File size: 2,585 Bytes
fbf3c28 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 | #!/bin/bash
# /data/adaptai/projects/elizabeth/blueprint/scripts/verify_forge.sh
# Robust script to verify the operational status of the Nova GPU forge.
echo "🔍 [QUARTZ] Starting MLOps Verification Protocol"
echo "==================================================="
# Configuration
# Assuming local port forwarding is set up from your machine to the Vast.ai instance.
# Forward 20037 -> 8000 (for 7B model)
# Forward 20038 -> 8001 (for 14B model)
ENDPOINTS=( "http://localhost:20037" "http://localhost:20038" )
MODELS=( "Qwen/Qwen2-7B-Instruct" "Qwen/Qwen2-14B-Instruct" )
ALL_CHECKS_PASSED=true
# 1. Verify Model Endpoints
echo "1. 🧠 Verifying Nova Model Endpoints..."
for i in ${!ENDPOINTS[@]};
do
endpoint=${ENDPOINTS[$i]}
model_name=${MODELS[$i]}
echo " Pinging ${model_name} at ${endpoint}..."
# Check if the vLLM server is alive
response_code=$(curl -s -o /dev/null -w "%{{http_code}}" -m 5 "${endpoint}/v1/models")
if [ "$response_code" = "200" ]; then
echo " ✅ Port $(echo $endpoint | cut -d: -f3): vLLM Server ONLINE"
else
echo " ❌ Port $(echo $endpoint | cut -d: -f3): vLLM Server OFFLINE or ERROR (HTTP $response_code)"
ALL_CHECKS_PASSED=false
continue
fi
# 2. Perform a warm-up inference to check model loading and performance
echo " 🚀 Performing warm-up inference for ${model_name}..."
TEST_PROMPT="Hello, Nova. This is a connectivity and performance verification. Please respond with 'OPERATIONAL' and nothing else."
START_TIME=$(date +%s%N)
RESPONSE=$(curl -s -X POST "${endpoint}/v1/chat/completions"
-H "Content-Type: application/json"
-d @- <<EOF
{
"model": "${model_name}",
"messages": [{"role": "user", "content": "$TEST_PROMPT"}],
"max_tokens": 5,
"temperature": 0
}
EOF
)
END_TIME=$(date +%s%N)
LATENCY_MS=$(($((END_TIME - START_TIME))/1000000))
RESPONSE_TEXT=$(echo $RESPONSE | jq -r '.choices[0].message.content // "ERROR"')
if [[ "$RESPONSE_TEXT" == "OPERATIONAL" ]]; then
echo " ✅ Model Response: OK ('OPERATIONAL')"
echo " ✅ Latency: ${LATENCY_MS} ms"
else
echo " ❌ Model Response FAILED. Raw response:"
echo " ${RESPONSE}"
ALL_CHECKS_PASSED=false
fi
done
if [ "$ALL_CHECKS_PASSED" = true ]; then
echo "\n✅ [QUARTZ] All MLOps checks PASSED. Forge is OPERATIONAL."
else
echo "\n❌ [QUARTZ] Some MLOps checks FAILED. Review logs above."
fi
echo "==================================================="
|