I was working on using Ansible to format and mount persistent disks on Google Compute Engine (GCE). The question I had was how to determine if the disk needed to be formatted and mounted.
In this case the solution turned out to be simple since there are sym-links to /dev/sdbX in /dev/disk/by-id/. The latter folder contains the disk-name assigned to the disk when it’s attached to the GCE instance. The only caveat is that google- is prefixed to the name.
With that knowledge, the same steps provided in the Google Cloud Platform (GCP) documentation can be used: https://cloud.google.com/compute/docs/disks/add-persistent-disk. The only change is replacing the paths.
After testing those commands manually, I wrote a simple Ansible role to automate the process.
mount_disk/tasks/main.yml:
- name: Check if disk exists
shell: "file -sL /dev/disk/by-id/google-"
register: disk_exists
- name: Format disk
shell: "mkfs.ext4 -m 0 -F -E lazy_itable_init=0,lazy_journal_init=0,discard /dev/disk/by-id/google-"
when: "'UUID=' not in disk_exists.stdout"
- name: Create mount directory
file:
dest: "/mnt/disks/"
state: directory
owner: root
group: root
mode: 0755
- name: Mount drive
mount:
path: "/mnt/disks/"
src: "/dev/disk/by-id/google-"
fstype: ext4
opts: discard,defaults
state: mounted
mount_disk/vars/main.yml:
disk_name: disk-name
This role will only format and mount a single disk specified via the disk_name variable. I felt this was the safest way to approach things. In my use case I didn’t have many disks to attach to my instances.
The first step in this role is to check if the disk is formatted. This could be more robust if I handled the case where the disk was not attached.
Next, if the disk exists, we’ll format the disk if it does not contain UUID= which appears in file -sL when the disk is formatted.
Finally, the built-in Ansible file and mount modules can be used for creating a mount directory and then mounting the disk to that location.