Description

Ansible playbooks to migrate a single root partition scheme or volume to DISA STIG/CIS approved LVM layout.

Requirements

A known good backup if running against a production system.

A second disk attached to your machine/instance. Examples:

• Physical disk
• Virtual hard disk
• AWS EBS volume

Tested Operating Systems

• CentOS 6.10
• CentOS 7
• RHEL 6.10
• RHEL 7.6
• Debian 9.9
• Ubuntu 16.04
• Ubuntu 18.04

Migration Process

Typical secondary disk notations:

• AWS EC2: /dev/xvdb
• AWS EC2 with NMVe: /dev/nvme1n1
• VMware ESXi: /dev/sdb

Step 1 - Create LVM layout:

ansible-playbook -i inventory --extra-vars='new_disk=/dev/sdb' lvm2.yml

Step 2 - Rsync current disk to new disk:

ansible-playbook -i inventory sync.yml

Step 3 - Update fstab:

sudo ansible-playbook -i inventory --extra-vars='new_disk=/dev/sdb' fstab.yml

Step 4 - Update grub config:

sudo ansible-playbook -i inventory --extra-vars='new_disk=/dev/sdb' grub.yml

Step 5 - Power off instance and detach original volume.

Step 6 - Attach secondary volume as primary and power on.

Variable Override Examples

• Check vars/main.yml for additional override options

AWS EC2:

ansible-playbook -i inventory --extra-vars='new_disk=/dev/xvdb' lvm2.yml

AWS EC2 with NVMe:

ansible-playbook -i inventory --extra-vars='new_disk=/dev/nvme1n1' \
--extra-vars='boot_part=p1' --extra-vars='root_part=p2' lvm2.yml

CentOS/RHEL 6 - ext4 vs default xfs for root:

ansible-playbook -i inventory --extra-vars='new_disk=/dev/sdb' --extra-vars='root_fs=ext4' lvm2.yml

RHEL 8 - xfs vs default ext4 for boot:

ansible-playbook -i inventory --extra-vars='new_disk=/dev/sdb' --extra-vars='boot_fs=xfs' lvm2.yml

Skip swap created on LVM:

sudo ansible-playbook -i inventory --extra-vars='new_disk=/dev/sdb' \
--extra-vars='create_swap=false' grub.yml

CIS disk layout to add /var/tmp:

ansible-playbook -i inventory --extra-vars='new_disk=/dev/sdb' --extra-vars='cis_layout=true' lvm2.yml

Gotchas

• On AWS EC2, when reattaching the secondary EBS volume as the primary volume, use /dev/sda1.
• On VMware vCenter, when removing both disks DO NOT select "Delete files from datastore". Then Add New Device > Existing Hard Disk. Choose YOUR_VM_1.vmdk as SCSI 0:0.
• CentOS/RHEL 6 only updates grub/initrd for default kernel.
• RHEL 8 gather_facts is noticeably slower, in my testing, for ansible chroot operations (Steps 3 and 4).

License and Author

• Author:: Nick Adams