Configuring the Data Processing (sahara) service (optional)


This feature is experimental at this time and it has not been fully production tested yet.

Sahara provide users with a simple means to provision data processing frameworks (such as Hadoop, Spark and Storm) on OpenStack.

Sahara is configured using the /etc/openstack_deploy/conf.d/sahara.yml file and the /etc/openstack_deploy/user_variables.yml file.

Configuring target hosts

Modify /etc/openstack_deploy/conf.d/sahara.yml by adding a list containing the infrastructure target hosts in the sahara-infra_hosts section:

In sahara.yml:


Replace *_IP_ADDRESS with the IP address of the br-mgmt container management bridge on each target host.

This hosts will be used to deploy the containers where sahara will be installed.

Configuring the cluster network

Sahara is configured to use the neutron implementation of OpenStack Networking.

Floating IP management

By default sahara is configured to use fixed IP addresses for access. This is controlled by the sahara_use_floating_ips variable. By changing sahara_use_floating_ips to True the user may specify a floating IP address pool for each node group directly.

In user_variables.yml:

sahara_use_floating_ips: False


When using floating IP addresses for management every instance in the cluster must have a floating IP address, otherwise sahara will not be able to utilize that cluster.

When using fixed IP addresses (sahara_use_floating_ips=False) the user will be able to choose the fixed IP network for all instances in a cluster. It is important to ensure that all instances running sahara have access to the fixed IP networks.

Object Storage access using proxy users

By default sahara is configured to use proxy users and delegated trusts for Object Storage access. In that way, users are not required to enter credentials for their data sources and job binaries referenced in Object Storage. To disable this functionality change the following variable to False.

In user_variables.yml:

sahara_use_domain_for_proxy_users: True

Also, is it possible to change which roles the trust users will receive in the proxy domain, by default it receives the _member_ role.

In user_variables.yml:

sahara_proxy_user_role_names: _member_


In the context of the proxy user, any roles that are required for Object Storage access by the project owning the object store must be delegated to the proxy user for authentication to be successful.

Configuring cluster instances NTP

By default sahara will enable the NTP service on all cluster instances if the NTP package is included in the image. The default NTP server will be this can be overridden using the sahara_default_ntp_server variable.

In user_variables.yml:

sahara_default_ntp_server: ""

Configuring plugins

The following plugins are installed and loaded by default:

  - vanilla
  - spark
  - cdh
  - ambari

To add/remove plugins, just change the sahara_plugin_base variable accordingly, in the user_variables.yml file.

Configuring notifications

Sahara can be configured to send notifications to the OpenStack Telemetry module. By default, the variable is set to true if there are any Ceilometer hosts in the environment. To change this, the following variable must be set:

In user_variables.yml:

sahara_ceilometer_enabled: True


To enable the Data Processing panel on horizon, the following variable should be set:

In user_variables.yml:

horizon_enable_sahara_ui: True

Setting up Sahara

Run the setup-hosts playbook, to create the sahara containers, and the repo-build playbook to update the repository with the sahara packages.

# cd /opt/openstack-ansible/playbooks
# openstack-ansible setup-hosts.yml
# openstack-ansible repo-build.yml

Run the sahara and horizon playbooks to install sahara and enable the Data Processing panel in horizon:

# cd /opt/openstack-ansible/playbooks
# openstack-ansible os-sahara-install.yml
# openstack-ansible os-horizon-install.yml