Inventory

Each Pigsty deployment has an inventory that describes the infrastructure and database clusters.

Configuration File

By default, Pigsty uses Ansible YAML configuration format. The inventory is a single file, pigsty.yml.

~/pigsty
  ^---- pigsty.yml   # <---- Default configuration file

Edit this file to customize the deployment. Alternatively, use the configure wizard to generate it automatically.

Configuration Structure

The inventory uses Ansible YAML configuration format. It contains global parameters in all.vars and groups in all.children.

Define clusters in all.children. Define infrastructure settings in all.vars, as this example shows:

all:                  # Top-level object: all
  vars: {...}         # Global parameters
  children:           # Group definitions
    infra:            # Group definition: 'infra'
      hosts: {...}        # Group members: 'infra'
      vars:  {...}        # Group parameters: 'infra'
    etcd:    {...}    # Group definition: 'etcd'
    pg-meta: {...}    # Group definition: 'pg-meta'
    pg-test: {...}    # Group definition: 'pg-test'
    redis-test: {...} # Group definition: 'redis-test'
    # ...

Cluster Definition

Each Ansible group can represent a cluster. Examples include node, PostgreSQL, Redis, Etcd, and Silo clusters.

A cluster definition contains members (hosts) and parameters (vars). Define members in <cls>.hosts. Set configuration parameters in <cls>.vars. This example defines a three-node HA PostgreSQL cluster:

all:
  children:    # Ansible group list
    pg-test:   # Ansible group name
      hosts:   # Ansible group instances (cluster members)
        10.10.10.11: { pg_seq: 1, pg_role: primary } # Host 1
        10.10.10.12: { pg_seq: 2, pg_role: replica } # Host 2
        10.10.10.13: { pg_seq: 3, pg_role: offline } # Host 3
      vars:    # Ansible group variables (cluster parameters)
        pg_cluster: pg-test

Cluster vars override global parameters. Instance vars override both cluster and global parameters.

Splitting Configuration

For a large deployment, you can split the inventory into multiple files. This can also help organize and maintain configuration.

inventory/
├── hosts.yml              # Host and cluster definitions
├── group_vars/
│   ├── all.yml            # Global default variables (corresponds to all.vars)
│   ├── infra.yml          # infra group variables
│   ├── etcd.yml           # etcd group variables
│   └── pg-meta.yml        # pg-meta cluster variables
└── host_vars/
    ├── 10.10.10.10.yml    # Specific host variables
    └── 10.10.10.11.yml

Put cluster members in hosts.yml. Put cluster configuration parameters in the corresponding files under group_vars.

Switching Configuration

When you run a playbook, use -i to temporarily select a different inventory.

./pgsql.yml -i another_config.yml
./infra.yml -i nginx_config.yml

Ansible also accepts local yaml|ini configuration files, a CMDB, or other dynamic configuration scripts.

The ansible.cfg file in the Pigsty home directory sets the default inventory to pigsty.yml in that directory. You can change this setting.

[defaults]
inventory = pigsty.yml

You can store the inventory in a CMDB metabase to integrate it with existing systems.