HPC-S² Application and Usage Restrictions

Updated on 2024-05-17 GMT+08:00

View PDF

HPC-S² allows you to create HPC clusters with a few clicks. During the cluster creation, the schedulers, domain controllers, and runtime dependency are automatically deployed.
You can manage clusters and nodes and view their status on the HPC-S² console.

The HPC-S² solution is available only in the AP-Singapore region.
Only Slurm and SGE schedulers can be used for scheduling.
The SGE scheduler can be in the Kunpeng or x86 architecture, but the Slurm scheduler can only be in the x86 architecture.
A single user can create a maximum of three clusters.
A cluster can contain a maximum of 2,000 compute nodes.
Data disks cannot be automatically attached to compute nodes. If needed, you can manually attach them for use.
If nodes fail to be created or added, the cluster cannot be automatically restored.
Only CentOS 7 series public images can be used during cluster creation or capacity expansion.
If the EIPs or data disks of the nodes are not deleted when the nodes are deleted, they will still incur fees.
Do not perform operations on the nodes in the HPC cluster on the ECS console. Otherwise, the cluster may encounter with errors.
Do not modify or delete files in the /usr/local/.hpccluster directory. Otherwise, the cluster may encounter errors.
The authorized_keys file in the /usr/local/.hpccluster/role/.ssh/ directory is used only for system maintenance.
The clusters created by users with different member accounts cannot have duplicated names if such users belong to the same master account.