Docs

File Storage vePFS

FAQs

File System Management FAQ

This topic explains common issues related to vePFS file system management for your reference.

Category	Question
Creating file systems	Why does the file system initially use about 20 GiB of capacity after creation? What should I do if latency increases when creating files or folders in the vePFS root directory?
Expanding file systems	How do I expand or shrink a file system? What should I do if operations fail or get stuck while expanding the file system? What should I do if the capacity display is inconsistent during file system expansion?
Deleting file systems	The file system has been unbound from the mounting service, so why am I still unable to delete it? Deleting files with the rm command is slow. How can I speed up the deletion?

Why is there an initial capacity usage of about 20 GiB after the file system is created?

The initial capacity used after the file system is created includes file system metadata and the first batch of preallocated inodes, which are reserved for future file writes.

What should I do if there is increased latency when creating files or directories in the vePFS root directory?

vePFS is a strongly consistent parallel file system that uses distributed locks to ensure atomic and consistent access. The root directory is the entry point to the file system. Every node uses the root directory, so reasonable access to the root directory is critical for system stability, performance, and maintainability. If many nodes simultaneously create and delete subdirectories or files in the root directory at high frequency, it triggers contention for the root directory lock. This causes metadata operations to be serialized, significantly degrading performance.

To avoid high-frequency global lock contention in the root directory, plan subdirectories based on your specific business needs. Avoid frequently creating or deleting directories or files directly in the root directory. Additionally, as a special directory in vePFS, a Fileset has its own inode space and management capabilities. We recommend that subdirectories under the root directory be configured as Filesets to optimize file storage management within each Fileset.

If you have other requirements, submit a ticket to contact Volcano Engine for assistance.

How do I expand or reduce file system capacity?

vePFS provides storage services based on the configured capacity. You can manually expand the file system, but reducing its capacity is not supported.

If the current file system's capacity no longer meets your business needs, plan for expansion in advance and expand the file system before its usage reaches 90% to ensure stable performance and smooth business growth. For more information about expanding the file system, see Expanding a file system.

What should I do if I encounter an operation failure or get stuck during expansion?

When the file system reaches a high usage level, you may experience severe file fragmentation and slower disk access, resulting in reduced performance. To avoid these issues, plan your expansion before usage reaches 90%. If the expansion operation fails or gets stuck, it's likely due to high utilization. We recommend deleting some data first and uploading it again after a successful expansion. For detailed information about expansion, see Expanding a file system.

What should I do if the capacity display is inconsistent after expanding the file system?

After performing the expansion, the file system status shows "Expanding." When it changes to "Running," the expansion process is complete, and the console updates the file system capacity. However, the actual available storage in the file system has not yet been updated. Expanded capacity becomes effective only after disk balancing for expansion is completed.

Disk balancing after expansion is performed automatically by the system once the file system expansion process is finished. Depending on the expansion size and file system status, this process may take from several hours up to two days. Wait patiently during this period.

You can check the balancing progress in the vePFS Console. For more information, see Expand a File System.

You can log in to any management node of the file system and run the df -h command to view the actual storage capacity of the file system and confirm whether the expanded capacity has taken effect. Charges will apply once the expanded capacity becomes effective.

The file system has been unbound from the mounting service. Why can't it be deleted?

If you use the Machine Learning Platform (MLP), you must cancel the vePFS instance authorization for the MLP service in addition to unbinding the mounting service before deleting the file system. For more information about MLP operations, seeConfigure File Storage vePFS.

Deleting files with the rm command is slow. How can I improve the deletion speed?

You can use Data Insight to obtain a list of files to delete, then create a Data Flow Deletion task to batch delete files based on that list Data Flow Deletion tasks use concurrent deletion across multiple machines, delivering higher speed than executing the rm command on a single node You can also create multiple deletion tasks to split different subdirectories, increasing concurrency further

Last updated: 2026.05.19 14:09:10