ansible parted 自动创建分区并获取编号并动态启动

Question

我需要在需要添加文件系统的数百台服务器中自动创建分区。

这个想法是让 ansible 获取下一个分区的编号，并根据 parted 信息的输出开始。

---
- hosts: all
  become: true

  tasks:
    - name: get partition info
      parted:
         device: /dev/sda
         unit: GiB
      register: sda_info

    - name: create new partition
      parted:
        device: /dev/sda
        number: ???
        part_start: ???
        part_end: 3GiB
        state: present
      loop: '{{ sda_info.partitions }}

这是 sda_info 的输出

"sda_info": {
    "changed": false,
    "disk": {
        "dev": "/dev/sda",
        "logical_block": 512,
        "model": "VMware Virtual disk",
        "physical_block": 512,
        "size": 49.0,
        "table": "msdos",
        "unit": "gib"
    },
    "failed": false,
    "partitions": [
        {
            "begin": 0.0,
            "end": 0.49,
            "flags": [
                "boot"
            ],
            "fstype": "ext4",
            "name": "",
            "num": 1,
            "size": 0.49,
            "unit": "gib"
        },
        {
            "begin": 0.49,
            "end": 40.0,
            "flags": [
                "lvm"
            ],
            "fstype": "",
            "name": "",
            "num": 2,
            "size": 39.5,
            "unit": "gib"
        },
        {
            "begin": 40.0,
            "end": 46.0,
            "flags": [],
            "fstype": "",
            "name": "",
            "num": 3,
            "size": 6.0,
            "unit": "gib"
        }
    ],
    "script": "unit 'GiB' print"

我如何根据 sda_info 的输出计算数字和开始？

在手头的例子中，parted 应该是： item.num +1 = 4 和 part_start = 46 磁盘上最后一个分区结束的地方。

每个服务器都有不同的磁盘结构，所以我需要动态获取它。

我正在考虑使用 last 作为数字，例如：

'{{ (sda_info.partitions|last).num +1 }}'
'{{ (sda_info.partitions|last).end }}'

但是如何确保 sda_info.partitions 按顺序输出并且 last 实际上是最高数字而不仅仅是字典中的最后一项并最终破坏坏分区？

sda_info.partition字典和字典在python中没有排序。

Answer 1

你只需像这样添加排序：

- debug:
    msg: "maxnum: {{ maxnum | int + 1 }}, maxend: {{ maxend |float}}"
  vars:
    maxnum: "{{ ((sda_info.partitions|sort(attribute='num'))|last).num }}"
    maxend: "{{ ((sda_info.partitions|sort(attribute='end'))|last).end }}"

如果您确定 maxnum 和 maxend 在同一条记录中：您可以简化

- debug:
    msg: "maxnum: {{ lastsorted.num | int + 1 }}, maxend: {{ lastsorted.end |float}}"
  vars:
    lastsorted: "{{ sda_info.partitions|sort(attribute='num')|last }}"

Answer 2

请参阅下面的完整剧本示例。这个想法是把剧本分成两部分。在第一个块中获取磁盘信息并将其写入控制器的文件。在第二个块中创建分区。原因是为了让代码更安全、更健壮、更灵活。

• 第一部分是安全的。远程主机上没有任何变化。您可以在所有主机上运行一次并显示调试以检查详细信息。比如我们在4GB的闪存盘上测试

shell> parted -l
Model: Generic Flash Disk (scsi)
Disk /dev/sdb: 4089MB
Sector size (logical/physical): 512B/512B
Partition Table: msdos
Disk Flags: 

Number  Start  End     Size    Type     File system  Flags
 1      512B   34.1MB  34.1MB  primary  fat32        esp

设置变量

disk: sdb
part_size: 10
unit: MB

并运行剧本

shell> ansible-playbook pb.yml -e debug=true -e get_info=true

PLAY [localhost] *****************************************************************************

TASK [debug] *********************************************************************************
ok: [localhost] => 
  msg: |-
    disk: /dev/sdb
    part_unit: MB
    cache_update: False
    cache_path: /tmp/localhost-sdb-part_info.json

TASK [get partition info] ********************************************************************
ok: [localhost]

TASK [debug] *********************************************************************************
ok: [localhost] => 
  part_info:
    changed: false
    disk:
      dev: /dev/sdb
      logical_block: 512
      model: Generic Flash Disk
      physical_block: 512
      size: 4089.0
      table: msdos
      unit: mb
    failed: false
    partitions:
    - begin: 0.0
      end: 34.1
      flags:
      - esp
      fstype: fat32
      name: ''
      num: 1
      size: 34.1
      unit: mb
    script: unit 'MB' print

TASK [debug] *********************************************************************************
ok: [localhost] => 
  msg: |-
    part_size: 10
    last_number: 1
    last_end: 34.1
    next_number: 2
    next_part_start: 34.1MB
    next_part_end: 44.1MB

TASK [copy] **********************************************************************************
changed: [localhost]

TASK [set_fact] ******************************************************************************
skipping: [localhost]

TASK [debug] *********************************************************************************
skipping: [localhost]

TASK [debug] *********************************************************************************
skipping: [localhost]

TASK [create new partition] ******************************************************************
skipping: [localhost]

TASK [debug] *********************************************************************************
skipping: [localhost]

PLAY RECAP ***********************************************************************************
localhost: ok=5    changed=1    unreachable=0    failed=0    skipped=5    rescued=0    ignored=0

可以看到任务复制是变了.这意味着文件是写在控制器上的。看一看

shell> cat /tmp/localhost-sdb-part_info.json 
{"changed": false, "disk": {"dev": "/dev/sdb", "size": 4089.0, "unit": "mb", "table": "msdos", "model": "Generic Flash Disk", "logical_block": 512, "physical_block": 512}, "partitions": [{"num": 1, "begin": 0.0, "end": 34.1, "size": 34.1, "fstype": "fat32", "name": "", "flags": ["esp"], "unit": "mb"}], "script": "unit 'MB' print", "failed": false}

这复制参数force设置为错误的.这意味着该文件不会被替换。仅当目标不存在时才会传输文件。这就是你想让游戏幂等的原因。创建新分区后，您不想用分区信息替换缓存。但是，如果您出于某种原因想要更新缓存集 cache_update=true 并运行剧本

shell> ansible-playbook pb.yml -e debug=true -e get_info=true -e cache_update=true

• 第二部分将创建新分区。先以--check模式运行，查看调试输出

shell> ansible-playbook pb6.yml -e debug=true -e create_part=true -C

给出（删节）

TASK [debug] *********************************************************************************
ok: [localhost] => 
  msg: |-
    part_size: 10
    last_number: 1
    last_end: 34.1
    unit: MB
    next_number: 2
    next_part_start: 34.1MB
    next_part_end: 44.1MB

这是对的。我们要创建大小为 10MB 的分区号​​ 2。让我们创造它

shell> ansible-playbook pb.yml -e debug=true -e create_part=true

PLAY [localhost] ********************************************************************************

TASK [debug] ************************************************************************************
skipping: [localhost]

TASK [get partition info] ***********************************************************************
skipping: [localhost]

TASK [debug] ************************************************************************************
skipping: [localhost]

TASK [debug] ************************************************************************************
skipping: [localhost]

TASK [copy] *************************************************************************************
skipping: [localhost]

TASK [set_fact] *********************************************************************************
ok: [localhost]

TASK [debug] ************************************************************************************
ok: [localhost] => 
  part_info:
    changed: false
    disk:
      dev: /dev/sdb
      logical_block: 512
      model: Generic Flash Disk
      physical_block: 512
      size: 4089.0
      table: msdos
      unit: mb
    failed: false
    partitions:
    - begin: 0.0
      end: 34.1
      flags:
      - esp
      fstype: fat32
      name: ''
      num: 1
      size: 34.1
      unit: mb
    script: unit 'MB' print

TASK [debug] ************************************************************************************
ok: [localhost] => 
  msg: |-
    part_size: 10
    last_number: 1
    last_end: 34.1
    unit: MB
    next_number: 2
    next_part_start: 34.1MB
    next_part_end: 44.1MB

TASK [create new partition] *********************************************************************
changed: [localhost]

TASK [debug] ************************************************************************************
ok: [localhost] => 
  result:
    changed: true
    disk:
      dev: /dev/sdb
      logical_block: 512
      model: Generic Flash Disk
      physical_block: 512
      size: 4089.0
      table: msdos
      unit: mb
    failed: false
    partitions:
    - begin: 0.0
      end: 34.1
      flags:
      - esp
      fstype: fat32
      name: ''
      num: 1
      size: 34.1
      unit: mb
    - begin: 34.1
      end: 44.1
      flags: []
      fstype: ''
      name: ''
      num: 2
      size: 10.0
      unit: mb
    script: unit MB mkpart primary 34.1MB 44.1MB

PLAY RECAP **************************************************************************************
localhost: ok=5    changed=1    unreachable=0    failed=0    skipped=5    rescued=0    ignored=0

您可以看到新分区已创建。该剧本是幂等的。您可以重复运行它以确保分区存在。一切正常时，不会报告任何更改。

shell> parted -l
Model: Generic Flash Disk (scsi)
Disk /dev/sdb: 4089MB
Sector size (logical/physical): 512B/512B
Partition Table: msdos
Disk Flags: 

Number  Start   End     Size    Type     File system  Flags
 1      512B    34.1MB  34.1MB  primary  fat32        esp
 2      34.1MB  44.1MB  10.0MB  primary

• 当您第一次在大量远程主机上运行第二部分时，将远程主机分成较小的组可能是个好主意。创建分区后，您可以将其全部运行一次以进行审计。

---

完整剧本的示例

---
# All rights reserved (c) 2022, Vladimir Botka <vbotka@gmail.com>
# Simplified BSD License, https://opensource.org/licenses/BSD-2-Clause

- hosts: localhost
  gather_facts: false
  become: true

  vars:

    disk: sda          # change this
    part_size: 3       # change this
    unit: GiB          # change this
    cache_update: false
    cache_path: "/tmp/{{ inventory_hostname }}-{{ disk }}-part_info.json"
    # next variables are calculated from part_info
    last_number: "{{ part_info.partitions|map(attribute='num')|max }}"
    last_end: "{{ part_info.partitions|map(attribute='end')|max }}"
    next_number: "{{ last_number|int + 1 }}"
    next_part_start: "{{ last_end }}{{ unit }}"
    next_part_end: "{{ last_end|float + part_size|float }}{{ unit }}"

  tasks:

    - name: Get info and write part_info to cache_path
      block:
        - debug:
            msg: |-
              disk: /dev/{{ disk }}
              unit: {{ unit }}
              cache_update: {{ cache_update }}
              cache_path: {{ cache_path }}
          when: debug|d(false)|bool
        - name: get partition info
          parted:
             device: "/dev/{{ disk }}"
             unit: "{{ unit }}"
          register: part_info
        - debug:
            var: part_info
          when: debug|d(false)|bool
        - debug:
            msg: |-
              part_size: {{ part_size }}
              last_number: {{ last_number }}
              last_end: {{ last_end }}
              next_number: {{ next_number }}
              next_part_start: {{ next_part_start }}
              next_part_end: {{ next_part_end }}
          when: debug|d(false)|bool
        - copy:
            dest: "{{ cache_path }}"
            force: "{{ cache_update|bool }}"
            content: |
              {{ part_info|to_json }}
          delegate_to: localhost
      when: get_info|d(false)|bool

    - name: Read part_info from cache_path and create partition
      block:
        - set_fact:
            part_info: "{{ lookup('file', cache_path)|from_yaml }}"
        - debug:
            var: part_info
          when: debug|d(false)|bool
        - debug:
            msg: |-
              part_size: {{ part_size }}
              last_number: {{ last_number }}
              last_end: {{ last_end }}
              unit: {{ unit }}
              next_number: {{ next_number }}
              next_part_start: {{ next_part_start }}
              next_part_end: {{ next_part_end }}
          when: debug|d(false)|bool
        - name: create new partition
          parted:
            device: "/dev/{{ disk }}"
            number: "{{ next_number }}"
            part_start: "{{ next_part_start }}"
            part_end: "{{ next_part_end }}"
            align: optimal
            unit: "{{ unit }}"
            state: present
          register: result
        - debug:
            var: result
          when: debug|d(false)|bool
      when: create_part|d(false)|bool

Answer 3

这是我与unit: s（扇区为整数）一起使用的简单剧本，以避免由于浮动而导致的一些错误：

- name: get partition info
  community.general.parted:
    device: /dev/sda
    unit: s
  register: sda_info
- name: create new partition
  community.general.parted:
    device: /dev/sda
    label: gpt
    number: "{{ lastsorted.num | int + 1 }}"
    part_start: "{{ lastsorted.end | int + 1}}s"
    part_end: "100%"
    state: present
  vars:
    lastsorted: "{{ sda_info.partitions|sort(attribute='num')|last }}"

它使用所有可用空间在磁盘末端创建一个新分区。