configure_rhel

Runs ansible/configure_rhel.yml, the general RHEL host baseline, letting the user choose which of its 43 domains actually apply for a given run via the configure_rhel_domains variable — without editin

SKILL.md

GitHub

configure_rhel

Runs ansible/configure_rhel.yml — the general host baseline playbook (see .claude/skills/soe/SKILL.md and docs/ARCHITECTURE.md for how it fits into the six-playbook layout). This skill is specifically about controlling which of its domains apply for a given run.

Policy & workload awareness

Each of this playbook's domains carries its own policy/workload staleness check in its own SKILL.md (see docs/ARCHITECTURE.md's "Policy & workload awareness"). When running a broad configure_rhel pass across many domains at once, prefer soe's repo-wide scan (.claude/skills/soe/SKILL.md) over checking each domain one at a time. This playbook's own file, and configure_rhel_domains's default list specifically, can itself need a change when a new workload or policy item implies a domain should move from off-by-default to on-by-default (or vice versa) — that's a change to ansible/configure_rhel.yml itself, not to any one role, so it follows the "cross-cutting change" path in docs/ARCHITECTURE.md's contribution workflow, not a single-domain branch.

The mechanism

Every one of the 43 in-repo roles this playbook can run is listed as:

- role: <name>
  tags: ["<name>"]
  when: "'<name>' in configure_rhel_domains"

configure_rhel_domains is a single list variable, defined once near the top of the playbook's vars: block. A role's tasks are only considered at all if its name is in that list — this is the actual on/off switch, not a comment to remove. --tags/--skip-tags still work as a second, independent filter on top: a role only runs if it's in configure_rhel_domains and matches the requested tags (both are verified separately, ANDed — confirmed by testing, not just by reading the YAML). --tags <domain> alone, without <domain> also being in configure_rhel_domains, does nothing.

Default value — reproduces exactly what this playbook applied before this variable existed:

ipv6_setup, etc_hosts, guest_agent, resolver_configuration, timesync,
packages_remove, boot_parameters, system_locale, audit_setup, certificates,
firewall, watchdog, packages_install, performance_tuning, root_password,
security_hardening, sebooleans, service_state, sshd_configuration,
system_init

Available, off by default — in the roles: list, fully functional, just not in the default configure_rhel_domains value:

system_hostname, repository_setup, dns_cache, timezone, system_keyboard,
multipath_setup, system_update_report_pre, system_update, accounts_policy,
accounts_local, domain_ad, ima_evm_setup, motd_issue, mount_setup,
rescue_image, packages_verify, shell_profile, troubleshooting_tools,
usbguard_setup, splunk_forwarder, system_coredump, scap_compliance,
cron_setup

That's 43 total. See each domain's own SKILL.md for what it actually does — this skill is about turning them on/off, not what each one enforces.

Not covered by configure_rhel_domains at all (stays commented-out in the file regardless, needs a direct edit — not a -e flag — to ever use): a handful of redhat.rhel_system_roles.*/community.general.* entries interleaved in the roles: list (external collection roles, not part of this repo's own ansible/roles/), and one stale reference to ipaclient, which was never an actual role in this repo (no ansible/roles/ipaclient/ exists) — don't treat it as a real, selectable domain.

-e replaces the list, it does not merge into it. To add usbguard_setup to an otherwise-normal baseline run, pass the full 21-item default list plus usbguard_setup — not just ["usbguard_setup"] on its own (that would run only usbguard_setup and skip everything else).

Some domains also gate themselves internally, independent of configure_rhel_domains — e.g. shell_profile's task still no-ops unless shell_profile_file is set, packages_verify still no-ops unless packages_verify_enable: true. Being in configure_rhel_domains gets a role a chance to run; it doesn't override that role's own "do nothing until configured" defaults. Check the domain's own SKILL.md/ defaults/main.yml for what else needs setting.

What to do

Figure out what the user actually wants applied. In a live conversation, this is a good fit for ask_user_input_v0 rather than asking the user to hand-write a domain list — e.g. present the 23 available-but-off domains (grouped sensibly: accounts/auth (accounts_local, accounts_policy, domain_ad), naming/discovery (system_hostname, dns_cache, timezone, system_keyboard), compliance/hardening (usbguard_setup, ima_evm_setup, scap_compliance, packages_verify), housekeeping (repository_setup, motd_issue, shell_profile, cron_setup, rescue_image, system_coredump, troubleshooting_tools), storage (multipath_setup, mount_setup), updates (system_update_report_pre, system_update), and splunk_forwarder) and ask which to add to the default 20. On the CLI with no interactive elicitation available, ask directly in plain text instead.

Audit (read-only, safe to run any time) — default domains:

ansible-playbook ansible/configure_rhel.yml --check --diff

Audit with a custom domain selection — pass the full list (default 20 plus whatever's being added/removed):

ansible-playbook ansible/configure_rhel.yml --check --diff \
  -e '{"configure_rhel_domains": ["ipv6_setup","etc_hosts","guest_agent","resolver_configuration","timesync","packages_remove","boot_parameters","system_locale","audit_setup","certificates","firewall","watchdog","packages_install","performance_tuning","root_password","security_hardening","sebooleans","service_state","sshd_configuration","system_init","usbguard_setup"]}'

(That example adds usbguard_setup to the default set. Swap in whatever the user actually chose. Removing a normally-active domain for one run works the same way — leave it out of the list, or use --skip-tags instead if that's clearer for a one-off exclusion without touching the variable at all.)

Audit one domain regardless of whether it's in the default set — combine --tags with a domain list that includes it:

ansible-playbook ansible/configure_rhel.yml --tags usbguard_setup --check --diff \
  -e '{"configure_rhel_domains": ["usbguard_setup"]}'

(Here it's fine for the configure_rhel_domains list to contain only the one domain being audited — --tags is already restricting to it, so the other defaults being absent from the list doesn't additionally narrow anything.)

Remediate: same commands without --check, only after the user has seen the diff and explicitly confirmed. Before remediating with a newly-added domain, check that domain's own SKILL.md for anything that needs deliberate confirmation first — several of the available-but-off domains are exactly the ones this repo is most cautious about:

  • usbguard_setup enables and starts USB enforcement (policy: reject) the moment it actually runs — confirm the host has console/physical access as a fallback, or a reviewed HID allow rule, before remediating.
  • accounts_local can delete users/groups (including home directories) if accounts_local_users_delete/accounts_local_groups_delete are set.
  • system_update can update dozens of packages and reboot the host.
  • multipath_setup reboots by default once it actually applies a change.
  • scap_compliance remediates (not just audits) whenever scap_compliance_remediate: true is set, regardless of --check.

See each domain's own SKILL.md for the full safety detail — this skill only decides whether a domain runs, not what it does once it does.

Propose a change to the default configure_rhel_domains list itself (e.g. "usbguard should just be on by default for all hosts now") — that's a persistent, cross-cutting change, not a one-off -e flag. Follow the same branch + PR workflow as any other change: branch soe/configure_rhel/<short-desc>, edit the default list in ansible/configure_rhel.yml, validate locally, push, open a PR titled [configure_rhel] <what changed>, then stop for human review — see docs/ARCHITECTURE.md's "Contribution workflow". Don't make this change unilaterally just because a user asked for one run with a domain enabled; a single -e flag for that run is the right scope unless they explicitly ask for the default to change.

Notes

  • This mechanism is specific to configure_rhel.yml. The other five playbooks (connect_linux.yml, load_balancer_setup.yml, nfs_client_setup.yml, nfs_server_setup.yml, update_rhel.yml) are each already scoped to one purpose and don't have (or need) an equivalent domain-selection variable — see their own SKILL.mds.
  • --tags/--skip-tags against configure_rhel.yml now work correctly for every domain, including ones outside the default configure_rhel_domains value (as long as the domain list is set to include them too — see above). This was verified directly: previously, bare role names in the roles: list had no implicit tag matching their own name, so --tags <domain> silently matched nothing for any domain — a pre-existing bug across this repo's docs, fixed as part of adding this toggle (every entry now carries an explicit tags: ["<name>"]). The same fix was applied to load_balancer_setup.yml, nfs_client_setup.yml, nfs_server_setup.yml, and update_rhel.yml for the same reason.