跳到主要内容

Bohrium CLI Frequently Asked Questions

When using the Bohrium CLI for the first time, you need to bind an AccessKey. Please ensure that the bound AccessKey is correct. Please visit Bohrium-AccessKey to generate an AccessKey (Recreating the AccessKey will cause the original one to become invalid.

ACCESS_KEY值

After obtaining the ACCESS_KEY value, configure the environment variables for ACCESS_KEY as follows:

Note: Please replace the XXXX after ACCESS_KEY with your own AccessKey.

ZSH

If you are using ZSH, use the following command to write the ACCESS_KEY into ~/.zshrc.

echo 'export ACCESS_KEY=XXXX' >> ~/.zshrc
source ~/.zshrc

BASH

If you are using ZSH, use the following command to write the ACCESS_KEY into ~/.bashrc.

echo 'export ACCESS_KEY=XXXX' >> ~/.bashrc
source ~/.bashrc

Windows

Open Command Prompt (CMD) and run the following commands (note that the setx command requires reopening the Command Prompt to take effect):

setx ACCESS_KEY XXXX

Alternatively, you can manually add these variables through the system settings.

2. What should I do if an error suddenly occurs while creating a dataset using the Bohrium CLI?

Since bohr supports resuming from breakpoints, if an interruption occurs due to network issues or other factors, you can resume by re-executing the same command. Then, follow the prompt and enter y to recover the previous files, allowing the process to resume from the breakpoint.

3. Can’t find the project group ID created with Bohrium CLI on the Bohrium platform?

The job group ID created with bohr job_group create is used to submit multiple jobs to the same job group. It is not the same as the job group ID on the Bohrium platform; this job group ID is only applicable when submitting jobs using the Bohrium CLI.。

If you want to submit all jobs to the same job group, first use bohr job_group create to create a job group ID. Then, write this job group ID into the job_group_id field in the job.json file used for submitting jobs. An example is as follows:

{
"job_group_id":0000
}

4. When submitting jobs using Bohrium CLI, how to set the maximum running time for jobs?

Add max_run_time in the --file -i json file, the unit is minutes. The job will be automatically terminated and the files will be recycled after exceeding the maximum running time.

{
"max_run_time":2
}

5. Why do jobs abnormally interrupt? How to auto-retry when submitting jobs via Bohrium CLI?

During the computing node operation, there is a low probability of encountering abnormal issues, such as machine reclaiming or physical machine failure.

In this case, your job will be interrupted. The job results will still be retrieved rieved and can be downloaded from the job details page.

If you want the system to auto-retry after a job is abnormally interrupted, you can make the following settings in job.json:

  • If your computing software supports resuming calculations, you can:

    • Set the command to support resuming calculations, For example, using if else statements to determine whether there is a checkpoint file, and then executing different commands for calculations.
  • If your computing software does not support resuming calculations.

    • You only need to add the variable max_reschedule_times in the --file -i json file, which represents the number of retries you can accept. After a task is interrupted, the system will perform a full recalculation of the task. To set the retry to 2 times, the example is as follows:

      {
      "max_reschedule_times":2
      }