Submit a job - Elastic High Performance Computing - Alibaba Cloud Documentation Center

Prerequisites

Before you begin, make sure that you have:

A cluster and cluster nodes in the Running state
A cluster user (see Manage users)
(Console only) A job file prepared on the cluster
(CLI only) A cluster scheduler that is Slurm or OpenPBS

Submit a job using the console

Go to the Cluster Details page.
1. Log on to the E-HPC console.
2. In the left part of the top navigation bar, select a region.
3. In the left-side navigation pane, click Cluster.
4. On the Cluster List page, find the cluster you want to manage and click its cluster ID.
In the left-side navigation pane, click Job Management.

Click Create Job and configure the following parameters.

Basic settings

Parameter	Required	Description
Job Name	Yes	A name for the job.
Scheduler Queue	Yes	The scheduler queue to run the job in.
Start Job Array	No	Enables the job array feature. Set Minimum Value (start index), Maximum Value (end index), and Step Size (increment between indexes; default: 1).
Job Priority	No	An integer greater than 0. A higher value means higher priority. Jobs with higher priority run first when the cluster scheduling policy is set to prioritize by job priority.
Run Command	Yes	The command or script to run. Use a relative path for a script file (e.g., `./job.pbs`), or prefix a text command with `--` (e.g., `--/opt/mpi/bin/mpirun /home/test/job.pbs`).
Nodes	No	Number of compute nodes to allocate for the job.
Requested CPUs	No	vCPUs per node. Allocate enough resources for the job — insufficient resources may cause the job to fail.
Requested Memory	No

Advanced settings

Parameter	Required	Description
Stdout Path	No	File path for stdout output. The cluster user must have write permission on this path. If left blank, the scheduler generates the output file automatically.
Stderr Path	No
Environment Variables	No	Runtime variables passed to the job, accessible via environment variables in the executable.

Click Confirm Create.

Submit a job using the CLI

Connect to the logon node of your cluster (see Connect to a cluster).
Write a job script and submit it. The commands differ by scheduler.

In auto scaling scenarios, E-HPC does not support memory-based scaling. Specify the number of required vCPUs when submitting a job.

Submit a job with OpenPBS

A PBS job script has three parts: the shell interpreter line, #PBS directives that set resource requirements and options, and the commands to run.

Create a job script file.

vim jobscript.pbs

The following is a sample jobscript.pbs. For the full list of PBS directives, see the OpenPBS documentation.

#!/bin/sh
# PBS -l ncpus=4,mem=1gb    # Computing resources: CPUs and memory
# PBS -l walltime=00:10:00  # Maximum run time (HH:MM:SS)
# PBS -o test_pbs.log       # stdout output file
# PBS -j oe                 # Merge stderr into the stdout file
cd $HOME
test.py -i test.data

Submit the job.
```
qsub jobscript.pbs
```

Submit a job with Slurm

A Slurm job script has three parts: the shell interpreter line, #SBATCH directives that set resource requirements and options, and the commands to run.

Create a job script file.

vim jobscript.slurm

The following is a sample jobscript.slurm. For the full list of Slurm directives, see the Slurm documentation.

#!/bin/sh
# SBATCH --job-name=slurm-quickstart  # Job name
# SBATCH --output=test_slurm.log      # stdout output file
# SBATCH --nodes=1                    # Number of nodes
# SBATCH --ntasks=1                   # Number of tasks
# SBATCH --cpus-per-task=1            # vCPUs per task
# SBATCH --time=00:10:00              # Maximum run time (HH:MM:SS)
# SBATCH --mem-per-cpu=1024           # Memory per vCPU (MB)
cd $HOME
test.py test.data

Submit the job.
```
sbatch jobscript.slurm
```

Submit a job using E-HPC Portal

For steps on submitting jobs through E-HPC Portal, see Submit jobs by using submitter.

What's next

Call the CreateJob API operation to submit jobs programmatically.