Skip to content

[All platforms] Re-onboard does not clean up orphaned gateway container from interrupted Ctrl+C #1582

@zNeill

Description

@zNeill

Description

[Description]
Ctrl+C during gateway start leaves orphaned Docker container; re-onboard fails with port conflict instead of auto-cleanup.

[Environment]

Device: DGX Spark
Node.js: v22.22.2
npm: 10.9.7
Docker: Docker Engine 29.1.3
OpenShell CLI: 0.0.23
NemoClaw: 0.0.7
OpenClaw: 2026.3.11 (29dc654)

[Steps to Reproduce]

  1. Ensure no existing gateway
  2. Run: nemoclaw onboard
  3. Wait for "[2/8] Starting OpenShell gateway" and "Starting gateway cluster..."
  4. Press Ctrl+C after ~10-20 seconds (before sandbox image build)
  5. Verify orphaned container: docker ps -a | grep openshell
    → shows openshell-cluster-nemoclaw (Up, health: starting)
  6. Run: nemoclaw onboard

[Expected Behavior]

Step 6: Re-onboard automatically detects and cleans up the orphaned Docker
container from the failed gateway start. No manual cleanup required.

[Actual Behavior]
Step 6: Re-onboard fails at preflight check:

[1/8] Preflight checks
✓ Docker is running
✓ Container runtime: docker
✓ openshell CLI: openshell 0.0.23

!! Port 8080 is not available.
OpenShell gateway needs this port.

 Blocked by: docker-pr (PID 667159)

 To fix, stop the conflicting process:

   sudo kill 667159

Manual cleanup required:
docker stop openshell-cluster-nemoclaw
docker rm openshell-cluster-nemoclaw


[NVB# 6056049]

[NVB#6056049]

[NVB#6072048]

Metadata

Metadata

Assignees

Labels

NV QABugs found by the NVIDIA QA Teamneeds: triageAwaiting maintainer classification

Type

No fields configured for Bug.

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions