FAQ

Are there any restrictions on the use of file storage?

There are some limitations on file storage, such as file system capacity, maximum file length, protocol version, and so on. Please refer to the “Product Restrictions” section for detailed information.

What access protocols does file storage support?

Currently, new instances of Capacity-based file storage products support both NFS and SMB protocols, while new instances of Performance-based file storage products only support the NFS protocol.

Does file storage allow cross-regional access?

The network of the file storage product is based on VPC. On the UFS console, you can specify hosts under multiple VPCs in the same region as the file system instance for access. If you need to access the file system in other regions, you can connect the VPC where the host that needs to access the file system and the VPC where the file system mount point is located through the UDPN service. We strongly recommend that the file system and the host using the file system be located in the same region, as the UDPN service may incur extra charges and bring higher latency than accessing the same region.

Does file storage permit cross-project group access?

The network of the file storage product is based on VPC. On the UFS console, users are allowed to specify hosts under multiple VPCs in the project group where the file system instance is located for access, and VPCs not in the project group are not allowed to create mount points by default. You can enable access to the file system from the hosts in the VPC of the project group where the mount point of the file system is located and the VPC corresponding to the project group of the hosts that need to access the file system on the VPC page.

Linux NFS client kernel defect

The file storage service depends on the NFS client’s service, which has some defects in its implementation on Linux, which could lead to serious problems such as IO card stuck without response.

We strongly recommend that you upgrade the kernel version of your host to the recommended version above. If you are unable to upgrade and encounter an unresponsive scene, you need to restart the host to recover.

Linux SMB client kernel defect

The SMB client in some early versions of Linux has some defects in its implementation, which can cause serious problems such as IO card stuck without response.

We strongly recommend that you upgrade the kernel version of your host to the recommended version above. If you are unable to upgrade and encounter an unresponsive scene, you need to restart the host to recover.

Is there a limit to the number of host clients that can access a single file system?

The client host accesses through the mount point corresponding to the file system. Since the mount point limits the VPC allowed to access rather than the specific host IP, any number of hosts within the VPC can access as long as you reasonable plan the size of each subnet within the VPC. Although we do have certain restrictions on the number of mount points (please refer to the “Product Restrictions” section for details), there is no restriction on the number of hosts allowed to access the mount point.

Can physical cloud and hosted cloud hosts access file storage?

Physical cloud hosts can access file storage by choosing their VPC on the file storage console and setting up mount points. However, hosted cloud VPCs are not allowed to be connected on the file storage console by default. You need to contact the hosted cloud technical support to enable the VPC where the hosted cloud host is located and the VPC where the file system mount point is located to access.

How to change the concurrency of NFS requests?

The NFS client limits the number of simultaneous requests allowed. The default value for this in many Linux distributions is 2, which leads to a lower throughput that the client can achieve. For this reason, it is recommended to change this parameter to improve performance for a single client.

Step One, Install NFS Client

For detailed instructions on installing the NFS client, please refer to the “Mount File System” section in the “Operations Guide” chapter.

Step Two, Modify System Parameters

Run the following command with root permission in the command line (this operation only needs to be performed once after the first time of the NFS Client installation is complete, and does not need to be run multiple times):

echo "options sunrpc tcp_slot_table_entries=128" >> /etc/modprobe.d/sunrpc.conf

echo "options sunrpc tcp_max_slot_table_entries=128" >> /etc/modprobe.d/sunrpc.conf

Step Three, Mount File System

For detailed instructions on mounting the file system, please refer to the “Mount File System” section in the “Operations Guide” chapter.

Step Four, Modify System Parameters Again

Modify the relevant system parameters by executing the following command:

sysctl -w sunrpc.tcp_slot_table_entries=128

After completing this, unmount the file system and re-mount it. Then run the following command to view the system parameter values. If the return value is 128, it means the setting is successful.

cat /proc/sys/sunrpc/tcp_slot_table_entries

Can the storage type of file storage (Capacity-based, Performance-based) be changed?

No, at present, it is not supported to change the type of already created file storage. If you have similar requirements, you can create a new file system instance and migrate the data from the old file system instance.

Do I need to do any special operations before deleting a file system?

Yes, before a file system is deleted, please make sure that the data in it will no longer be used. File storage currently does not provide data recovery functions, so please be cautious with the deletion operation. Before you delete a file system, please make sure that all hosts that are using the file system have unmounted it, otherwise the remaining deleted file system instances on the host will affect the operation of the host.

How to solve the problem of Windows using file system, prompt for no permission?

• Problem cause: NFS protocol checks the identity through GID and UID, and the default GID and UID values of Windows are -2, so there will be no permission to operate the file system.

• Solution:

1. Open a Command Prompt window, type regedit to enter the registry editor page.
2. Select the following values: HKEY_LOCAL_MACHINE > SOFTWARE > Microsoft > ClientForNFS > CurrentVersion > Default.
3. Right click Default, select New > DWORD(32-bit) value and create AnonymousGID and AnonymousUID two registry items.

How to solve the problem of the client getting stuck when Windows uses a file system to create files\folders?

• Problem cause: By default, Windows clients will use Server Message Block (SMB) ports 139 and 445 to access mounted file systems. As the file system is mounted using the NFS protocol, requests sent to ports 139 and 445 will remain unresponsive, causing the client to get stuck.

• Solution:

  Scheme one: Use SMB protocol to mount file system.

  Scheme two: If you are sure not to use the SMB protocol, you can close the smb client service using the following steps.

  1. Click on the start menu button in the lower left corner, open Windows Management Tools.
  2. Double-click Services.
  3. Find Workstation service.
  4. Right click Properties, in the General tab, set the startup type to Disabled and the service status to Stopped. In the Recovery tab, set the action type for First failure, Second failure, and Subsequent failures to No action. After the modification is completed, click Apply and OK.
  5. After modification, there will be no more client crashes when using the file system. If you need to use the SMB protocol in the future, you can start the Workstation service again.

How to solve the problem of “Invalid device” error when renaming files\folders when using file system in Windows?

• Problem Cause: NFS protocol implementation issue.

• Solution: Due to an NFS protocol implementation issue, when the mounted file system is a subdirectory, the renaming operation will return an “Invalid device” error. Please mount the root directory of the file system for use.