feat(bootstrap): add rescue block with VM cleanup on failure

README.md

@@ -91,7 +91,7 @@ all:
       username: root@pam
       password: !vault |
         $ANSIBLE_VAULT...
-      host: pve01
+      node: pve01
       storage: local-lvm
 
   children:
@@ -268,7 +268,7 @@ The first user's credentials are prompted interactively via `vars_prompt` unless
 | `url`        | string | --      | API host (Proxmox/VMware)                            |
 | `username`   | string | --      | API username                                         |
 | `password`   | string | --      | API password                                         |
-| `host`       | string | --      | Proxmox node name                                    |
+| `node`       | string | --      | Target compute node (Proxmox node / VMware ESXi host; mutually exclusive with `cluster` on VMware) |
 | `storage`    | string | --      | Storage identifier (Proxmox/VMware)                  |
 | `datacenter` | string | --      | VMware datacenter                                    |
 | `cluster`    | string | --      | VMware cluster                                       |

inventory_example.yml

@@ -6,7 +6,7 @@ all:
       url: "pve01.example.com"
       username: "root@pam"
       password: "CHANGE_ME"
-      host: "pve01"
+      node: "pve01"
       storage: "local-lvm"
     boot_iso: "local:iso/archlinux-x86_64.iso"
   children:

main.yml

@@ -110,32 +110,81 @@
       ansible.builtin.import_role:
         name: system_check
 
-  roles:
-    - role: virtualization
-      when: system_cfg.type == "virtual"
-      become: false
-      vars:
-        ansible_connection: local
+  tasks:
+    - name: Bootstrap pipeline
+      block:
+        - name: Record that no pre-existing VM was found
+          ansible.builtin.set_fact:
+            _vm_absent_before_bootstrap: true
 
-    - role: environment
-      vars:
-        ansible_connection: "{{ 'vmware_tools' if hypervisor_type == 'vmware' else 'ssh' }}"
+        - name: Create virtual machine
+          when: system_cfg.type == "virtual"
+          ansible.builtin.include_role:
+            name: virtualization
+            public: true
+          vars:
+            ansible_connection: local
+            ansible_become: false
 
-    - role: partitioning
-      vars:
-        partitioning_boot_partition_suffix: 1
-        partitioning_main_partition_suffix: 2
+        - name: Configure environment
+          ansible.builtin.include_role:
+            name: environment
+            public: true
 
-    - role: bootstrap
+        - name: Partition disks
+          ansible.builtin.include_role:
+            name: partitioning
+            public: true
+          vars:
+            partitioning_boot_partition_suffix: 1
+            partitioning_main_partition_suffix: 2
 
-    - role: configuration
+        - name: Install base system
+          ansible.builtin.include_role:
+            name: bootstrap
+            public: true
 
-    - role: cis
-      when: system_cfg.features.cis.enabled | bool
+        - name: Apply system configuration
+          ansible.builtin.include_role:
+            name: configuration
+            public: true
 
-    - role: cleanup
-      when: system_cfg.type in ["virtual", "physical"]
-      become: false
+        - name: Apply CIS hardening
+          when: system_cfg.features.cis.enabled | bool
+          ansible.builtin.include_role:
+            name: cis
+            public: true
+
+        - name: Clean up and finalize
+          when: system_cfg.type in ["virtual", "physical"]
+          ansible.builtin.include_role:
+            name: cleanup
+            public: true
+          vars:
+            ansible_become: false
+
+      rescue:
+        - name: Delete VM on bootstrap failure
+          when:
+            - _vm_absent_before_bootstrap | default(false) | bool
+            - virtualization_vm_created_in_run | default(false) | bool
+            - system_cfg.type == "virtual"
+          ansible.builtin.include_role:
+            name: virtualization
+            tasks_from: delete
+          vars:
+            ansible_connection: local
+            ansible_become: false
+          tags:
+            - rescue_cleanup
+
+        - name: Fail host after bootstrap rescue
+          ansible.builtin.fail:
+            msg: >-
+              Bootstrap failed for {{ hostname }}.
+              {{ 'VM was deleted to allow clean retry.'
+                 if (virtualization_vm_created_in_run | default(false))
+                 else 'VM was not created in this run (kept).' }}
 
   post_tasks:
     - name: Set post-reboot connection flags

roles/environment/tasks/_configure_network.yml

@@ -13,6 +13,14 @@
         | default('')
       }}
 
+- name: Bring up network interface
+  when:
+    - hypervisor_type == "vmware"
+    - environment_interface_name | default('') | length > 0
+  ansible.builtin.command: "ip link set {{ environment_interface_name }} up"
+  register: environment_link_result
+  changed_when: environment_link_result.rc == 0
+
 - name: Set IP-Address
   when:
     - hypervisor_type == "vmware"
@@ -32,13 +40,31 @@
   register: environment_gateway_result
   changed_when: environment_gateway_result.rc == 0
 
+- name: Configure DNS resolvers
+  when:
+    - hypervisor_type == "vmware"
+    - system_cfg.network.dns.servers | default([]) | length > 0
+  ansible.builtin.copy:
+    dest: /etc/resolv.conf
+    content: |
+      {% for server in system_cfg.network.dns.servers %}
+      nameserver {{ server }}
+      {% endfor %}
+      {% if system_cfg.network.dns.search | default([]) | length > 0 %}
+      search {{ system_cfg.network.dns.search | join(' ') }}
+      {% endif %}
+    mode: "0644"
+
 - name: Synchronize clock via NTP
   ansible.builtin.command: timedatectl set-ntp true
   register: environment_ntp_result
   changed_when: environment_ntp_result.rc == 0
 
 - name: Configure SSH for root login
-  when: hypervisor_type == "vmware" and hypervisor_cfg.ssh | bool
+  when:
+    - hypervisor_type == "vmware"
+    - hypervisor_cfg.ssh | default(false) | bool
+    - system_cfg.network.ip is defined and system_cfg.network.ip | string | length > 0
   block:
     - name: Allow login
       ansible.builtin.replace:
@@ -58,7 +84,18 @@
         name: sshd
         state: reloaded
 
-    - name: Set SSH connection for VMware
+    - name: Switch to SSH connection
       ansible.builtin.set_fact:
         ansible_connection: ssh
         ansible_user: root
+        ansible_password: ""
+        ansible_host: "{{ system_cfg.network.ip }}"
+        ansible_ssh_extra_args: "-o UserKnownHostsFile=/dev/null -o StrictHostKeyChecking=no"
+
+    - name: Reset connection for SSH switchover
+      ansible.builtin.meta: reset_connection
+
+    - name: Verify SSH connectivity
+      ansible.builtin.wait_for_connection:
+        timeout: 30
+        delay: 2

roles/environment/tasks/_prepare_installer.yml

@@ -79,6 +79,13 @@
 # bootstrapping RHEL-family distros from the Arch ISO, where the
 # host rpm/dnf does not trust target distro GPG keys. Package
 # integrity is verified by the target system's own rpm after reboot.
+- name: Create RPM macros directory
+  when: is_rhel | bool
+  ansible.builtin.file:
+    path: /etc/rpm
+    state: directory
+    mode: "0755"
+
 - name: Relax RPM Sequoia signature policy for RHEL bootstrap
   when: is_rhel | bool
   ansible.builtin.copy:

roles/environment/tasks/main.yml

@@ -1,6 +1,6 @@
 ---
 - name: Configure work environment
-  become: "{{ hypervisor_type != 'vmware' }}"
+  become: "{{ (hypervisor_type | default('none')) != 'vmware' }}"
   block:
     - name: Detect and validate live environment
       ansible.builtin.include_tasks: _detect_live.yml

roles/global_defaults/defaults/main.yml

@@ -46,7 +46,7 @@ hypervisor_defaults:
   url: ""
   username: ""
   password: ""
-  host: ""
+  node: ""
   storage: ""
   datacenter: ""
   cluster: ""
@@ -136,10 +136,10 @@ system_defaults:
 # All virtual types additionally require network bridge or interfaces.
 hypervisor_required_fields:
   proxmox:
-    hypervisor: [url, username, password, host, storage]
+    hypervisor: [url, username, password, node, storage]
     system: [id]
   vmware:
-    hypervisor: [url, username, password, datacenter, cluster, storage]
+    hypervisor: [url, username, password, datacenter, storage]
     system: []
   xen:
     hypervisor: []

roles/global_defaults/tasks/_normalize_system.yml

@@ -77,7 +77,12 @@
           if (system_raw.mirror | default('') | string | trim | length) > 0
           else _mirror_defaults[system_raw.os | default('') | string | lower] | default('')
         }}
-      path: "{{ system_raw.path | default('') | string }}"
+      path: >-
+        {{
+          (system_raw.path | default('') | string)
+          if (system_raw.path | default('') | string | length > 0)
+          else (hypervisor_cfg.folder | default('') | string)
+        }}
       packages: >-
         {{
           (

roles/global_defaults/tasks/main.yml

@@ -32,12 +32,19 @@
       api_host: "{{ hypervisor_cfg.url }}"
       api_user: "{{ hypervisor_cfg.username }}"
       api_password: "{{ hypervisor_cfg.password }}"
-      node: "{{ hypervisor_cfg.host }}"
+      node: "{{ hypervisor_cfg.node }}"
   no_log: true
 
 - name: Normalize system inputs
   ansible.builtin.include_tasks: system.yml
 
+- name: Inherit folder from hypervisor when system path is empty
+  when:
+    - system_cfg.path | default('') | string | length == 0
+    - hypervisor_cfg.folder | default('') | string | length > 0
+  ansible.builtin.set_fact:
+    system_cfg: "{{ system_cfg | combine({'path': hypervisor_cfg.folder | string}, recursive=True) }}"
+
 - name: Validate variables
   ansible.builtin.include_tasks: validation.yml
 
@@ -85,3 +92,12 @@
   when: hypervisor_type == "vmware"
   ansible.builtin.set_fact:
     ansible_connection: vmware_tools
+    ansible_vmware_host: "{{ hypervisor_cfg.url }}"
+    ansible_vmware_port: 443
+    ansible_vmware_user: "{{ hypervisor_cfg.username }}"
+    ansible_vmware_password: "{{ hypervisor_cfg.password }}"
+    ansible_vmware_guest_path: "/{{ hypervisor_cfg.datacenter }}/vm{{ system_cfg.path }}/{{ hostname }}"
+    ansible_vmware_validate_certs: "{{ hypervisor_cfg.certs | bool }}"
+    ansible_vmware_tools_user: root
+    ansible_vmware_tools_password: "{{ system_cfg.root.password }}"
+  no_log: true

roles/global_defaults/tasks/system.yml

@@ -48,6 +48,28 @@
         }, recursive=True)
       }}
 
+- name: Populate primary network fields from first interface (pre-computed)
+  when:
+    - system_cfg is defined
+    - _bootstrap_needs_enrichment | default(false) | bool
+    - system_cfg.network.interfaces | default([]) | length > 0
+    - system_cfg.network.bridge | default('') | string | length == 0
+  vars:
+    _primary: "{{ system_cfg.network.interfaces[0] }}"
+  ansible.builtin.set_fact:
+    system_cfg: >-
+      {{
+        system_cfg | combine({
+          'network': system_cfg.network | combine({
+            'bridge': _primary.bridge | default(''),
+            'vlan': _primary.vlan | default(''),
+            'ip': _primary.ip | default(''),
+            'prefix': _primary.prefix | default(''),
+            'gateway': _primary.gateway | default('')
+          })
+        }, recursive=True)
+      }}
+
 - name: Derive convenience facts from pre-computed system_cfg
   when:
     - system_cfg is defined

roles/global_defaults/tasks/validation.yml

@@ -166,6 +166,23 @@
     label: "hypervisor.{{ item }}"
   no_log: true
 
+- name: Validate VMware placement (cluster or node required, mutually exclusive)
+  when:
+    - system_cfg.type == "virtual"
+    - hypervisor_type == "vmware"
+  ansible.builtin.assert:
+    that:
+      - >-
+        (hypervisor_cfg.cluster | default('') | string | length > 0)
+        or (hypervisor_cfg.node | default('') | string | length > 0)
+      - >-
+        (hypervisor_cfg.cluster | default('') | string | length == 0)
+        or (hypervisor_cfg.node | default('') | string | length == 0)
+    fail_msg: >-
+      VMware requires either hypervisor.cluster or hypervisor.node (mutually exclusive).
+      cluster targets a vSphere cluster; node targets a specific ESXi host.
+    quiet: true
+
 - name: Validate hypervisor-specific required system fields
   when:
     - system_cfg.type == "virtual"
@@ -293,8 +310,8 @@
     system_disk_mounts: >-
       {{
         (system_cfg.disks | default([]))
-        | map(attribute='mount')
-        | map(attribute='path')
+        | map(attribute='mount', default={})
+        | map(attribute='path', default='')
         | map('string')
         | map('trim')
         | reject('equalto', '')

roles/system_check/tasks/main.yml

@@ -45,7 +45,7 @@
       block:
         - name: Query Proxmox for existing VM
           community.proxmox.proxmox_vm_info:
-            node: "{{ hypervisor_cfg.host }}"
+            node: "{{ hypervisor_cfg.node }}"
             vmid: "{{ system_cfg.id }}"
             name: "{{ hostname }}"
             type: qemu
@@ -66,6 +66,7 @@
     - name: Check VM existence in vCenter
       when: hypervisor_type == "vmware"
       delegate_to: localhost
+      become: false
       module_defaults:
         community.vmware.vmware_guest_info:
           hostname: "{{ hypervisor_cfg.url }}"
@@ -106,6 +107,7 @@
     - name: Check if VM already exists on Xen
       when: hypervisor_type == "xen"
       delegate_to: localhost
+      become: false
       ansible.builtin.command:
         argv:
           - xl

roles/virtualization/tasks/delete.yml

@@ -0,0 +1,79 @@
+---
+- name: Delete VMware VM
+  when: hypervisor_type == "vmware"
+  delegate_to: localhost
+  community.vmware.vmware_guest:
+    hostname: "{{ hypervisor_cfg.url }}"
+    username: "{{ hypervisor_cfg.username }}"
+    password: "{{ hypervisor_cfg.password }}"
+    validate_certs: "{{ hypervisor_cfg.certs | bool }}"
+    datacenter: "{{ hypervisor_cfg.datacenter }}"
+    name: "{{ hostname }}"
+    folder: "{{ system_cfg.path if system_cfg.path | string | length > 0 else omit }}"
+    state: absent
+    force: true
+  no_log: true
+
+- name: Delete Proxmox VM
+  when: hypervisor_type == "proxmox"
+  delegate_to: localhost
+  community.proxmox.proxmox_kvm:
+    api_host: "{{ hypervisor_cfg.url }}"
+    api_user: "{{ hypervisor_cfg.username }}"
+    api_password: "{{ hypervisor_cfg.password }}"
+    node: "{{ hypervisor_cfg.node }}"
+    vmid: "{{ system_cfg.id | default(omit, true) }}"
+    name: "{{ hostname }}"
+    state: absent
+    force: true
+  no_log: true
+
+- name: Destroy libvirt VM
+  when: hypervisor_type == "libvirt"
+  delegate_to: localhost
+  block:
+    - name: Stop libvirt VM
+      community.libvirt.virt:
+        name: "{{ hostname }}"
+        state: destroyed
+        uri: "{{ libvirt_uri | default('qemu:///system') }}"
+      failed_when: false
+
+    - name: Undefine libvirt VM
+      community.libvirt.virt:
+        name: "{{ hostname }}"
+        command: undefine
+        uri: "{{ libvirt_uri | default('qemu:///system') }}"
+      failed_when: false
+
+    - name: Remove libvirt disk images
+      ansible.builtin.file:
+        path: "{{ item.path }}"
+        state: absent
+      loop: "{{ virtualization_libvirt_disks | default([]) }}"
+      loop_control:
+        label: "{{ item.path | default('unknown') }}"
+
+    - name: Remove libvirt cloud-init disk
+      ansible.builtin.file:
+        path: "{{ virtualization_libvirt_cloudinit_path | default('/dev/null') }}"
+        state: absent
+      when: virtualization_libvirt_cloudinit_path is defined
+
+- name: Destroy Xen VM
+  when: hypervisor_type == "xen"
+  delegate_to: localhost
+  block:
+    - name: Stop Xen VM
+      ansible.builtin.command:
+        argv:
+          - xl
+          - destroy
+          - "{{ hostname }}"
+      failed_when: false
+
+    - name: Remove Xen VM config
+      ansible.builtin.file:
+        path: "/etc/xen/{{ hostname }}.cfg"
+        state: absent
+      failed_when: false

roles/virtualization/tasks/vmware.yml

@@ -32,7 +32,8 @@
           {%- endfor -%}
           {{ ns.out }}
       community.vmware.vmware_guest:
-        cluster: "{{ hypervisor_cfg.cluster }}"
+        cluster: "{{ hypervisor_cfg.cluster if (hypervisor_cfg.node | default('') | length == 0) else omit }}"
+        esxi_hostname: "{{ hypervisor_cfg.node if (hypervisor_cfg.node | default('') | length > 0) else omit }}"
         folder: "{{ system_cfg.path if system_cfg.path | string | length > 0 else omit }}"
         name: "{{ hostname }}"
         # Generic guest ID — VMware auto-detects OS post-install

vars_example.yml

@@ -7,7 +7,7 @@ hypervisor:
   url: "pve01.example.com"
   username: "root@pam"
   password: "CHANGE_ME"
-  host: "pve01"
+  node: "pve01"
   storage: "local-lvm"
   datacenter: "dc01"
   cluster: "cluster01"