Updated on 2025-05-21 GMT+08:00

Quick Start

  1. Log in to HPC.

    Visit the EIP associated with the master node of the cockpit to switch to the login page. Then, use admin/admin for login.

  2. Add a partition.

    On the partition management page, choose Cluster Management > Partition Management.

    The parameters are described as follows.

    Parameter

    Description

    Specifications

    Compute node flavor, for example, c6s.xlarge.2

    Mirroring

    ID of the private image of the compute nodes

    AZ

    AZ where compute nodes are located, for example, cn-southwest-2a

    VPC ID

    ID of the VPC where compute nodes are located. It is recommended that the VPC ID be the same as that of the master node.

    Subnet ID list

    ID of the subnet where compute nodes are located. It is recommended that the subnet ID be the same as that of the master node.

    Security group

    ID of the security group where compute nodes are located. It is recommended that the security group ID be the same as that of the master node.

    SFS Mount List

    Mount address of the SFS Turbo file system. For example, if xxx.sfsturbo.internal:/ is the sharing path, you need to mount xxx.sfsturbo.internal:/ to /mnt/sfs_turbo_1 and mount xxx.sfsturbo.internal:/home to /home.

  3. Create a compute cluster.

    Go to the Resource Management page and click Add Compute Node.

    The parameters are described as follows.

    Parameter

    Description

    Payment mode

    Billing mode of compute nodes, which can be Yearly/Monthly or Pay-per-use.

    Partitioning

    Partition where nodes are located

    Specifications

    Compute node flavor

    Quantity

    Number of compute nodes to create

    Availability zone

    (Optional) AZ where the compute nodes are located. If this parameter is not set, the AZ set for the partition will be used.

    Node Prefix

    (Optional) Compute node prefix. If this parameter is not set, the prefix set for the partition will be used.

    Image ID

    (Optional) Image used by compute nodes. If this parameter is not set, the image set for the partition will be used.

  4. Create a template.

    On the template management page, choose System Management > Job Template Management.

  5. Submit a job.

    On the service management page, submit a job.

    The parameters are described as follows.

    Parameter

    Description

    Job Template

    Choose a template from the drop-down list.

    Execution User

    User who executes tasks. By default, the current user is used.

    Environment variables

    Environment variables that need to be transferred to Slurm

    Job Command

    Task script

    partitioning

    Partition to which tasks are submitted

    Feature

    Features can be added to compute nodes. When tasks are submitted, they can be scheduled based on the features.

    Execution path

    Path where the script is executed. The path must be a mounted public path.

    Standard output

    (Optional) Name of the log file generated by the job

    Error output

    (Optional) Exception log file of a job

    Number of nodes

    Number of nodes required by a job

    Number of Tasks on a Single Node

    Number of tasks running on each node

    Number of vCPUs on a Single Node

    Number of vCPUs on each node

    Memory of a Single Node

    Single-node memory required by tasks

    After a job is submitted, you can check job logs in real time.

    After the job is complete, you can check the job list, statistics, and resource usage.