SaltStack and Reclass metadata model

SaltStack and Reclass metadata model

SaltStack is an automation tool that executes formulas. Each SaltStack formula defines one component of the MCP cluster, such as MySQL, RabbitMQ, OpenStack services, and so on. This approach enables MCP product developers to combine the components as needed so that services do not interfere with each other and can be reused in multiple scenarios.

Reclass is an external node classifier (ENC) which enables cloud operators to manage an inventory of nodes by combining different classes into MCP cluster configurations. Reclass operates classes which you can view as tags or categories of metadata parameters.

The metadata model itself consists of hierarchically structured classes and corresponding parameters.

The following diagram displays the Mirantis Reclass metadata model’s hierarchy of classes:

MCP reclass classes
Service class System class Cluster class

A service class defines one service, or a group of related services, and the most specific configuration parameters for them. The parameters in this layer of the metadata model are translated directly into values in the configuration files for the corresponding service, and so on.

The service classes are provided by and match the Salt formulas installed onto the Salt Master node. A metadata parameter value defined in one of the service classes might be overridden by values from higher levels in the hierarchy, which include the system and cluster levels.

A system class defines a role (with different granularity) that is applied to a node, logical or physical. System classes typically include and combine service classes and other system classes in a way to describe completely configured, integrated, and ready-to-use system.

The system classes are distributed as a Git repository. The repository is copied to the Salt Master node during the bootstrap of DriveTrain.

A metadata parameter value set in a system class could be overridden by the values from a higher level in the hierarchy, which is the cluster level.

A cluster class defines configuration of a specific MCP cluster. This kind of classes can combine system classes according to the architecture of the cluster.

A cluster metadata model is typically generated using the automation pipeline that is executed by DriveTrain Jenkins. This pipeline uses Cookiecutter as a templating tool to generate the cluster model.

The cluster metadata model is distributed as a Git repository.