Start with a read-only inventory
Use lsblk -f, findmnt, df -h and df -i to identify devices, filesystems, mounts, capacity and inode usage. Never infer that a disk is empty from its device name, and do not begin with a formatting command.
Verify recovery first
Back up important data and partition information and test restoration. Determine whether the system uses partitions, LVM or a whole-device filesystem. A larger cloud disk does not automatically expand every layer inside the guest.
Change the appropriate layers
Follow the procedure for the actual device, partition or logical volume and filesystem. Shrinking has different support and safety constraints; it is not simply the reverse of expansion. Stop writes when unmounting or maintenance is required.
Verify persistence after restart
Check capacity, application reads and writes, and persistent mount configuration. Prefer stable identifiers and understand boot-time failure behavior. Remove old storage only after the application is using the new location and recovery has been verified.