Parameters and roles

The Deploy Ceph cluster plugin has the following parameters:

ceph_facts_template

The path to custom Ceph facts template. Decapod deploys the custom facts module on the nodes that collect the Ceph-related facts. Usually, you do not need to configure this parameter.

ceph_stable

Set to true if it is required to install Ceph from the stable repository.

ceph_stable_repo, ceph_stable_release, ceph_stable_distro_source

The options define the repository where to obtain Ceph. In case of Ubuntu Xenial, you will get the following repository string:

deb {{ ceph_stable_repo }} {{ ceph_stable_distro_source }} main

cluster

Defines the cluster name.

Important

Some tools require the ceph cluster name only. The default name allows executing the ceph utility without an explicit cluster name and with the --cluster option.

cluster_network

Defines the cluster network.

copy_admin_key

Copies the admin key to all nodes. This is required if you want to run the ceph utility from any cluster node. Keep this option as true. Otherwise, it may break some playbooks that maintain the lifecycle after deployment.

fsid

The unique identifier for your object store. Since you can run multiple clusters on the same hardware, you must specify the unique ID of the object store when bootstrapping a monitor.

journal_collocation

Defines if the OSD will place its journal on the same disk with the data. It is false by default.

If you want to have separate disks for journals (SSDs) and data (rotationals), set this to false. Also, set raw_multi_journal to true and list journal disks as raw_journal_devices.

raw_multi_journal

This option is the opposite to journal_collocation.

Note

The raw_multi_journal and journal_collocation options must have different values. For example, if journal_collocation is set to true, set raw_multi_journal to false.

dmcrypt_journal_collocation

This option has the same meaning as journal_collocation but both journal and data disks are encrypted by dmcrypt.

dmcrypt_dedicated_journal

This option has the same meaning as journal_collocation set to false. If dmcrypt_dedicated_journal is set to true, the journal and data will be placed on different disks and encrypted with dmcrypt.

journal_size

OSD journal size in megabytes.

max_open_files

Sets the number of open files to have on a node.

nfs_file_gw

Set to true to enable file access through NFS. Requires an MDS role.

nfs_obj_gw

Set to true to enable object access through NFS. Requires an RGW role.

os_tuning_params

Different kernels parameters. This is the list of dicts where name is the name of the parameter and value is the value.

public_network

Defines the public network.

monitor_interface

The option defines the NIC on the host that is connected to the public network.

devices

Defines the disks where to place the OSD data. If collocation is enabled, then journal devices, raw_journal_devices, are not used.

raw_journal_devices

Defines the disks where to place the journals for OSDs. If collocation is enabled, this option is not used.

The ceph-ansible project supports two deployment modes of a Ceph cluster: with journal collocation and on separate drives, and also with dmcrypt and without it. Therefore, there are four possible combinations.

The following table lists the possible combinations:

Setting	Combination 1	Combination 2	Combination 3	Combination 4
collocation	true	true	false	false
dmcrypt	true	false	true	false
journal_collocation	false	true	false	false
raw_multi_journal	true	false	false	true
dmcrypt_journal_collocation	false	false	false	false
dmcrypt_dedicated_journal	false	false	true	false
Data devices option name	devices	devices	devices	devices
Journal devices option name	–	–	raw_journal_devices	raw_journal_devices

Consider the different meaning of devices and raw_journal_devices in different modes: if no collocation is defined, then devices means disks with data. Journals are placed on raw_journal_devices disks. Otherwise, define devices only. In this case, the journal will be placed on the same device as the data.