceph-ansible]# cp group_vars/all.yml.sample group_vars/all.yml
   [root@admin ceph-ansible]# cp group_vars/osds.yml.sample group_vars/osds.yml
   [root@admin ceph-ansible]# cp site-docker.yml.sample site-docker.yml

5. Edit the copied files.

   1. Edit the group_vars/all.yml file. See the table below for the most common required and optional parameters to uncomment. Note that the table does not include all parameters.

      Important

      Do not set the cluster: ceph parameter to any value other than ceph because using custom cluster names is not supported.

      Table 1.1. General Ansible Settings

      Option	Value	Required	Notes
      monitor_interface	The interface that the Monitor nodes listen to	monitor_interface, monitor_address, or monitor_address_block is required
      monitor_address	The address that the Monitor nodes listen to
      monitor_address_block	The subnet of the Ceph public network		Use when the IP addresses of the nodes are unknown, but the subnet is known
      ip_version	ipv6	Yes if using IPv6 addressing
      journal_size	The required size of the journal in MB	No
      public_network	The IP address and netmask of the Ceph public network	Yes	The Verifying the Network Configuration for Red Hat Ceph Storage section in the Installation Guide for Red Hat Enterprise Linux
      cluster_network	The IP address and netmask of the Ceph cluster network	No
      ceph_docker_image	rhceph/rhceph-3-rhel7, or cephimageinlocalreg if using a local Docker registry	Yes
      containerized_deployment	true	Yes
      ceph_docker_registry	registry.access.redhat.com, or <local-host-fqdn> if using a local Docker registry	Yes

      An example of the all.yml file can look like:

      monitor_interface: eth0
      journal_size: 5120
      public_network: 192.168.0.0/24
      ceph_docker_image: rhceph/rhceph-3-rhel7
      containerized_deployment: true
      ceph_docker_registry: registry.access.redhat.com

      For additional details, see the all.yml file.

   2. Edit the group_vars/osds.yml file. See the table below for the most common required and optional parameters to uncomment. Note that the table does not include all parameters.

      Important

      Use a different physical device to install an OSD than the device where the operating system is installed. Sharing the same device between the operating system and OSDs causes performance issues.

      Table 1.2. OSD Ansible Settings

      Option	Value	Required	Notes
      osd_scenario	collocated to use the same device for write-ahead logging and key/value data (BlueStore) or journal (FileStore) and OSD data non-collocated to use a dedicated device, such as SSD or NVMe media to store write-ahead log and key/value data (BlueStore) or journal data (FileStore) lvm to use the Logical Volume Manager to store OSD data	Yes	When using osd_scenario: non-collocated, ceph-ansible expects the numbers of variables in devices and dedicated_devices to match. For example, if you specify 10 disks in devices, you must specify 10 entries in dedicated_devices.
      osd_auto_discovery	true to automatically discover OSDs	Yes if using osd_scenario: collocated	Cannot be used when devices setting is used
      devices	List of devices where ceph data is stored	Yes to specify the list of devices	Cannot be used when osd_auto_discovery setting is used. When using lvm as the osd_scenario and setting the devices option, ceph-volume lvm batch mode creates the optimized OSD configuration.
      dedicated_devices	List of dedicated devices for non-collocated OSDs where ceph journal is stored	Yes if osd_scenario: non-collocated	Should be nonpartitioned devices
      dmcrypt	true to encrypt OSDs	No	Defaults to false
      lvm_volumes	A list of FileStore or BlueStore dictionaries	Yes if using osd_scenario: lvm and storage devices are not defined using devices	Each dictionary must contain a data, journal and data_vg keys. Any logical volume or volume group must be the name and not the full path. The data, and journal keys can be a logical volume (LV) or partition, but do not use one journal for multiple data LVs. The data_vg key must be the volume group containing the data LV. Optionally, the journal_vg key can be used to specify the volume group containing the journal LV, if applicable. See the examples below for various supported configurations.
      osds_per_device	The number of OSDs to create per device.	No	Defaults to 1
      osd_objectstore	The Ceph object store type for the OSDs.	No	Defaults to bluestore. The other option is filestore. Required for upgrades.

      The following are examples of the osds.yml file when using the three OSD scenarios: collocated, non-collocated, and lvm. The default OSD object store format is BlueStore, if not specified.

      Collocated

      osd_objectstore: filestore
      osd_scenario: collocated
      devices:
        - /dev/sda
        - /dev/sdb

      Non-collocated - BlueStore

      osd_objectstore: bluestore
      osd_scenario: non-collocated
      devices:
       - /dev/sda
       - /dev/sdb
       - /dev/sdc
       - /dev/sdd
      dedicated_devices:
       - /dev/nvme0n1
       - /dev/nvme0n1
       - /dev/nvme1n1
       - /dev/nvme1n1

      This non-collocated example will create four BlueStore OSDs, one per device. In this example, the traditional hard drives (sda, sdb, sdc, sdd) are used for object data, and the solid state drives (SSDs) (/dev/nvme0n1, /dev/nvme1n1) are used for the BlueStore databases and write-ahead logs. This configuration pairs the /dev/sda and /dev/sdb devices with the /dev/nvme0n1 device, and pairs the /dev/sdc and /dev/sdd devices with the /dev/nvme1n1 device.

      Non-collocated - FileStore

      osd_objectstore: filestore
      osd_scenario: non-collocated
      devices:
        - /dev/sda
        - /dev/sdb
        - /dev/sdc
        - /dev/sdd
      dedicated_devices:
         - /dev/nvme0n1
         - /dev/nvme0n1
         - /dev/nvme1n1
         - /dev/nvme1n1

      LVM simple

      osd_objectstore: bluestore
      osd_scenario: lvm
      devices:
        - /dev/sda
        - /dev/sdb

      or

      osd_objectstore: bluestore
      osd_scenario: lvm
      devices:
        - /dev/sda
        - /dev/sdb
        - /dev/nvme0n1

      With these simple configurations ceph-ansible uses batch mode (ceph-volume lvm batch) to create the OSDs.

      In the first scenario, if the devices are traditional hard drives or SSDs, then one OSD per device is created.

      In the second scenario, when there is a mix of traditional hard drives and SSDs, the data is placed on the traditional hard drives (sda, sdb) and the BlueStore database (block.db) is created as large as possible on the SSD (nvme0n1).

      LVM advance

      osd_objectstore: filestore
      osd_scenario: lvm
      lvm_volumes:
         - data: data-lv1
           data_vg: vg1
           journal: journal-lv1
           journal_vg: vg2
         - data: data-lv2
           journal: /dev/sda
           data_vg: vg1

      or

      osd_objectstore: bluestore
      osd_scenario: lvm
      lvm_volumes:
        - data: data-lv1
          data_vg: data-vg1
          db: db-lv1
          db_vg: db-vg1
          wal: wal-lv1
          wal_vg: wal-vg1
        - data: data-lv2
          data_vg: data-vg2
          db: db-lv2
          db_vg: db-vg2
          wal: wal-lv2
          wal_vg: wal-vg2

      With these advance scenario examples, the volume groups and logical volumes must be created beforehand. They will not be created by ceph-ansible.

      Note

      If using all NVMe SSDs set the osd_scenario: lvm and osds_per_device: 4 options. For more information, see the Configuring OSD Ansible settings for all NVMe Storage section in the Red Hat Ceph Storage Container Guide.

      For additional details, see the comments in the osds.yml file.

6. Edit the Ansible inventory file located by default at /etc/ansible/hosts. Remember to comment out example hosts.

   1. Add the Monitor nodes under the [mons] section:

      [mons]
      <monitor-host-name>
      <monitor-host-name>
      <monitor-host-name>

   2. Add OSD nodes under the [osds] section. If the nodes have sequential naming, consider using a range:

      [osds]
      <osd-host-name[1:10]>

      Note

      For OSDs in a new installation, the default object store format is BlueStore.

      Alternatively, you can colocate Monitors with the OSD daemons on one node by adding the same node under the [mons] and [osds] sections. See Chapter 2, Colocation of Containerized Ceph Daemons for details.

7. Optionally, for all deployments, bare-metal or in containers, you can create a custom CRUSH hierarchy using ansible-playbook:

   1. Setup your Ansible inventory file. Specify where you want the OSD hosts to be in the CRUSH map’s hierarchy by using the osd_crush_location parameter. You must specify at least two CRUSH bucket types to specify the location of the OSD, and one bucket type must be host. By default, these include root, datacenter, room, row, pod, pdu, rack, chassis and host.

      Syntax

      [osds]
      CEPH_OSD_NAME osd_crush_location="{ 'root': ROOT_BUCKET_', 'rack': 'RACK_BUCKET', 'pod': 'POD_BUCKET', 'host': 'CEPH_HOST_NAME' }"

      Example

      [osds]
      ceph-osd-01 osd_crush_location="{ 'root': 'default', 'rack': 'rack1', 'pod': 'monpod', 'host': 'ceph-osd-01' }"

   2. Set the crush_rule_config and create_crush_tree parameters to True, and create at least one CRUSH rule if you do not want to use the default CRUSH rules. For example, if you are using HDD devices, edit the paramters as follows:

      crush_rule_config: True
      crush_rule_hdd:
          name: replicated_hdd_rule
          root: root-hdd
          type: host
          class: hdd
          default: True
      crush_rules:
        - "{{ crush_rule_hdd }}"
      create_crush_tree: True

      If you are using SSD devices, then edit the parameters as follows:

      crush_rule_config: True
      crush_rule_ssd:
          name: replicated_ssd_rule
          root: root-ssd
          type: host
          class: ssd
          default: True
      crush_rules:
        - "{{ crush_rule_ssd }}"
      create_crush_tree: True

      Note

      The default CRUSH rules fail if both ssd and hdd OSDs are not deployed because the default rules now include the class parameter, which must be defined.

      Note

      Add the custom CRUSH hierarchy to the OSD files in the host_vars directory as described in a step below to make this configuration work.

   3. Create pools, with created crush_rules in group_vars/clients.yml file.

      Example

      copy_admin_key: True
      user_config: True
      pool1:
        name: "pool1"
        pg_num: 128
        pgp_num: 128
        rule_name: "HDD"
        type: "replicated"
        device_class: "hdd"
      pools:
        - "{{ pool1 }}"

   4. View the tree.

      [root@mon ~]# ceph osd tree

   5. Validate the pools.

      # for i in $(rados lspools);do echo "pool: $i"; ceph osd pool get $i crush_rule;done
      
      pool: pool1
      crush_rule: HDD

8. For all deployments, bare-metal or in containers, open for editing the Ansible inventory file, by default the /etc/ansible/hosts file. Comment out the example hosts.

   1. Add the Ceph Manager (ceph-mgr) nodes under the [mgrs] section. Colocate the Ceph Manager daemon with Monitor nodes.

      [mgrs]
      <monitor-host-name>
      <monitor-host-name>
      <monitor-host-name>

9. As the Ansible user, ensure that Ansible can reach the Ceph hosts:

   [user@admin ~]$ ansible all -m ping

10. As root, create the /var/log/ansible/ directory and assign the appropriate permissions for the ansible user:

    [root@admin ~]# mkdir /var/log/ansible
    [root@admin ~]# chown ansible:ansible  /var/log/ansible
    [root@admin ~]# chmod 755 /var/log/ansible

    1. Edit the /usr/share/ceph-ansible/ansible.cfg file, updating the log_path value as follows:

       log_path = /var/log/ansible/ansible.log

11. As the Ansible user, change to the /usr/share/ceph-ansible/ directory:

    [user@admin ~]$ cd /usr/share/ceph-ansible/

12. Run the ceph-ansible playbook:

    [user@admin ceph-ansible]$ ansible-playbook site-docker.yml

    Note

    If you deploy Red Hat Ceph Storage to Red Hat Enterprise Linux Atomic Host hosts, use the --skip-tags=with_pkg option:

    [user@admin ceph-ansible]$ ansible-playbook site-docker.yml --skip-tags=with_pkg

    Note

    To increase the deployment speed, use the --forks option to ansible-playbook. By default, ceph-ansible sets forks to 20. With this setting, up to twenty nodes will be installed at the same time. To install up to thirty nodes at a time, run ansible-playbook --forks 30 PLAYBOOK FILE. The resources on the admin node must be monitored to ensure they are not overused. If they are, lower the number passed to --forks.

13. Using the root account on a Monitor node, verify the status of the Ceph cluster:

    docker exec ceph-<mon|mgr>-<id> ceph health

    Replace:

    • <id> with the host name of the Monitor node:

    For example:

    [root@monitor ~]# docker exec ceph-mon-mon0 ceph health
    HEALTH_OK

Procedure

1. Set osd_scenario: lvm and osds_per_device: 4 in group_vars/osds.yml:

   osd_scenario: lvm
   osds_per_device: 4

2. List the NVMe devices under devices:

   devices:
     - /dev/nvme0n1
     - /dev/nvme1n1
     - /dev/nvme2n1
     - /dev/nvme3n1

3. The settings in group_vars/osds.yml will look similar to this example:

   osd_scenario: lvm
   osds_per_device: 4
   devices:
     - /dev/nvme0n1
     - /dev/nvme1n1
     - /dev/nvme2n1
     - /dev/nvme3n1

1. As the root user, navigate to the /usr/share/ceph-ansible/ directory.

   [root@admin ~]# cd /usr/share/ceph-ansible/

2. Uncomment the radosgw_interface parameter in the group_vars/all.yml file.

   radosgw_interface: interface

   Replace interface with the interface that the Ceph Object Gateway nodes listen to.

3. Optional. Change the default variables.

   1. Create a new copy of the rgws.yml.sample file located in the group_vars directory.

      [root@admin ceph-ansible]# cp group_vars/rgws.yml.sample group_vars/rgws.yml

   2. Edit the group_vars/rgws.yml file. For additional details, see the rgws.yml file.

4. Add the host name of the Ceph Object Gateway node to the [rgws] section of the Ansible inventory file located by default at /etc/ansible/hosts.

   [rgws]
   gateway01

   Alternatively, you can colocate the Ceph Object Gateway with the OSD daemon on one node by adding the same node under the [osds] and [rgws] sections. See Colocation of containerized Ceph daemons for details.

5. As the Ansible user, run the ceph-ansible playbook.

   [user@admin ceph-ansible]$ ansible-playbook site-docker.yml --limit rgws

   Note

   If you deploy Red Hat Ceph Storage to Red Hat Enterprise Linux Atomic Host hosts, use the --skip-tags=with_pkg option:

   [user@admin ceph-ansible]$ ansible-playbook site-docker.yml --skip-tags=with_pkg

6. Verify that the Ceph Object Gateway node was deployed successfully.

   1. Connect to a Monitor node as the root user:

      ssh hostname

      Replace hostname with the host name of the Monitor node, for example:

      [user@admin ~]$ ssh root@monitor

   2. Verify that the Ceph Object Gateway pools were created properly:

      [root@monitor ~]# docker exec ceph-mon-mon1 rados lspools
      rbd
      cephfs_data
      cephfs_metadata
      .rgw.root
      default.rgw.control
      default.rgw.data.root
      default.rgw.gc
      default.rgw.log
      default.rgw.users.uid

   3. From any client on the same network as the Ceph cluster, for example the Monitor node, use the curl command to send an HTTP request on port 8080 using the IP address of the Ceph Object Gateway host:

      curl http://IP-address:8080

      Replace IP-address with the IP address of the Ceph Object Gateway node. To determine the IP address of the Ceph Object Gateway host, use the ifconfig or ip commands:

      [root@client ~]# curl http://192.168.122.199:8080
      <?xml version="1.0" encoding="UTF-8"?><ListAllMyBucketsResult xmlns="http://s3.amazonaws.com/doc/2006-03-01/"><Owner><ID>anonymous</ID><DisplayName></DisplayName></Owner><Buckets></Buckets></ListAllMyBucketsResult>

   4. List buckets:

      [root@monitor ~]# docker exec ceph-mon-mon1 radosgw-admin bucket list

1. Add a new section [mdss] to the /etc/ansible/hosts file:

   [mdss]
   hostname
   hostname
   hostname

   Replace hostname with the host names of the nodes where you want to install the Ceph Metadata Servers.

   Alternatively, you can colocate the Metadata Server with the OSD daemon on one node by adding the same node under the [osds] and [mdss] sections. See Colocation of containerized Ceph daemons for details.

2. Navigate to the /usr/share/ceph-ansible directory:

   [root@admin ~]# cd /usr/share/ceph-ansible

3. Optional. Change the default variables.

   1. Create a copy of the group_vars/mdss.yml.sample file named mdss.yml:

      [root@admin ceph-ansible]# cp group_vars/mdss.yml.sample group_vars/mdss.yml

   2. Optionally, edit parameters in mdss.yml. See mdss.yml for details.

4. As the Ansible user, run the Ansible playbook:

   [user@admin ceph-ansible]$ ansible-playbook site-docker.yml --limit mdss

5. After installing Metadata Servers, configure them. For details, see the Configuring Metadata Server Daemons chapter in the Ceph File System Guide for Red Hat Ceph Storage 3.

1. Create the nfss file from the sample file:

   [root@ansible ~]# cd /usr/share/ceph-ansible/group_vars
   [root@ansible ~]# cp nfss.yml.sample nfss.yml

2. Add gateway hosts to the /etc/ansible/hosts file under an [nfss] group to identify their group membership to Ansible. If the hosts have sequential naming, use a range. For example:

   [nfss]
   <nfs_host_name_1>
   <nfs_host_name_2>
   <nfs_host_name[3..10]>

3. Navigate to the Ansible configuration directory, /etc/ansible/:

   [root@ansible ~]# cd /usr/share/ceph-ansible

4. To copy the administrator key to the Ceph Object Gateway node, uncomment the copy_admin_key setting in the /usr/share/ceph-ansible/group_vars/nfss.yml file:

   copy_admin_key: true

5. Configure the FSAL (File System Abstraction Layer) sections of the /usr/share/ceph-ansible/group_vars/nfss.yml file. Provide an ID, S3 user ID, S3 access key and secret. For NFSv4, it should look something like this:

   ###################
   # FSAL RGW Config #
   ###################
   #ceph_nfs_rgw_export_id: <replace-w-numeric-export-id>
   #ceph_nfs_rgw_pseudo_path: "/"
   #ceph_nfs_rgw_protocols: "3,4"
   #ceph_nfs_rgw_access_type: "RW"
   #ceph_nfs_rgw_user: "cephnfs"
   # Note: keys are optional and can be generated, but not on containerized, where
   # they must be configered.
   #ceph_nfs_rgw_access_key: "<replace-w-access-key>"
   #ceph_nfs_rgw_secret_key: "<replace-w-secret-key>"

   Warning

   Access and secret keys are optional, and can be generated.

6. Run the Ansible playbook:

   [user@admin ceph-ansible]$ ansible-playbook site-docker.yml --limit nfss

Procedure

1. As the root user, open and edit the /etc/ansible/hosts file. Add a node name entry in the iSCSI gateway group:

   Example

   [iscsigws]
   ceph-igw-1
   ceph-igw-2

2. Navigate to the /usr/share/ceph-ansible directory:

   [root@admin ~]# cd /usr/share/ceph-ansible/

3. Create a copy of the iscsigws.yml.sample file and name it iscsigws.yml:

   [root@admin ceph-ansible]# cp group_vars/iscsigws.yml.sample group_vars/iscsigws.yml

   Important

   The new file name (iscsigws.yml) and the new section heading ([iscsigws]) are only applicable to Red Hat Ceph Storage 3.1 or higher. Upgrading from previous versions of Red Hat Ceph Storage to 3.1 will still use the old file name (iscsi-gws.yml) and the old section heading ([iscsi-gws]).

   Important

   Currently, Red Hat does not support the following options to be installed using ceph-ansible for container-based deployments:

   • gateway_iqn
   • rbd_devices
   • client_connections

   See the Configuring the Ceph iSCSI gateway in a container section for instructions on configuring these options manually.

4. Open the iscsigws.yml file for editing.

5. Configure the gateway_ip_list option by adding the iSCSI gateway IP addresses, using IPv4 or IPv6 addresses:

   Example

   gateway_ip_list: 192.168.1.1,192.168.1.2

   Important

   You cannot use a mix of IPv4 and IPv6 addresses.

6. Optionally, uncomment the trusted_ip_list option and add the IPv4 or IPv6 addresses accordingly, if you want to use SSL. You will need root access to the iSCSI gateway containers to configure SSL. To configure SSL, do the following steps:

   1. If needed, install the openssl package within all the iSCSI gateway containers.

   2. On the primary iSCSI gateway container, create a directory to hold the SSL keys:

      # mkdir ~/ssl-keys
      # cd ~/ssl-keys

   3. On the primary iSCSI gateway container, create the certificate and key files:

      # openssl req -newkey rsa:2048 -nodes -keyout iscsi-gateway.key -x509 -days 365 -out iscsi-gateway.crt

      Note

      You will be prompted to enter the environmental information.

   4. On the primary iSCSI gateway container, create a PEM file:

      # cat iscsi-gateway.crt iscsi-gateway.key > iscsi-gateway.pem

   5. On the primary iSCSI gateway container, create a public key:

      # openssl x509 -inform pem -in iscsi-gateway.pem -pubkey -noout > iscsi-gateway-pub.key

   6. From the primary iSCSI gateway container, copy the iscsi-gateway.crt, iscsi-gateway.pem, iscsi-gateway-pub.key, and iscsi-gateway.key files to the /etc/ceph/ directory on the other iSCSI gateway containers.

7. Optionally, review and uncomment any of the following iSCSI target API service options accordingly:

   #api_user: admin
   #api_password: admin
   #api_port: 5000
   #api_secure: false
   #loop_delay: 1
   #trusted_ip_list: 192.168.122.1,192.168.122.2

8. Optionally, review and uncomment any of the following resource options, updating them according to the workload needs:

   # TCMU_RUNNER resource limitation
   #ceph_tcmu_runner_docker_memory_limit: 1g
   #ceph_tcmu_runner_docker_cpu_limit: 1
   
   # RBD_TARGET_GW resource limitation
   #ceph_rbd_target_gw_docker_memory_limit: 1g
   #ceph_rbd_target_gw_docker_cpu_limit: 1
   
   # RBD_TARGET_API resource limitation
   #ceph_rbd_target_api_docker_memory_limit: 1g
   #ceph_rbd_target_api_docker_cpu_limit: 1

9. As the Ansible user, run the Ansible playbook:

   [user@admin ceph-ansible]$ ansible-playbook site-docker.yml --limit iscsigws

   For Red Hat Enterprise Linux Atomic, add the --skip-tags=with_pkg option:

   [user@admin ceph-ansible]$ ansible-playbook site-docker.yml --limit iscsigws --skip-tags=with_pkg

10. Once the Ansible playbook has finished, open TCP ports 3260 and the api_port specified in the iscsigws.yml file on each node listed in the trusted_ip_list option.

    Note

    If the api_port option is not specified, the default port is 5000.