Docs

Torch Object Storage

TOS Import（will be deprecated soon）

TOS Import tool FAQs

This article introduces common issues encountered when using the TOS Import tool.

How to configure flow control?

You need to set the total migration bandwidth rate and the total file count flow control for migration tasks in the task configuration file by configuring the task_bandwidth and task_file_mps parameters before creating a migration task. The default total migration bandwidth rate is 100MB/s, and the total file count flow control for migration tasks is 1000 files/s. For detailed information about the task configuration file, see configuration file description.

How to upgrade the TOS Import tool?

TOS Import does not support overwrite upgrades or update self-checks. You need to download the latest package from the Volcengine official website and extract it to a new directory.

Does the TOS Import distributed mode support changing the main node (server node) by modifying the server.yaml file's work_nodes parameter?

TOS Import does not support replacing the server node. The node where the tool is deployed serves as the server node.

tip

To ensure TOS Import can be deployed in various network configurations, it does not strictly require the server node addresses in work_nodes in distributed mode. When the server.log file displays the following warning, it means the tool has detected that the server node address is not found on any network interface. You should check if the data center network is accessible. If the network is confirmed accessible, you can ignore this warning.

What to do if creating a migration task with the same name fails?

TOS Import does not support creating migration tasks with duplicate names; task names must be unique. Depending on the creation scenario, choose one of the following methods according to your situation:

If you want to create a new migration task, rename the task name in the task configuration file.

If you want to retry a failed migration task, use the ./tos-import resume command to retry; the migration tool will resume the task from the checkpoint.

If you want to retry the migration task from the beginning and consider the previous migration report and list useless, use the ./tos-import delete command to delete the task and then create the migration task again.

After creating a migration task, if the progress shows transferring but no data is migrating, what should be done?

Go to the /tos-import/log directory and check the server.log file for any error messages. If there are errors, fix them according to the corresponding messages.

tip

If running in distributed mode, check whether the first address in work_nodes within server.yaml is your machine's external address and ensure all child nodes can access it. If not, update it to the correct address.

Last updated: 2025.07.16 12:04:18