Method 2: Deploying an HPC Cluster

Scenarios

If you want to use existing cloud servers to deploy an HPC cluster, perform operations in this section.

Select a master node and then compute nodes. These nodes must have the same CPU architecture and OS version and in the same VPC and security groups.

Notes

When you use VPC for the first time, the system automatically creates a VPC for you, including a default security group and NIC. You can also create a VPC by yourself. For more information about VPC, see Virtual Private Cloud User Guide.
You can use the default security group provided by the system or create a security group by yourself.
To allow secure access to cloud servers in the cluster, ensure that port 22 in the security group rule is open. For details, see Adding a Security Group Rule.
Scalable File Service (SFS) is used for shared file access spanning cloud servers. For more information about SFS, see Scalable File Service User Guide.

Procedure

Log in to the HPC-S² console.
Click in the upper left corner and select your region and project.
Click Deploy HPC Cluster.
The Deploy HPC Cluster page is displayed.
Select a region.
Cloud servers in different regions cannot communicate with each other over an intranet. For low network latency and quick resource access, select the nearest region.

Set a cluster name.

The name must comply with the naming rules in the following table.

**Table 1** Cluster naming rules
Item	Requirements	Example Value
Cluster name	Consists of 1 to 52 characters. Contains only letters, digits, hyphens (-), and periods (.). Cannot start or end with a hyphen (-) or period (.). Cannot contain consecutive hyphens (-) and periods (.) or the combination of them, for example, --, .., -., or .-.	hpc-001.p1

Select a cloud server as the master node.
The master node must:
- Be in the region selected in 2.
- Be in the Running state.
- Have an EIP bound.
- Not be included in other clusters under the current account.
Select security groups.
A security group controls the access to the cloud servers added to this group by defining access rules. You can select multiple security groups.
Select compute nodes.
The compute nodes must:
- Be in the region selected in 2.
- Be in the Running state.
- Use the same CPU architecture as the master node.
- Be in the same VPC and security groups as the master node.
- Use the same major OS version as the master node.
- Be no more than 50 in quantity.
Select a VPC.
This parameter is automatically set to the VPC of the master node.

Set the login mode.

Only Password is available. If any of the selected cloud servers is using key pair authentication, go to the corresponding service console to change it to password authentication. In addition, set a strong password to protect yourself from malicious attacks.

The initial password of user root will be used for authenticating Linux login, while that of user Administrator will be used for authenticating Windows login. The password must meet the following requirements.

**Table 2** Password complexity requirements
Item	Requirements	Example Value
Password	Consists of 8 to 26 characters. Contains at least three of the following character types: Uppercase letters Lowercase letters Digits Special characters: $!@%-_=+[]:./^,{}? Cannot contain the username or the username spelled backwards. Cannot contain more than two characters in the same sequence as they appear in the username. (This requirement applies only to Windows cloud servers.)	YNbUwp!dUc9MClnv NOTE: The example password is generated randomly. Do not copy it for use.

Select a file system.
Available file systems are filtered and displayed in the drop-down list based on the network settings. If no files systems are available, you can create one and add it to the VPC specified in 9. For details, see Create a File System.
Select a scheduler.
Currently, only the slurm and sge schedulers are available. The default one is slurm.
Confirm the configurations and click Deploy HPC Cluster.
After the order is submitted, the system returns back to the Clusters page, where you can see that the cluster is in the Deploying state. Wait until the cluster status changes to Running which indicates that the cluster is successfully deployed.