Introduction

To run in Tencent Cloud, a CVM must have a kernel supporting virtio drivers, including the block device driver virtio_blk and the NIC driver virtio_net. To ensure that a CVM created with a custom image can start up properly, please check whether your image support virtio drivers in the source server before importing the image. This document uses CentOS as an example to describe how to check if an image supports virtio drivers.

Directions

Step 1: Checking whether the kernel supports virtio drivers

Execute the following command to check whether the current kernel supports virtio drivers:

grep -i virtio /boot/config-$(uname -r)

A response similar to the following will be returned:

• If the value of CONFIG_VIRTIO_BLK and CONFIG_VIRTIO_NET is m in the response，please go to Step 2.
• If the value of CONFIG_VIRTIO_BLK and CONFIG_VIRTIO_NET is y in the response, which means the OS contains the virtio dirvers, you can import the custom image to Tencent Cloud. For details, see Import Images > Overview.
• If you cannot find CONFIG_VIRTIO_BLK and CONFIG_VIRTIO_NET in the response, it means that images with the OS cannot be imported to Tencent Cloud. Please download and compile kernel.

Step 2: Checking whether virtio drivers are in the temporary file system

If the value of the parameters is m in Step 1, you need to check whether initramfs or initrd contains the virtio drivers. Please execute the corresponding command according to the operating system:

• For CentOS 6/CentOS 7/CentOS 8/RedHat 6/RedHat 7:

  lsinitrd /boot/initramfs-$(uname -r).img | grep virtio

• For RedHat 5/CentOS 5:

  mkdir -p /tmp/initrd && cd /tmp/initrd
  zcat /boot/initrd-$(uname -r).img | cpio -idmv
  find . -name "virtio*"

• For Debian/Ubuntu:

  lsinitramfs /boot/initrd.img-$(uname -r) | grep virtio

If a result similar to the following is returned:

It means that initramfs contains the virtio_blk driver and virtio.ko, virtio_pci.ko, and virtio_ring.ko on which the driver depends. In this case, you can import the custom image to Tencent Cloud. For details, see Import Images > Overview.
If initramfs or initrd does not contain the virtio drivers, please go to Step 3.

Step 3: Reconfigure the temporary file system

If you find that initramfs or initrd does not contain the virtio drivers in [Step 2]](#CheckVirtioForInitramfs), you will need to reconfigure the temporary file system to make sure that initramfs or initrd contains the virtio drivers. Please execute the corresponding command according to the operating system:

• For CentOS 6/CentOS 7/RedHat 6/RedHat 7:

  cp /boot/initramfs-$(uname -r).img /boot/initramfs-$(uname -r).img.bak
  mkinitrd -f --with=virtio_blk --with=virtio_pci /boot/initramfs-$(uname -r).img $(uname -r)

• For RedHat 5/CentOS 5:

  cp /boot/initrd-$(uname -r).img /boot/initrd-$(uname -r).img.bak
  mkinitrd -f --with=virtio_blk --with=virtio_pci /boot/initrd-$(uname -r).img $(uname -r)

• For Debian/Ubuntu:

  echo -e "virtio_pci\nvirtio_blk" >> /etc/initramfs-tools/modules
  update-initramfs  -u

Appendix

Downloading and compiling the kernel

Downloading the kernel installation package

1. Execute the following command to install the components necessary for kernel compilation.

   yum install -y ncurses-devel gcc make wget

2. Execute the following command to view the current version of the kernel.

   uname -r

   A response similar to the following will be returned, indicating the current kernel version is 2.6.32-642.6.2.el6.x86_64.
   
3. Go to Linux Kernel Download Page to download the source code of the corresponding kernel version.
   For example, for the 2.6.32-642.6.2.el6.x86_64 version, you should download linux-2.6.32.tar.gz at https://mirrors.edge.kernel.org/pub/linux/kernel/v2.6/linux-2.6.32.tar.gz.
4. Execute the following command to switch directory.

   cd /usr/src/

5. Execute the following command to download the installation package.

   wget https://mirrors.edge.kernel.org/pub/linux/kernel/v2.6/linux-2.6.32.tar.gz

6. Execute the following command to decompress the installation package.

   tar -xzf linux-2.6.32.tar.gz

7. Execute the following command to make connection.

   ln -s linux-2.6.32 linux

8. Execute the following command to switch directory.

   cd /usr/src/linux

Compling the kernel

1. Execute the following commands to compile the kernel.

   make mrproper
   cp /boot/config-$(uname -r) ./.config
   make menuconfig

   Enter the “Linux Kernel vX.X.XX Configuration” interface as shown below:
   

   Note：

   If you are not taken to the "Linux Kernel vX.X.XX Configuration" interface, please go to [Step 18](# OptionalStep).
   “Linux Kernel vX.X.XX Configuration” interface:

   • Press “Tab” or “↑” “↓” to move the cursor.
   • Press “Enter” to select or execute the item selected by the cursor.
   • Press the space bar to select the item selected by the cursor. “*” means compiling to the kernel, and "M" means compiling to a module.

2. Press the "↓" key to move the cursor to "Virtualization" and press the space bar to select "Virtualization".
3. Press "Enter" to enter the Virtualization details interface.
4. In the Virtualization details interface, check whether the Kernel-based Virtual Machine (KVM) support option is selected as shown below:
   
   If it is not selected, press the space bar to select the “Kernel-based Virtual Machine (KVM) support” option.
5. Press "Esc" to return to the "Linux Kernel vX.X.XX Configuration" main interface.
6. Press the "↓" key to move the cursor to "Processor type and features" and press "Enter" to enter the Processor type and features details interface.
7. Press the "↓" key to move the cursor to "Paravirtualized guest support" and press "Enter" to enter the detailed interface of Paravirtualized guest support.
8. In the Paravirtualized guest support details interface, check whether "KVM paravirtualized clock" and "KVM Guest support" are selected as shown below:
   
   If they are not selected, press the space bar to select the "KVM paravirtualized clock" and "KVM Guest support" options.
9. Press "Esc" to return to the "Linux Kernel vX.X.XX Configuration" main interface.
10. Press the "↓" key to move the cursor to "Device Drivers" and press "Enter" to enter the Device Drivers details interface.
11. Press the "↓" key to move the cursor to "Block devices" and press "Enter" to enter the Block devices details interface.
12. In the Block devices details interface, check whether "Virtio block driver (EXPERIMENTAL)" is selected as shown below:
    
    If it is not selected, press the space bar to select the "Virtio block driver (EXPERIMENTAL)" option.
13. Press “Esc” to return to the Device Drivers details interface.
14. Press the "↓" key to move the cursor to "Network device support" and press "Enter" to enter the Network device support details interface.
15. In the Network device support details interface, check whether "Virtio network driver (EXPERIMENTAL)" is selected as shown below:
    
    If it is not selected, press the space bar to select the "Virtio network driver (EXPERIMENTAL)" option.
16. Press "Esc" to exit the kernel configuration interface, and select "YES" to save the .config file.
17. Take Step 1: Checking whether the kernel supports the virtio drivers to verify whether the virtio drivers have been configured correctly.
18. (Optional) Execute the following command to manually edit the .config file.

    Note：

    This step is recommended if any of the following two is true:

    • The kernel still contains no configuration information related to the virtio drivers after you finish checking.
    • When compiling the kernel, you can not enter the kernel configuration interface or save the .config file.

    make oldconfig
    make prepare
    make scripts
    make
    make install

19. Execute the following commands to check the installation of the virtio drivers.

    find /lib/modules/"$(uname -r)"/ -name "virtio.*" | grep -E "virtio.*"
    grep -E "virtio.*" < /lib/modules/"$(uname -r)"/modules.builtin

    If any of the commands returns a list of files such as virtio_blk,virtio_pci.virtio_console, it indicates that you have installed the virtio drivers correctly.