curl -s $BLF_ICL_ENDPOINT/health | python3 -m json.tool

{
  "status": "ok",
  "chain_connected": true,
  "mongo_connected": true
}

# Check if your node is in the active pool
curl -s "$BLF_ICL_ENDPOINT/v1/nodes/active" | python3 -m json.tool | grep -A5 "your_address"

# Check node metrics
curl -s "$BLF_ICL_ENDPOINT/v1/nodes/{your_address}" | python3 -m json.tool

{
  "operator_address": "0xdDef3Cf5A4d0A6404Bc084D74de3E2c0d6147dA5",
  "model_tiers": [0],
  "active": true,
  "metrics": {
    "tasks_completed": 42,
    "tasks_accepted": 40,
    "tasks_rejected": 2,
    "reputation_score": 9520,
    "total_slash_amount": 0,
    "last_heartbeat": 1779078687
  }
}

# Tail logs in real-time
tail -f ~/.blindference/node.log

# Or if running in foreground, logs go to stdout
grep -E "ERROR|WARNING" ~/.blindference/node.log

# Count heartbeats in last hour
grep "Heartbeat sent" node.log | wc -l
# Expected: ~60 (1 per minute)

# Count heartbeat failures
grep "Heartbeat failed" node.log | wc -l
# Expected: 0

# Total jobs received
grep "Received.*assignment" node.log | wc -l

# Jobs by role
grep "starting as leader" node.log | wc -l
grep "starting as verifier" node.log | wc -l

# Job outcomes
grep "Leader result accepted" node.log | wc -l
grep "Verdict: match" node.log | wc -l
grep "Verdict: no-match" node.log | wc -l

# Average inference time (if logged at DEBUG level)
grep "Inference complete" node.log | awk '{print $NF}' | sort -n | awk '
  {sum+=$1; count++}
  END {print "Avg:", sum/count, "ms"}'

# Model usage distribution
grep "Running inference" node.log | sed 's/.*via //' | sort | uniq -c

# Decrypt successes/failures
grep "Decrypted prompt key" node.log | wc -l
grep "CoFHE decrypt failed" node.log | wc -l

# IPFS operations
grep "Downloaded prompt blob" node.log | wc -l
grep "uploaded encrypted output" node.log | wc -l

#!/bin/bash
# check_node_health.sh

HEALTH=$(curl -s -o /dev/null -w "%{http_code}" $BLF_ICL_ENDPOINT/health)
if [ "$HEALTH" != "200" ]; then
  echo "ALERT: ICL health check failed with status $HEALTH" | mail -s "Blindference Alert" admin@example.com
fi

# Check if node is in active pool
ACTIVE=$(curl -s "$BLF_ICL_ENDPOINT/v1/nodes/active" | grep -c "your_address")
if [ "$ACTIVE" -eq 0 ]; then
  echo "ALERT: Node not in active pool" | mail -s "Blindference Alert" admin@example.com
fi

*/5 * * * * /path/to/check_node_health.sh

# Add to node_loop.py
from prometheus_client import Counter, Histogram, start_http_server

jobs_completed = Counter('blindference_jobs_completed', 'Jobs completed', ['role'])
inference_duration = Histogram('blindference_inference_seconds', 'Inference duration')
heartbeat_failures = Counter('blindference_heartbeat_failures', 'Heartbeat failures')

# Start metrics server
start_http_server(9090)

[YYYY-MM-DDTHH:MM:SS] [LEVEL   ] [address] Message

[2026-05-18T10:25:00] [INFO   ] [0xdDef3Cf5] Daemon starting ...

# Extract all ERROR lines
grep "ERROR" node.log | jq -R 'split("]") | {time: .[0], level: .[1], address: .[2], message: .[3]}'

# Count errors by type
grep "ERROR" node.log | sed 's/.*ERROR.*]//' | sort | uniq -c | sort -rn