Platform for AI (PAI) supports mounting CPFS for Lingjun in DLC, DSW, and EAS. Store data, code, and models once — access them everywhere with consistent paths. No copying or code changes needed.
Prerequisites
A CPFS for Lingjun file system. See Create a file system.
Additional requirements based on your resource type:
Resource type | Requirement |
Lingjun resources | CPFS availability zone and cluster ID must match your Lingjun GPU |
General computing resources |
For resource type details, see PAI computing resource overview.
Mount in DLC
When creating a DLC training job, configure CPFS storage in Environment Information. For complete job settings, see Create a training job.
For general computing resources, select the same VPC as your CPFS. If CPFS has no VPC mount target, add one first.
Choose Direct mount for quick experiments. Choose Dataset mount for team collaboration or frequent job creation.
Direct mount
Log in to the PAI console and open your workspace.
Go to Distributed Training (DLC) → Create Job.
In Environment Information → Storage Mount, click CPFS for Lingjun.
Configure mount parameters:
Parameter
Description
File System
Select your CPFS for Lingjun file system
Mount Target
Select VPC mount target. Required for general computing resources only — Lingjun resources skip this
File System Path
Directory to mount.
/for root, or specify subdirectoryMount Path
Path inside container (e.g.,
/mnt/data/)Read-only
Set read/write permissions
Complete other settings and submit the job.
Dataset mount
Datasets save mount configurations for reuse. Create a dataset first.
Log in to the PAI console and open your workspace.
Go to Distributed Training (DLC) → Create Job.
In Dataset Mount, click Custom Dataset.
Switch to the File Storage - CPFS for Lingjun tab, select your dataset, and click Confirm.
Verify the mount path. Set Read-only as needed.
Complete other settings and submit the job.
Mount in DSW
When creating a DSW instance, configure CPFS storage. For complete instance settings, see Create and manage DSW instances.
For general computing resources (including public resources), select the same VPC as your CPFS. If CPFS has no VPC mount target, add one first.
Choose Direct mount for quick experiments. Choose Dataset mount for team collaboration or frequent instance creation.
Direct mount
Log in to the PAI console and open your workspace.
Go to Interactive Modeling (DSW) → Create Instance.
In Environment Information → Storage Mount, click CPFS for Lingjun.
Configure mount parameters:
Parameter
Description
File System
Select your CPFS for Lingjun file system
Mount Target
Select VPC mount target. Required for general computing resources only
File System Path
Directory to mount.
/for root, or specify subdirectoryMount Path
Path inside container (e.g.,
/mnt/data/)Read-only
Set read/write permissions
Complete other settings and create the instance.
Dataset mount
Datasets save mount configurations for reuse. Create a dataset first.
Log in to the PAI console and open your workspace.
Go to Interactive Modeling (DSW) → Create Instance.
In Dataset Mount, click Custom Dataset.
Switch to the File Storage - CPFS for Lingjun tab, select your dataset, and click Confirm.
Complete other settings and create the instance.
Mount in EAS
When deploying an EAS service, configure CPFS storage in Environment Information. For complete deployment settings, see Deploy a model service.
EAS only supports direct mount. Dataset mount is not available.
For general computing resources, select the same VPC as your CPFS VPC mount target. If CPFS has no VPC mount target, add one first.
Log in to the PAI console and open your workspace.
Go to Model Online Service (EAS) → Inference Services tab.
Click Deploy Service → Custom Deployment.
In Environment Information → Storage Mount, click CPFS for Lingjun.
Configure mount parameters:
Parameter
Description
Storage Permission Resource Group
Select a resource group with CPFS access
File System
Select your CPFS for Lingjun file system
File System Path
Directory to mount.
/for root, or specify subdirectoryMount Path
Path inside container (e.g.,
/mnt/data/)Read-only
Set read/write permissions
Complete other settings and deploy the service.
Create a dataset
Datasets are mount configuration templates that store file system, directory path, and mount path settings. Create once, reuse across multiple DLC and DSW tasks.
EAS does not support dataset mount.
Log in to the PAI console and open your workspace.
Go to AI Asset Management → Datasets.
In the Custom Datasets tab, click Create Dataset and configure:
Section
Parameter
Description
Dataset
Name
Dataset name
Type
Select Basic
Storage Type
Select File Storage (CPFS for Lingjun)
Content Type
Select data type: Image, Text, Audio, Video, Table, or General
Version
Storage Permission Resource Group
Select a resource group with CPFS access
File System
Select your CPFS for Lingjun file system
File System Path
Directory in CPFS.
/for root, or subdirectory like/datasets/trainDefault Mount Path
Path inside container (e.g.,
/mnt/data/)Click OK.
Troubleshooting
Error: not support cross azone attach
CPFS availability zone or cluster ID doesn't match your Lingjun GPU. Select a CPFS file system in the same zone and cluster.
No VPC mount target options when creating dataset
The CPFS file system has no VPC mount target configured. Add a VPC mount target first.