Why Is My Interactive Workspace Pending?

Jobs can stay in Pending when they exceed quotas or when requested resources are not available. This page explains how to check the reason in the browser and common causes.

Investigating Pending Jobs (in the Browser)

To see why your job is Pending:

  1. Go to Workload Manager > Workloads

  2. Select your job

  3. Click Show details (top-right)

  4. Choose the Event History tab

  5. Look at the latest entries for the Pending reason

Common Reasons and Fixes

Exceeding User Quota

Rule: For interactive jobs, each user has a limited quota. If you request more than this, the job will remain Pending.

Fix: For heavier GPU or CPU needs, submit a training job.

If you exceed your GPU quota, the Event History tab may show messages like:

Node-Pool 'l40s': NonPreemptibleOverQuota: Non-preemptible workload is over quota.
Workload requested 1 GPUs, but <project_name>-l40s quota is 0.17 GPUs, while 0 GPUs are already allocated
for non-preemptible pods. Use a preemptible workload to go over quota.

Changing Compute Resources

Symptom: Pending if you select CPU/GPU/Memory values outside your project’s quota.

Fix:

  • Verify your requests are within the project’s quota.

  • Avoid modifying resource values beyond limits during submission.

Quick Checklist

  1. Open Show details → Event History.

  2. Confirm your request is within quotas.

  3. Avoid modifying compute resources beyond your allowed quota.

  4. Resubmit once adjusted.