Elastic High Performance Computing:Submit jobs using the submitter

Prerequisites

The cluster is in the Running state.

Before you begin

Before you submit a job, prepare the job file and upload it to your E-HPC cluster.

The E-HPC Portal supports the following methods for uploading job files:

• Use the Data Management feature: You can create and edit job files directly within a cluster folder, upload files from your local computer to the cluster, or download files from an OSS bucket to the cluster.

• Use the Remote Connection feature: Click the icon in the top-right corner to connect to the cluster. Then, you can run commands to create the job file.

Procedure

1. Log on to the E-HPC Portal.

   For more information, see Log on to the E-HPC Portal.

2. In the top navigation bar, choose Tasks.

3. At the top of the page, click submitter.

4. On the Create Job page, configure the following job parameters.

   Note

   If you need to submit the same job multiple times, you can click Save as Template after you configure the job information. This saves the current configuration as a template, which you can select from the left-side navigation pane for future use.

   • Basic parameters

     Parameter	Description
     Username	If you have a Slurm cluster and are logged in to the E-HPC Portal as the root user, you can choose to submit the job as root or as another specified user. Important If you specify a non-root user, ensure that user has previously logged on to the E-HPC Portal and submitted at least one job.
     Job name	The name of the job. If the job file is automatically downloaded and decompressed, the job name is also used as the name of the destination directory.
     Input file	The input file for the job. First, enter a command-line flag (for example, -input), and then select an input file (for example, /home/testuser/in.txt).
     Output file	The output file for the job. First, enter a command-line flag (for example, -output), and then enter a path for the output file (for example, /home/testuser/out.txt).
     Job queue	The job queue in which to run the job. If compute nodes are assigned to a specific job queue, you must submit the job to that queue. Otherwise, the job fails. If compute nodes are not assigned to any queue, the job is submitted to the scheduler's default queue.
     Run Command	The command submitted to the scheduler to run the job. The command can be a script file or a block of text. You can specify the command in one of the following ways: Online editing Local file Uploaded file Note If the script file is executable, you can specify a relative path, such as ./job.pbs. If the script file does not have execute permissions, you must specify an execution command, such as /opt/mpi/bin/mpirun /home/test/job.pbs.
     Priority	The priority of the job. Valid values range from 0 to 9. A larger value indicates a higher priority. If the cluster's scheduling policy is based on priority, jobs with higher priority are scheduled to run first.
     Number of nodes	The number of compute nodes to use for this job.
     Number of tasks	The number of tasks (processes) to run per compute node for this job.
     Number of threads	The number of threads for each task. If left empty, the default is 1 thread.
     Number of GPUs	The number of GPUs to use per compute node for this job. When you configure this parameter, make sure that the compute nodes are GPU-accelerated instances.

   • Advanced parameters

     Parameter	Description
     MPI performance profiling	Enable or disable MPI performance profiling.
     Maximum memory	The maximum memory per compute node for this job. If left empty, there is no memory limit. The format is Value+Unit, such as 1 GB or 200 MB.
     Maximum walltime	The maximum walltime for the job. If the job exceeds this time, it fails. If left empty, there is no time limit. For example, to specify 1 hour, enter 01:00:00.
     Log file	The path for saving job-related logs. Ensure that the specified path has write permissions. If left empty, the scheduler generates log files according to its default behavior.
     Environment variables	Add custom environment variables for the job.

5. Click Submit.

Next steps

After you submit the job, you can query its status on the Task Management page. For more information, see Query jobs.