如何在Ansible中自动识别并分配AWS服务器上的最大分区至/data目录

Great question—this is super common with AWS NVMe devices since their naming can shift depending on whether the instance uses a separate root volume or combines storage into a single device. The simplest way to dynamically find the largest block device (that meets your 1TB+ requirement) without hardcoding variables or relying on static inventory is to leverage Ansible's built-in facts and Jinja2 filters.

Step 1: Dynamically Identify the Target Device #

First, we'll use set_fact to parse Ansible's collected device facts and pick the largest non-removable block device with at least 1TB of capacity. Here's how:

- name: Gather device-specific facts (if not already collected)
  setup:
    gather_subset:
      - '!all'
      - '!min'
      - devices
    gather_timeout: 30

- name: Determine the largest block device (>=1TB)
  set_fact:
    target_device: >-
      {{ 
        ansible_devices | dict2items 
        | selectattr('value.removable', 'equalto', false) 
        | selectattr('value.size_total', '>=', 1024*1024*1024*1024) 
        | sort(attribute='value.size_total', reverse=True) 
        | first 
        | attr('key') 
      }}

- name: Fail if no qualifying device is found
  fail:
    msg: "No block device with 1TB+ capacity detected. Check instance storage configuration."
  when: target_device is not defined

Let's break down the filter chain:

• dict2items: Converts the ansible_devices dictionary (keys are device names, values are device details) into a list of key-value pairs for easier filtering.
• selectattr('value.removable', 'equalto', false): Filters out removable devices (like USB drives) to focus on fixed NVMe disks.
• selectattr('value.size_total', '>=', 1024*1024*1024*1024): Keeps only devices with 1TB or more of total capacity.
• sort(attribute='value.size_total', reverse=True): Sorts the remaining devices from largest to smallest.
• first | attr('key'): Grabs the top (largest) device and extracts its name (e.g., nvme0n1 or nvme1n1).

Step 2: Partition, Format, and Mount the Device #

Once we have target_device defined, use standard Ansible modules to set up the /data mount:

- name: Create a single partition covering the entire device
  parted:
    device: "/dev/{{ target_device }}"
    number: 1
    state: present
    part_start: 0%
    part_end: 100%
    label: gpt

- name: Format the partition as ext4
  filesystem:
    fstype: ext4
    dev: "/dev/{{ target_device }}1"
    force: no

- name: Ensure /data directory exists
  file:
    path: /data
    state: directory
    mode: '0755'

- name: Mount /data and add to fstab
  mount:
    path: /data
    src: "/dev/{{ target_device }}1"
    fstype: ext4
    state: mounted
    backup: yes

Key Notes #

• Facts Collection: If you already have full facts enabled (the default), you can skip the explicit setup task—Ansible will collect ansible_devices automatically.
• Edge Cases: If an instance has multiple 1TB+ devices, this picks the largest one. Adjust the sort logic or add additional filters (like checking device paths) if you need to target a specific device type.
• Idempotency: All these modules are idempotent, so rerunning the playbook won't break existing setups (it will skip tasks if the partition/mount already exists).

内容的提问来源于stack exchange，提问作者dsollen