Getting help
Look here first
Section titled “Look here first”Before reaching out to your admin, a few quick checks can often tell you what happened:
- Workspace card / detail page - the stop reason is shown directly:
OOM killed,Exited unexpectedly (code N),stopped by host restart. - Container logs - on the workspace detail page, click Load logs to see the last N lines of the container’s output. Increase the tail count to 500 or 1000 for more history.
- Browser console - if you see a blank page or a broken UI, open the browser DevTools (F12) and check the Console and Network tabs for errors.
For GPU issues: open the LabPod Terminal and run nvidia-smi to see what the GPU reports.
What to tell your admin
Section titled “What to tell your admin”If you do need help, give your admin:
- The workspace name or ID - visible on the workspace card and in the URL
(
/workspaces/<id>). - What you did - which button you clicked, what command you ran.
- What you expected vs. what happened - paste error messages exactly; a screenshot is often faster than a description.
- Where you saw the error - in the browser UI, in a launcher (e.g. Jupyter kernel died),
in the terminal, or from
nvidia-smi.
The admin has access to audit logs (every start/stop/delete action with timestamps) and per-workspace container logs - they can correlate from your workspace ID.
Common self-service fixes
Section titled “Common self-service fixes”| Problem | Fix |
|---|---|
| Forgot LabPod password | Ask the admin to run labpod admin set-password <username> as root, or with sudo where appropriate |
Start reports image_not_present | Confirm the Pull prompt on the workspace form, or build a private template from Templates → My templates, then start again. Ask your admin to prepare a global template’s shared image |
| Launcher won’t start | Click Refresh in Apps, then try Start again; check container logs |
Bus error in DataLoader | You probably started the container yourself; add --shm-size to your podman run command |
| CUDA OOM | Reduce batch size or request more VRAM (whole GPU or larger shared fraction) |
/work full | Delete old checkpoints; prune unused images from the admin if you can’t free enough |
Changing your own password
Section titled “Changing your own password”Go to Settings (your username in the top navigation) → Change password. See Quotas & usage for details.
Personal access tokens
Section titled “Personal access tokens”If you drive your own workspaces from a terminal or a script, create a personal access token
instead of typing your password each time. Create one with labpod token create --name <name>
(the token is shown only once — copy it), then review or revoke your tokens in the browser under
Settings → API tokens. See CLI → Personal access tokens
for full usage.