tiCrypt 2.18.1

Active Jobs

Prerequisites

User Role:

Sub-admin, Admin, Super-admin

Permission Requirements

• Slurm
  • View Slurm job information

For a detailed diagram of how a job flows through the dual Slurm architecture, see the Secure HPC Batch Processing article.

A Slurm active job is a job currently scheduled or running in Global Slurm. This view reads the platform scheduling layer, so it shows what was provisioned and for whom, never the job script or the data the job is working on. Those stay inside the project's enclave with Local Slurm.

Job failures propagate up to Global Slurm, so a failed job appears here even though its contents do not. See Slurm for the full job path.

Access Slurm Active Jobs

1. Go to the Management icon in the top left taskbar.
2. Navigate to the Slurm section.
3. Click Active Jobs in the left panel.
4. View all the active Slurm jobs in the list.

The table refreshes every five minutes. Rows of FAILED jobs are red and rows of RUNNING jobs are green.

The following table describes the columns of the Active Jobs table.

Column	Content
Name	The job name.
Job ID	The job ID in Global Slurm.
State	The job state, such as PENDING, RUNNING or FAILED.
User	The tiCrypt user the job runs for.
VM	The VM that submitted the job.
Partition	The partition the job runs in.
Submitted	When the job was submitted.
Start time	The start time Slurm reports for the job.
End time	The end time Slurm reports for the job.
Reason/Nodes	For a pending job, the reason it waits; for a running job, the nodes it runs on.
Time limit	The time limit of the job.
Time used	How long the job has run.
Time left	How long the job may still run.
Priority	The scheduling priority of the job.
Account	The Slurm account the job is charged to.
QOS	The quality of service of the job.
Max CPUs	The maximum number of CPUs requested.
Min memory	The minimum memory requested.
Num nodes	The number of nodes requested.

View a Slurm Job

Location: Management > Slurm > Active Jobs

1. Select the job you want to view.
2. Click the View button in the top right panel.
3. Review the Slurm job panel. Its header shows the job state.

The following table describes the content of the Slurm job panel.

Section	Content
Basic Info	Job ID, State with a description of the state, User and VM with links to their panels, Partition, Account, QOS, and Reason for a pending job or Nodes for a running one.
Timing	Submitted, Eligible Time, Start Time, End Time, Time Limit, Time Used and Time Left, for the values Slurm reports.
Resources	Max CPUs, Min Memory, Num Nodes and Priority, for the values Slurm reports.

View All Details of a Slurm Job

Location: Management > Slurm > Active Jobs

1. Select the job you want to view.
2. Click the Open full menu button in the top right panel.
3. Select View all details.
4. In the Job details (<job ID>) pop-up, click Copy to copy the details or Download to download them.
5. Click Close.

View Details of a Slurm Job in JSON Format

Location: Management > Slurm > Active Jobs

1. Select the job you want to view.
2. Click the Open full menu button in the top right panel.
3. Select View details (JSON).
4. In the Job details (<job ID>) pop-up, click Copy to copy the JSON or Download to download it.
5. Click Close.

Cancel Slurm Job(s)

bulk-action

Location: Management > Slurm > Active Jobs

1. Select the job(s) you want to cancel. Cancel is not offered when every selected job is COMPLETED.
2. Click the Cancel button in the top right panel.
3. In the Cancel <job | N jobs>? prompt, which reads "This will remove the job(s) from the queue.", click Yes.

Result:

A Cancelling task asks Global Slurm to stop the selected jobs, in batches of up to 1,000 jobs, and the table reloads when the task ends.

Related

• Past Jobs: the accounting record of a job once it has ended.
• Worker VMs: the VMs Global Slurm started for running jobs.
• Partition Assignments: which VMs may submit jobs to each partition.