spark-status

name	spark-status
description	Check status of DGX Spark infrastructure (containers, Ray, vLLM)
allowed-tools	Bash(ssh *) Bash(curl *) Bash(source *) Read

Check DGX Spark Status

Prerequisites

Source the environment configuration:

source playbooks/dual-dgx-spark-setup/.env

Steps

1. Check containers:

   ssh -t $SPARK1_HOST "docker ps --filter name=$CONTAINER_NAME --format '{{.Names}} {{.Status}}'"
   ssh -t $SPARK2_HOST "docker ps --filter name=$CONTAINER_NAME --format '{{.Names}} {{.Status}}'"
   

2. Check Ray cluster (if container running):

   ssh -t $SPARK1_HOST "docker exec $CONTAINER_NAME ray status 2>/dev/null || echo 'Ray not running'"
   

3. Check vLLM server:

   curl -s http://$SPARK1_HOST:8000/health || echo "vLLM not responding"
   curl -s http://$SPARK1_HOST:8000/v1/models || echo "No models loaded"
   

4. Report status summary showing:

   • Container status on both Sparks
   • Ray cluster state (nodes, GPUs)
   • vLLM server health and loaded model (if any)