Run tests on different platforms
Run on Azure
VM can be deployed on Azure using images from vhd, shared image gallery or marketplace. If multiple types are specified, the first non-empty type is picked in the following order : vhd, shared image gallery and marketplace.
Use vhd
To run using vhd, add the following to runbook :
platform:
- type: azure
...
requirement:
...
azure:
...
vhd: "<VHD URL>"
hyperv_generation: <1 or 2>
The <VHD URL>
can either be a SAS url or a blob url. If it is a SAS url, the image is copied to the resource group: lisa_shared_resource
, storage
account: lisat{location}{subscription_id[last 8 digits]}
and container:
lisa-sas-copied
in the subscription used to run LISA, which could potentially
increase the runtime. The copied VHD has to be manually deleted by the user.
If the selected VM Size’s Hypervisor Generation is ‘2’, hyperv_generation parameter is necessary, and should be specified as 2.
Use marketplace image
To run using marketplace image, add the following to runbook:
platform:
- type: azure
...
requirement:
...
azure:
...
marketplace: "<Publisher> <Offer> <Sku> <Version>"
Use existing VMs
In addition to deploying a new Azure server and running tests every time, you can use a deployed resource group or pre-existing resource group. The execution time is much shorter than deploying a new VM, because it skips deploying VMs, and avoiding to installing prerequisites packages for some test cases.
If the pre-existing deployment is not created by LISA, the VM names may need to be specified in the runbook.
If there is no deployment to reuse, run with the variables to keep the environment after test passed. If there is an existing deployment, skip this step.
lisa -r ./microsoft/runbook/azure.yml <other required variables, like subscription id> -v keep_environment:always
Specify the resource group name, and deploy to false to reuse an environment. If the environment is deployed by above step, you can find the resource group name from the log.
lisa -r ./microsoft/runbook/azure.yml <other required variables, like subscription id> -v deploy:false -v resource_group_name:"<resource group name>
Set other Azure parameters
The other parameters, like location, vm size, can be specified during deployment.
platform:
- type: azure
...
azure:
virtual_network_resource_group: $(vnet_resource_group)
virtual_network_name: $(vnet_name)
subnet_prefix: $(subnet_name)
use_public_address: "<true or false>"
requirement:
...
ignored_capability:
- SerialConsole
- Isolated_Resource
azure:
...
location: "<one or multiple locations, split by comma>"
vm_size: "<vm size>"
maximize_capability: "<true or false>"
osdisk_size_in_gb: <disk size in gb>
virtual_network_resource_group. Specify if an existing virtual network should be used. If virtual_network_resource_group is not provided, a virtual network will be created in the default resource group. If virtual_network_resource_group is provided, an existing virtual network will be used.
virtual_network_name. Specify the desired virtual network name. If virtual_network_resource_group is not provided, a virtual network will be created and the resulting virtual network name will be <virtual_network_name>. If virtual_network_resource_group is provided, an existing virtual network, with the name equal to virtual_network_name, will be used.
subnet_prefix. Specify the desired subnet prefix. If virtual_network_resource_group is not provided, a virtual network and subnet will be created and the resulting subnets will look like <subnet_profile>0, <subnet_profile>1, and so on. If virtual_network_resource_group is provided, an existing virtual network and subnet, with the name equal to subnet_prefix, will be used.
use_public_address. True means to connect to the Azure VMs with their public IP addresses. False means to connect with the private IP addresses. If not provided, the connections will default to using the public IP addresses.
ignored_capability. Specify feature names which will be ignored in test requirement. You can find the feature name from its name method in source code. For example, IsolatedResource feature’s name defined in
lisa/features/isolated_resource.py
as below:@classmethod def name(cls) -> str: return FEATURE_NAME_ISOLATED_RESOURCE
Then, you can add
isolated_resource
toignored_capability
.location. Specify which locations is used to deploy VMs. It can be one or multiple locations. For example, westus3 or westus3,eastus. If multiple locations are specified, it means each environment deploys VMs in one of location. To test multiple locations together, the combinator is needed.
vm_size. Specify which vm_size is used to deploy.
maximize_capability. True means to ignore test requirement, and try best to run all test cases. Notice, there are some features are conflict by natural, so some test cases may not be picked up. This setting is useful to force run perf tests on not designed VM sizes.
osdisk_size_in_gb is used to specify the size of the OS disk. If the specified size is smaller than the default size, the default size will be used. For range of disk size refer
Run on Ready computers
If you have prepared a Linux computer for testing, please run LISA with
ready
runbook:
Get the IP address of your computer for testing.
Get the SSH public/private key pair which can access this computer.
Run LISA with parameters below:
lisa -r ./microsoft/runbook/ready.yml -v public_address:<public address> -v "user_name:<user name>" -v "admin_private_key_file:<private key file>"
The advantage is it’s not related to any infra. The shortage is that, some test cases won’t run in Ready platform, for example, test cases cannot get serial log from a VM directly.
ready
runbook also supports tests which require multiple computers (for
example, networking testing); and, it supports password authentication too.
Learn more from runbook reference.
For a comprehensive introduction to LISA supported test parameters and runbook schema, please read command-line reference and runbook reference.
Run on Linux and QEMU
You can run the tests on Linux machine that has QEMU and KVM installed.
Currently, only the CBL-Mariner distro is supported. But it should be fairly straightforward to extend support to other distros. Also, only the the tier 0 tests are currently supported.
For CBL-Mariner:
Acquire a VHDX image of CBL-Mariner.
For example, you can build your own by following the VHDX and VHD images build instructions.
Convert image from VHDX to qcow2:
qemu-img convert -f vhdx -O qcow2 "<vhdx file>" "<qcow2 file>"
Run LISA with the parameters below:
./lisa.sh -r ./microsoft/runbook/qemu/CBL-Mariner.yml -v "admin_private_key_file:<private key file>" -v "qcow2:<qcow2 file>"
Run on AWS
Linux VM can be deployed on AWS using Amazon Machine Image (AMI) that provides the information required to launch an instance. At current all AWS resources will be deployed to the same configured region.
Configure the credentials for AWS. The credentials could be configured in multiple ways. Please create access keys for an AWS Identity and Access Management(IAM) user by following the cli configuration quick start. If you have the AWS CLI, then you can run “aws configure” to set up the credentials.
Or you could add the following configurations to aws runbook:
platform: - type: aws ... aws: aws_access_key_id: $(aws_access_key_id) aws_secret_access_key: $(aws_secret_access_key) aws_default_region: $(location) requirement: ... aws: ... marketplace: "<ami_image_id>"
Run LISA with the parameters below:
./lisa.sh -r ./microsoft/runbook/aws.yml -v "admin_username:<username>" -v "admin_private_key_file:<private key file>"
Update the default user name for the AMI you use to launch the instance. For an Ubuntu AMI, the user name is ubuntu. Please refer to the general prerequisites for connecting to the instance.
Run on WSL
WSL is supported cross all platforms by the guest layer in a node. So, it can be run with Local, Ready, Azure, AWS, BareMetal, etc. It supports below functionalities:
Provisioning WSL from a clean environment, or reuse existing WSL environment.
Replace the default kernel.
Install distro by names.
Support kernel format as tar.xz, unzipped kernel, or a folder which contains a file starting with “vmlinux-“.
The WSL configurations is under platform section as below.
platform:
- type: ready
guest_enabled: true # Default is false. Make sure set it to true to enable WSL.
guests:
- type: wsl
reinstall: false # Default is false. Set to true to reinstall WSL every time.
distro: # distro name in Windows store. Default is Ubuntu.
kernel: # path to replaced kernel
debug_console: # true or false. Default is false. Set it to true to pop up console for debugging.
If it needs to copy kernel to the Windows host, you can use the file_uploader transformer to upload the kernel during the “environment_connected” phase.
transformer:
- type: file_uploader
phase: environment_connected
source: D:\temp
destination: \temp
files:
- linux-5.15.123.1-microsoft-standard-WSL2.tar.xz