Create an E-HPC Instant job - Elastic High Performance Computing

Create an E-HPC Instant job from the console to run batch computations or long-running services on managed cloud nodes.

Prerequisites

Before you begin, ensure that you have:

Parameter	Required	Description
Job Name	Yes	A name for the job. Must be 2–64 characters, start with a letter, and contain only letters, digits, and hyphens (-).
Area	Yes	The region where the job runs. Affects network latency and costs.
Scheduling Policy	Yes	The scheduling policy. Currently only Standard is supported. To use the Economy policy, submit a ticket to apply.
Network and Availability Zone	Yes	Select a VPC and a vSwitch. E-HPC Instant automatically attaches a network interface card (NIC) to the job execution node to enable communication with your NAS storage.
Security Group	Yes	Select a security group. E-HPC Instant automatically adds it to the job execution node.
Single-node Elastic IP Address	No	(Optional) If enabled, an Elastic IP address (EIP) is automatically requested and assigned to the node, allowing internet access. Enabling an EIP incurs additional fees. Enable this only when needed.
Tags	No	(Optional) One or more tags for resource searches and cost analysis. See Tags.

Parameter	Required	Description
Computing Environment	Yes	The environment to run the job. Options: Cloud Server or Container.
Application	Yes	The application image for the job.
Operation Mode	Yes	How the job node is released after execution. Batch Job: the node is automatically released after the execution command completes — use this for finite computations. Background Service: the node is released only after you manually delete the job — use this for long-running services.
Pre-processing Script	No	(Optional) A script command to run before the job starts.
Execution Command	Conditionally required	The command or script to run on the job node. Required when Operation Mode is Batch Job. For a Linux application, the command must start with `#!` followed by the interpreter path, for example: `#!/bin/bash`. For a Windows application, the first line must be `[powershell]` or `[bat]`.
Number Of Execution Nodes	Yes	The total number of nodes to run the job.

Parameter	Required	Description
Cores Per Node	Yes	The number of CPU cores for each execution node. The ratio of CPU cores to memory is fixed. See Limits for available combinations.
Memory Per Node	Yes	The memory size for each execution node. Determined by the Cores Per Node selection.
Storage Space Per Node	Yes	The local storage size per node. Must be larger than the application image. Minimum: 40 GiB for system applications, 20 GiB for personal applications.

Configure this section to mount a NAS file system to the job execution nodes.

Parameter	Description
Mount Directory	The local path on the node where the file system is mounted.
Type	The file system type. Currently only General-purpose NAS File System is supported.
File System and Mount Target	The NAS file system ID and mount target. Make sure the file system has at least one available mount target.
File System Directory	The directory within the file system to mount.
Mount Options	The mount protocol to use.

Parameter	Description
Logon Password / Confirm Password	The password for remote node login. Available only when Computing Environment is Cloud Server.
Resource Scheduling Limit	Controls node placement. None: nodes can be scheduled anywhere in the VPC. Only In The Specified vSwitch: nodes are placed in the selected vSwitch.

After configuring all parameters, click Create Job at the bottom of the page.

After creation, the job appears in the job list. The status progresses in this order:

Pending → Initializing → Running → Completed