Docs

Elastic Compute Service

Images

Changing boot mode from BIOS to UEFI for custom Linux images

This topic describes how to change the boot mode of the operating system from BIOS to UEFI in a cloud environment.

Applicable scenario

Some ECS instance types provided by the Volcano Engine platform only support the UEFI boot mode. You cannot select such instance types if the custom image you want to use is in BIOS mode. You can follow the instructions provided in this topic to change the boot mode of the target custom image to UEFI.

Prerequisites

You have imported a Linux image that uses BIOS boot mode into the Volcano Engine platform. For more information, see Importing custom images.

This operation may compromise data security. We recommend that you back up your data in advance to ensure data security by performing operations such as copying images.

Procedure

In the top navigation bar, select the region and project where the target custom image is located.

Click the "Create Instances" button in the "Actions" column to the right of the target image to create a temporary instance by using the image.

You can use the image to create an instance with the lowest-specification instance type, such as ecs.g2i.large.

Execute the following command in the temporary instance to check whether the system volume of the instance has free space for creating an efi partition.

parted /dev/vda print free

If the following output is displayed, the system volume has no free space or less than 1 GB of free space. In this case, you need to scale up the system volume by approximately 1 GB. For more information, see Scaling up volumes (Linux).

If the following output is displayed, the system volume has at least 1 GB of free space. In this case, skip this step.

Use the GPT fdisk tool to partition the free space.

Execute the gdisk /dev/vda command to add two partitions: BIOS boot partition and EFI System.

Enter n and follow the prompts to add the BIOS boot partition, and then press Enter.

warning

The value of first sector must be 34.

Enter n , and add an EFI System partition and press Enter.

tip

A partition size of 200 MB is required.

Enter w , and enter y to confirm and complete the partitioning.

Execute the following command to read the partition table and update partition information.

partprobe /dev/vda

Execute the following command to view information about the BIOS boot partition and EFI System partitions.

parted -l
lsblk -l

Sample output:

Mount the UEFI partition.

Execute the following command to format the ESP.

tip

If the command cannot be found during execution, execute the yum install dosfstools command to install the tool.

Replace the digit 3 in vda3 with the Partition number of the EFI partition in Step 6.2.

mkfs.vfat /dev/vda3

Sample output:

Execute the following command to mount the UEFI partition to the /boot/efi directory.

If the /boot/efi directory does not exist, execute the mkdir /boot/efi command to create it.

mount /dev/vda3 /boot/efi

Execute the following command to check whether the mount is successful.

lsblk

Sample output:

Write the mount information of the UEFI partition to the fstab file.

Execute the following command to view the UUID of the UEFI partition.

In this practice, the UUID of the /dev/vda3 partition is F6BA-62E7.

blkid

Sample output:

Execute the following command to edit the fstab file.

vim /etc/fstab

Press the i key to enter insert mode, and add the UUID of the UEFI partition to the file.

UUID=F6BA-62E7 /boot/efi vfat umask=0077 0 1

Press the Esc key to exit edit mode, enter :wq, and then press Enter to save the changes and exit the file.

Reinstall BootLoader

Execute the following command to update the software package list.

CentOS image

yum makecache

Debian/Ubuntu image

apt-get update

Execute the following command to install efibootmgr.

CentOS image

yum install efibootmgr

Debian/Ubuntu image

apt install efibootmgr

Execute the following command to install shim.

CentOS image

yum install shim

Debian/Ubuntu image

apt install shim

Execute the following command to install grub2-efi.

CentOS image

yum install grub2-efi-x64 && yum install grub2-efi-x64-modules

Debian/Ubuntu image

apt install --reinstall grub-common grub-efi-amd64

Execute the following command to reinstall grub with UEFI specified as the bootloader.

CentOS image

grub2-install --target=x86_64-efi --removable --efi-directory=/boot/efi --bootloader-id=GRUB

Debian/Ubuntu image

grub-install --target=x86_64-efi --removable --efi-directory=/boot/efi --bootloader-id=GRUB

Execute the following command to regenerate the grub2-config file.

warning

The grub2-config file that is regenerated overwrites the original configurations. We recommend that you back up data in advance.

CentOS image

grub2-mkconfig -o /boot/grub2/grub.cfg

Debian/Ubuntu image

grub-mkconfig -o /boot/grub/grub.cfg

Execute the following command to modify the configuration of the grub2-mkconfig file.

If you are using a Debian or Ubuntu image, skip this step.

sed -i 's/linux16/linuxefi/g' /boot/grub2/grub.cfg
sed -i 's/initrd16/initrdefi/g' /boot/grub2/grub.cfg

To ensure you can properly use the instance types provided by the Volcano Engine platform, follow the instructions provided in the Install the Virtio1.1 driver section to install the Virtio1.1 driver.

Create a custom image from the operating system of the temporary instance. For more information, see Creating custom image based on instance (Linux).

At this point, you obtain a custom image in UEFI boot mode. Subsequently, you can release cloud resources such as the temporary instance, volumes, and EIPs based on your needs. You can also use the new custom image to purchase instances or change the operating systems for existing instances. For more information, see Purchasing instances based on custom images and Changing operating systems.

You can also execute the following commands in an instance created by using this image to verify whether the modification is successful.

df -h
efibootmgr

Sample output:

If the following output is displayed, the new image uses UEFI boot mode.

FAQ

Why does the "Boot mode" value of a newly created image still show as BIOS when I check its information in the Image console?

Cause:

The boot mode of the image in the console inherits the boot mode (BIOS) of the temporary instance used to create the image. Therefore, the displayed boot mode of the new image in the console is BIOS. You need to change the boot mode displayed in the console to UEFI to ensure proper image usage.

Solution:

You can use the console or call the ModifyImageAttribute operation to change the boot mode of the target custom image to UEFI.

Why does the screen display garbled characters when I log in via VNC to an instance created from the new image?

Cause:

This issue may occur because the image lacks the Cirrus graphics card driver for the corresponding kernel version, leading to display anomalies when you log in via VNC.

Solution:

Execute the following command to check whether the Cirrus graphics card driver is installed in the operating system.

modinfo cirrus

If the driver is already installed, submit a ticket to obtain technical support.

If not installed, execute the following command to install the Cirrus graphics driver for the corresponding kernel version.

CentOS image

yum makecache
install linux-modules-extra-$(uname -r) -y

Debian/Ubuntu image

apt-get update
apt-get install linux-modules-extra-$(uname -r) -y

Last updated: 2026.06.18 17:49:52