Install Hybrid HPC–Azure. Do one of the following:
- To deploy LiCO in local cluster:
# config EPEL repo
dnf install -y https://dl.fedoraproject.org/pub/epel/epel-release-latest-9.noarch.rpm
# install
dnf install -y lico-core-cloudscheduling-azure- To deploy LiCO in container:
# config EPEL repo
dnf install -y https://dl.fedoraproject.org/pub/epel/epel-release-latest-9.noarch.rpm
# install dependency packages
dnf install -y openvpn easy-rsa sshpassModify the configuration file
/etc/lico/lico.ini.d/cloudscheduling.ini
change the following content to the IP address and subnet mask of the LiCO management node:
[CLOUDSCHEDULING] # local head node ip address/netmask # for example: # inet 10.241.57.123/24 brd 10.241.57.255 # HEAD_NODE_ADDRESS = "127.0.0.1/24" HEAD_NODE_ADDRESS = "10.241.57.123/24" # head node name # REMOTE_AGENT = "localhost" REMOTE_AGENT = "head"Adjust the download URLs for drivers and other components based on the actual situation to ensure they can be downloaded correctly:
[CLOUDSCHEDULING.AZURE.GPU_DRIVER] GPU_DRIVER_URL = "https://us.download.nvidia.com/XFree86/Linux-x86_64/550.90.07/NVIDIA-Linux-x86_64-550.90.07.run" [CLOUDSCHEDULING.AZURE.MONITOR_COMPONENTS] DCGM_URL = "https://developer.download.nvidia.cn/compute/cuda/repos/rhel9/x86_64/datacenter-gpu-manager-3.3.9-1-x86_64.rpm"
The LiCO management node shares /opt/lico/cloud:
echo "/opt/lico/cloud *(ro,sync,no_subtree_check,no_root_squash)" >> /etc/exports
exportfs -aModify the LiCO management node slurm configuration file
/etc/slurm/slurm.conf, and add the following content at the
end of the file:
include /opt/lico/cloud/azure/slurm.confConfigure the autoscaling function of Hybrid HPC
Modify the value in the file
/etc/slurm/slurm.confto the following content:ResumeProgram=/opt/lico/pub/slurm/resume_script.sh SuspendProgram=/opt/lico/pub/slurm/suspend_script.shCreate resume_script.sh, suspend_script.sh, auto_scaling.sh in
/opt/lico/pub/slurm# Create the directory if it is not existed mkdir -p /opt/lico/pub/slurm# resume_script.sh #!/bin/bash /opt/lico/pub/slurm/auto_scaling.sh $1 on# suspend_script.sh #!/bin/bash /opt/lico/pub/slurm/auto_scaling.sh $1 off# auto_scaling.sh #!/bin/bash power_type=$2 echo "`date` Power $power_type invoked $0 $1" >> /var/log/slurm/lico_power_save.log hosts=`scontrol show hostnames $1` for host in $hosts;do list+=\"$host\", done list=${list%?} echo "`date` start power $power_type: $hosts" >> /var/log/slurm/power_save.log api_key="input your api key here" login_ip="input your login ip here" curl -X POST -H "Content-Type: application/json" -H "Authorization: token $api_key" -d '{"vms":['$list']}' -k https://$login_ip/api/cloudscheduling/vm/autoscaling/$power_type/ echo "`date` end power $power_type: $hosts" >> /var/log/slurm/lico_power_save.logInput your api key and login ip of LiCO in auto_scaling.sh You can get the api key by clicking Admin→API Key after logging into LiCO web portal.
Modify the scripts and directory permissions: ```shell chown -R slurm:slurm /opt/lico/pub/slurm/
chmod 755 /opt/lico/pub/slurm/*.sh ```
Run the following command to restart the slurmctld service on the LiCO management node:
systemctl restart slurmctldCreate an Azure Authenticator:
Register an application Attention：Before registering the application, please check your Azure AD permissions and subscription permissions. For details,please refer to https://learn.microsoft.com/en-us/azure/active-directory/develop/howto-create-service-principal-portal
Sign in to your Azure Account through the Azure portal and select Microsoft Entra ID
Select App registrations and Click New registration
Name the application, for example example-app. Select a supported account type, which determines who can use the application. After setting the values, select Register.
After registration is complete, copy the Application (client) ID and Directory (tenant) ID and store.
Assign a role to the application
In the Azure portal, assign a role at the subscription scope, search for and select Subscriptions, or select Subscriptions on the Home page.
Select the particular subscription to assign the application to.
Select Access control (IAM), Select Add > Add role assignment to open the Add role assignment page.
In the Role tab，select Privileged administrator roles->Owner
In the Members tab，Select Assign access to-> User, group, or service principal and then select Select members. By default, Azure AD applications aren’t displayed in the available options. To find your application, search by name (for example, “example-app”) and select it from the returned list. Click the Select button.
In the Conditions tab，Select Allow user to assign all roles except privileged administrator roles Owner, UAA, RBAC or Allow user to assign all roles.
Then click the Review + assign button
Create a new application secret
Select Microsoft Entra ID
From App registrations in Azure AD, select your application.
Select Certificates & secrets
Select Client secrets -> New client secret.
Provide a description of the secret, and a duration. When done, select Add.
After saving the client secret, the value of the client secret is displayed. Copy this value because you won’t be able to retrieve the key later.Store this value in the same location with the tenant ID and application ID.
Run the following command to import the azure authentication information into LiCO:
# Import the application (client) ID, directory (tenant) ID and client password obtained in 6.
# Follow the prompts and import them in sequence
lico azure_secret importCreate Public IP address
