The DebOps playbooks are split into several files to allow partial usage:
site.yml | ┌---- <- service/core.yml |-- <- common.yml └-┐ |-- <- sys.yml |-- <- env.yml |-- <- net.yml |-- <- srv.yml |-- <- app.yml |-- <- virt.yml └-- <- hw.yml
When you run the :command:`debops` script or :command:`ansible-playbook`, you can either run the main :file:`site.yml` playbook, or specify name of the playbook you want to use to narrow the set of roles; this makes the Ansible runs shorter and lowers the startup time.
The order of the playbooks apart from the common ones (:file:`common.yml`, :file:`core.yml`) is not significant, although running roles that are used many times as role dependencies first might make the whole playbook run faster.
Common for all hosts
Main playbook, run by DebOps scripts by default, includes all other playbooks.
Playbook which runs on all hosts included in Ansible inventory. It executes a set of common roles which configure base services like SMTP service, a set of user accounts, :command:`ip(6)tables` firewall, APT repositories, and so on.
This is a playbook required on all hosts that use DebOps roles, regardless if you are using the playbooks or not. It's included by the :file:`common.yml` playbook. It will set up custom Ansible facts required by some of the roles, like root paths for several directory types, host UUID, installation of scripts that generate facts on the fly, and so on.
It also gathers the IP address of the Ansible Controller, or IP of the closest router which leads to it, to allow connections from that IP address through the firewall.
Activated by host groups
This playbook includes roles that configure services and resources that might be required by other roles, such as user and group accounts, authentication services like LDAP, network filesystems like NFS. Anything that is expected to be used by other roles further down the playbook, but is not common enough to be included in the :file:`common.yml` playbook, should be added here.
This is a playbook focused on programming language environments, like Ruby, PHP, Java, NodeJS. Since these might be used by multiple roles further down the playbook, they are grouped here to be run first so that other roles might be executed faster.
Playbook which focuses on roles that manage various network-related services, like DHCP, DNS, creating subnetworks or tunnels.
This playbook manages separate services like a webserver, various databases, file servers and others. These are usually standalone services which might be used by other roles down the line.
This playbook manages either end-user applications which might use multiple services (usually web applications like GitLab or phpIPAM) or end-point applications which can be used by other hosts in the cluster, like iPXE, or rsnapshot.
This playbook focuses on virtualization and hypervisors, like OpenVZ, KVM/libvirt or LXC.
At the end are roles which directly manage resources and services related to hardware, for example RAID health monitoring and notification.