Ceph default configuration

Pelagia Deployment Controller provides the capability to specify configuration options for the Ceph cluster through the rookConfig key-value section of the CephDeployment CR as if they were set in a usual ceph.conf file.

Default configuration options

However, if rookConfig is empty, Pelagia Deployment Controller still specifies the following default configuration options for each Ceph cluster:

• Required network parameters that you can change through the network section:

  [global]
  cluster network = <spec.network.clusterNet>
  public network = <spec.network.publicNet>
  

• General default configuration options that you can override using the rookConfig parameter:

  [global]
  mon_max_pg_per_osd = 300
  mon_target_pg_per_osd = 100
  
  [mon]
  mon_warn_on_insecure_global_id_reclaim = false
  mon_warn_on_insecure_global_id_reclaim_allowed = false
  
  [osd]
  osd_class_dir = /usr/lib64/rados-classes
  bdev_enable_discard = true
  bdev_async_discard_threads = 1
  

• If the spec.objectStorage.rgw section is defined, Pelagia specifies the following Ceph RADOS Gateway default configuration options, for example, for the rgw-store object store:

  [client.rgw.rgw.store.a]
  rgw_bucket_quota_ttl = 30
  rgw_data_log_backing = omap
  rgw_dns_name = rook-ceph-rgw-rgw-store.rook-ceph.svc
  rgw_max_attr_name_len = 64
  rgw_max_attr_size = 1024
  rgw_max_attrs_num_in_req = 32
  rgw_thread_pool_size = 256
  rgw_trust_forwarded_https = true
  rgw_user_quota_bucket_sync_interval = 30
  rgw_user_quota_sync_interval = 30
  

Rockoon-related default configuration options

If Pelagia is integrated with Rockoon, Pelagia Deployment Controller specifies the OpenStack-related default configuration options for a Ceph Object Storage instance only if all of the following conditions are met:

• The spec.objectStorage.rgw section is defined
• The spec.pools section contains at least one OpenStack pool with the images, vms, backup, or volumes role
• The openstack-rgw-creds secret provided by Rockoon exists in the OpenStack-Ceph shared namespace (openstack-ceph-shared by default)

These options are added on top of the general Ceph RADOS Gateway defaults above:

• Ceph Object Gateway options that you can override using the rookConfig parameter:

  [client.rgw.rgw.store.a]
  rgw_swift_account_in_url = true
  rgw_keystone_accepted_roles = '_member_, Member, member, swiftoperator'
  rgw_keystone_accepted_admin_roles = 'admin, service'
  rgw_keystone_implicit_tenants = true
  rgw_swift_versioning_enabled = true
  rgw_enforce_swift_acls = true
  rgw_s3_auth_use_keystone = true
  rgw_enable_usage_log = true
  rgw_usage_log_tick_interval = 30
  rgw_usage_log_flush_threshold = 1024
  rgw_usage_max_shards = 32
  rgw_usage_max_user_shards = 1
  

• Parameters for the Keystone integration taken from the openstack-rgw-creds secret. You cannot override rgw_keystone_api_version:

  rgw_keystone_api_version = 3
  rgw_keystone_url = <keystoneAuthURL>
  rgw_keystone_admin_user = <keystoneUser>
  rgw_keystone_admin_password = <keystonePassword>
  rgw_keystone_admin_domain = <keystoneProjectDomain>
  rgw_keystone_admin_project = <keystoneProjectName>
  

• If the secret contains barbican_url, parameters for the Barbican integration:

  rgw_crypt_s3_kms_backend = barbican
  rgw_barbican_url = <barbicanURL>
  rgw_keystone_barbican_user = <keystoneUser>
  rgw_keystone_barbican_password = <keystonePassword>
  rgw_keystone_barbican_domain = <keystoneProjectDomain>
  rgw_keystone_barbican_project = <keystoneProjectName>
  

  You cannot override rgw_crypt_s3_kms_backend.

See also

CephDeployment custom resource