Skip to content
LabPod LabPod LabPod Docs

Reboot handling

What happens on reboot

Section titled “What happens on reboot”

Workspace containers are not automatically restarted after a host reboot.

When LabPod starts after a reboot, the reconciler detects that the host boot ID has changed and:

  1. Marks every workspace that was running before the reboot as stopped.
  2. Records a host_rebooted event in the audit log.
  3. Users see their workspaces listed as stopped and can click Start to bring them back.

In-memory state is lost: Jupyter kernels, running training processes, and terminal sessions do not survive a reboot. However, any work saved to /work (the bind mount to ~/work) persists.

Post-reboot verification

Section titled “Post-reboot verification”
Terminal window
# 1. Confirm the service is running
sudo systemctl status labpod


# 2. Check the API is responding
curl -s http://127.0.0.1:24680/api/version


# 3. In the UI: verify that previously-running workspaces show "stopped"
#    with a host_rebooted audit entry.
Terminal window
# Run these as root
# 1. Confirm the service is running
systemctl status labpod


# 2. Check the API is responding
curl -s http://127.0.0.1:24680/api/version


# 3. In the UI: verify that previously-running workspaces show "stopped"
#    with a host_rebooted audit entry.

If Start fails after reboot

Section titled “If Start fails after reboot”

Check that rootless Podman is still configured correctly for the workspace owner:

Terminal window
# 1. Verify user linger is enabled
sudo loginctl show-user <username> | grep Linger
# Should show:  Linger=yes


# 2. List the user's containers
sudo runuser -u <username> -- podman ps -a --filter name=labpod-ws-<id>


# 3. Check container logs for the failed start
sudo runuser -u <username> -- podman logs labpod-ws-<id>
Terminal window
# Run these as root
# 1. Verify user linger is enabled
loginctl show-user <username> | grep Linger
# Should show:  Linger=yes


# 2. List the user's containers
runuser -u <username> -- podman ps -a --filter name=labpod-ws-<id>


# 3. Check container logs for the failed start
runuser -u <username> -- podman logs labpod-ws-<id>

Common post-reboot causes:

  • Linger not set - run sudo loginctl enable-linger <username> and try again.
  • Image was pruned - Podman couldn’t find the image; the workspace will fail until it’s pulled again. The error shows in the workspace detail page.
  • Port conflict - rare, since LabPod allocates ports from a managed pool.

Run sudo labpod admin doctor for a general health check after any reboot.