Fix: Use correct health check endpoints for LiteLLM and Letta

Root cause: Status API was using wrong health endpoints: - LiteLLM /health requires API key auth (returns 401) -> changed to /health/readiness (no auth) - Letta /api/health returns 404 -> changed to /v1/health/ (correct endpoint)
2026-07-02 05:21:08 +00:00
commit cc9c903d6d
3 changed files with 140 additions and 0 deletions
@@ -0,0 +1,9 @@
+FROM node:20-alpine
+RUN apk add --no-cache curl
+WORKDIR /app
+COPY package.json ./
+RUN npm install --production
+COPY server.js ./
+EXPOSE 3100
+HEALTHCHECK --interval=30s --timeout=10s --retries=3 CMD curl -sf http://localhost:3100/health || exit 1
+CMD ["node", "server.js"]
@@ -0,0 +1,13 @@
+{
+  "name": "apex-status-api",
+  "version": "1.0.0",
+  "description": "APEX OS Platform Health Monitor",
+  "main": "server.js",
+  "scripts": {
+    "start": "node server.js",
+    "test": "node test.js"
+  },
+  "dependencies": {
+    "express": "^4.18.0"
+  }
+}
@@ -0,0 +1,118 @@
+const express = require('express');
+const { execSync } = require('child_process');
+const app = express();
+const PORT = process.env.PORT || 3100;
+const START_TIME = Date.now();
+
+function exec(cmd) {
+  try {
+    return execSync(cmd, { timeout: 10000, encoding: 'utf-8' }).trim();
+  } catch (e) {
+    return null;
+  }
+}
+
+app.get('/health', (req, res) => {
+  try {
+    const containers = JSON.parse(exec('curl -s --unix-socket /var/run/docker.sock http://localhost/containers/json?all=true') || '[]');
+    const running = containers.filter(c => c.State === 'running').length;
+    const total = containers.length;
+    const unhealthy = containers.filter(c => c.Status && c.Status.includes('unhealthy')).length;
+    const status = unhealthy > 0 ? 'degraded' : (running === total ? 'healthy' : 'partial');
+    res.json({
+      status,
+      platform: 'APEX OS',
+      uptime_seconds: Math.floor((Date.now() - START_TIME) / 1000),
+      containers: { running, total, unhealthy },
+      timestamp: new Date().toISOString()
+    });
+  } catch (e) {
+    res.status(500).json({ status: 'error', error: e.message });
+  }
+});
+
+app.get('/containers', (req, res) => {
+  try {
+    const raw = exec('curl -s --unix-socket /var/run/docker.sock http://localhost/containers/json?all=true');
+    const containers = JSON.parse(raw || '[]').map(c => ({
+      name: (c.Names || ['/unknown'])[0].replace('/', ''),
+      image: c.Image,
+      state: c.State,
+      status: c.Status,
+      created: new Date(c.Created * 1000).toISOString(),
+      ports: (c.Ports || []).map(p => p.PublicPort ? `${p.PublicPort}:${p.PrivatePort}` : `${p.PrivatePort}`).filter(Boolean)
+    }));
+    res.json({ count: containers.length, containers });
+  } catch (e) {
+    res.status(500).json({ error: e.message });
+  }
+});
+
+app.get('/resources', (req, res) => {
+  try {
+    const memRaw = exec('free -m') || '';
+    const memLine = memRaw.split('\n').find(l => l.startsWith('Mem:'));
+    const memParts = memLine ? memLine.split(/\s+/) : [];
+    const diskRaw = exec('df -h /') || '';
+    const diskLine = diskRaw.split('\n')[1] || '';
+    const diskParts = diskLine.split(/\s+/);
+    const loadRaw = exec('cat /proc/loadavg') || '0 0 0';
+    const loadParts = loadRaw.split(' ');
+    res.json({
+      memory: {
+        total_mb: parseInt(memParts[1]) || 0,
+        used_mb: parseInt(memParts[2]) || 0,
+        available_mb: parseInt(memParts[6]) || 0,
+        usage_percent: memParts[1] ? Math.round((memParts[2] / memParts[1]) * 100) : 0
+      },
+      disk: {
+        total: diskParts[1] || 'unknown',
+        used: diskParts[2] || 'unknown',
+        available: diskParts[3] || 'unknown',
+        usage_percent: diskParts[4] || 'unknown'
+      },
+      load: {
+        avg_1m: parseFloat(loadParts[0]) || 0,
+        avg_5m: parseFloat(loadParts[1]) || 0,
+        avg_15m: parseFloat(loadParts[2]) || 0
+      },
+      timestamp: new Date().toISOString()
+    });
+  } catch (e) {
+    res.status(500).json({ error: e.message });
+  }
+});
+
+app.get('/services', async (req, res) => {
+  const services = [
+    { name: 'litellm', url: 'http://litellm:4000/health/readiness', critical: true },
+    { name: 'letta', url: 'http://letta:8283/v1/health/', critical: false },
+    { name: 'n8n', url: 'http://n8n:5678/healthz', critical: true },
+    { name: 'gitea', url: 'http://gitea:3000/api/v1/version', critical: true },
+    { name: 'grafana', url: 'http://grafana:3000/api/health', critical: false },
+    { name: 'ollama', url: 'http://ollama:11434/api/tags', critical: true }
+  ];
+  const results = [];
+  for (const svc of services) {
+    const start = Date.now();
+    const response = exec('curl -sf -m 5 ' + svc.url);
+    results.push({
+      name: svc.name,
+      status: response ? 'healthy' : 'unreachable',
+      response_ms: Date.now() - start,
+      critical: svc.critical
+    });
+  }
+  const healthy = results.filter(r => r.status === 'healthy').length;
+  res.json({
+    overall: healthy === results.length ? 'all_healthy' : 'degraded',
+    healthy_count: healthy,
+    total_count: results.length,
+    services: results,
+    timestamp: new Date().toISOString()
+  });
+});
+
+app.listen(PORT, '0.0.0.0', () => {
+  console.log('APEX Status API v1.1 - connectivity fix applied');
+});