mirror of https://github.com/cloud-hypervisor/cloud-hypervisor.git synced 2024-06-30 15:02:37 +00:00

Wei Liu 1a09381d1e docs: mention thread pool option for virtiofsd

Signed-off-by: Wei Liu <liuwe@microsoft.com>

2022-08-03 20:36:22 +00:00

3.0 KiB

Raw Blame History

How to use virtio-fs

In the context of virtualization, it is always convenient to be able to share a directory from the host with the guest.

virtio-fs, also known as vhost-user-fs is a virtual device defined by the VIRTIO specification which allows any VMM to perform filesystem sharing.

Pre-requisites

The daemon

This virtual device relies on the vhost-user protocol, which assumes the backend (device emulation) is handled by a dedicated process running on the host. This daemon is called virtiofsd and needs to be present on the host.

Build virtiofsd

git clone https://gitlab.com/virtio-fs/virtiofsd
pushd virtiofsd
cargo build --release
sudo setcap cap_sys_admin+epi target/release/virtiofsd

Create shared directory

mkdir /tmp/shared_dir

Run virtiofsd

./virtiofsd \
    --log-level debug \
    --socket-path=/tmp/virtiofs \
    --shared-dir=/tmp/shared_dir \
    --cache=never \
    --thread-pool-size=$N

The cache=never option is the default when using virtiofsd with Cloud Hypervisor. This prevents from using the host page cache, reducing the overall footprint on host memory. This increases the maximum density of virtual machines that can be launched on a single host.

The cache=always option will allow the host page cache to be used, which can result in better performance for the guest's workload at the cost of increasing the footprint on host memory.

The thread-pool-size option controls how many IO threads are spawned. For very fast storage like NVMe spawning enough worker threads is critical to getting an acceptable performance compared to native.

Kernel support

Modern Linux kernels (at least v5.10) have support for virtio-fs. Use of older kernels, with additional patches, are not supported.

Start the VM

Once the daemon is running, the option --fs from Cloud Hypervisor needs to be used.

Both direct kernel boot and EFI firmware can be used to boot a VM with virtio-fs, given that the cloud image contains a recent enough kernel.

Correct functioning of --fs requires --memory shared=on to facilitate interprocess memory sharing.

Assuming you have focal-server-cloudimg-amd64.raw and vmlinux on your system, here is the Cloud Hypervisor command you need to run:

./cloud-hypervisor \
    --cpus boot=1 \
    --memory size=1G,shared=on \
    --disk path=focal-server-cloudimg-amd64.raw \
    --kernel vmlinux \
    --cmdline "console=hvc0 root=/dev/vda1 rw" \
    --fs tag=myfs,socket=/tmp/virtiofs,num_queues=1,queue_size=512

Mount the shared directory

The last step is to mount the shared directory inside the guest, using the virtiofs filesystem type.

mkdir mount_dir
mount -t virtiofs myfs mount_dir/

The tag needs to be consistent with what has been provided through the Cloud Hypervisor command line, which happens to be myfs in this example.

DAX feature

Given the DAX feature is not stable yet from a daemon standpoint, it is not available in Cloud Hypervisor.

3.0 KiB Raw Blame History