fix: harden clinical assistant MCP connectivity

src/routes/clinicalAssistant.js

@@ -16,7 +16,8 @@ var logger = require('../utils/logger');
 
 router.use(authMiddleware);
 
-var MCP_URL = process.env.CLINICAL_ASSISTANT_MCP_URL || process.env.MCP_SERVER_URL || 'http://127.0.0.1:8100/mcp';
+var MCP_URLS = buildMcpUrls();
+var _lastGoodMcpUrl = MCP_URLS[0];
 var DEFAULT_BEHAVIOR = 'You are a concise pediatric clinical assistant. Use retrieved context only for factual claims. If the user input is a greeting or too vague, answer briefly and ask what they want to look up. Synthesize across sources, cite claims with numbered citations like [1], and list sources at the bottom by title/resource and page. Do not invent citations.';
 var GREETING_RE = /^(hi|hello|hey|yo|good\s+(morning|afternoon|evening)|thanks|thank you|ok|okay|sup)[\s.!?]*$/i;
 var TOPIC_SUGGESTIONS = {
@@ -70,7 +71,15 @@ router.get('/clinical-assistant/status', async function(req, res) {
     var imageModel = await getSetting('clinical_assistant.image_model', '');
     var searchLimit = clampInt(await getSetting('clinical_assistant.search_limit', '8'), 3, 20, 8);
     var contextChars = clampInt(await getSetting('clinical_assistant.context_chars', '1400'), 300, 4000, 1400);
-    res.json({ success: true, chatModel: chatModel, imageModel: imageModel, searchLimit: searchLimit, contextChars: contextChars, mcpUrl: MCP_URL.replace(/\/mcp$/, '/mcp') });
+    var mcpHealth = await getMcpHealth();
+    res.json({
+      success: true,
+      chatModel: chatModel,
+      imageModel: imageModel,
+      searchLimit: searchLimit,
+      contextChars: contextChars,
+      mcp: mcpHealth
+    });
   } catch (e) {
     res.status(500).json({ error: 'Request failed' });
   }
@@ -209,28 +218,104 @@ async function semanticSearch(query, opts) {
         context_chars: opts.contextChars
       }
     }
-  }, session.sessionId);
+  }, session.sessionId, session.mcpUrl);
   return search.result || search;
 }
 
-async function mcpRequest(payload, sessionId) {
+async function mcpRequest(payload, sessionId, preferredUrl) {
   var headers = {
     'Accept': 'application/json, text/event-stream',
     'Content-Type': 'application/json'
   };
   if (sessionId) headers['mcp-session-id'] = sessionId;
-  var resp = await axios.post(MCP_URL, payload, {
+  var resp = await postMcpWithRetry(payload, headers, preferredUrl);
-    headers: headers,
-    timeout: 90000,
-    responseType: 'text',
-    transformResponse: [function(data) { return data; }]
-  });
   var parsed = parseMcpResponse(resp.data);
   if (parsed.error) throw new Error(parsed.error.message || JSON.stringify(parsed.error));
   parsed.sessionId = resp.headers['mcp-session-id'] || sessionId || null;
+  parsed.mcpUrl = resp.config && resp.config.url ? resp.config.url : preferredUrl || _lastGoodMcpUrl;
   return parsed;
 }
 
+async function postMcpWithRetry(payload, headers, preferredUrl) {
+  var lastErr = null;
+  var urls = orderedMcpUrls(preferredUrl);
+  var isInitialize = payload && payload.method === 'initialize';
+  for (var round = 0; round < 2; round++) {
+    for (var i = 0; i < urls.length; i++) {
+      var url = urls[i];
+      try {
+        var resp = await axios.post(url, payload, {
+          headers: headers,
+          timeout: isInitialize ? 8000 : 90000,
+          responseType: 'text',
+          transformResponse: [function(data) { return data; }]
+        });
+        _lastGoodMcpUrl = url;
+        return resp;
+      } catch (e) {
+        lastErr = e;
+        if (!isTransientMcpError(e)) throw e;
+        console.warn('[clinical-assistant] MCP endpoint unavailable:', url, e.code || e.message);
+      }
+    }
+    await sleep(750 * (round + 1));
+  }
+  throw lastErr;
+}
+
+function orderedMcpUrls(preferredUrl) {
+  var candidates = [];
+  [preferredUrl, _lastGoodMcpUrl].concat(MCP_URLS).forEach(function(url) {
+    if (url && candidates.indexOf(url) === -1) candidates.push(url);
+  });
+  return candidates;
+}
+
+function isTransientMcpError(e) {
+  var code = e && (e.code || (e.cause && e.cause.code)) || '';
+  var status = e && e.response && e.response.status;
+  return code === 'ECONNREFUSED' || code === 'ECONNRESET' || code === 'ETIMEDOUT' || status === 502 || status === 503 || status === 504;
+}
+
+async function getMcpHealth() {
+  var results = [];
+  var urls = orderedMcpUrls();
+  for (var i = 0; i < urls.length; i++) {
+    var mcpUrl = urls[i];
+    var healthUrl = mcpUrl.replace(/\/mcp\/?$/, '/health/live');
+    try {
+      var resp = await axios.get(healthUrl, { timeout: 3000 });
+      results.push({ url: mcpUrl, healthUrl: healthUrl, ok: true, status: resp.status, data: resp.data });
+      _lastGoodMcpUrl = mcpUrl;
+      return { ok: true, activeUrl: mcpUrl, checked: results, candidates: MCP_URLS };
+    } catch (e) {
+      results.push({ url: mcpUrl, healthUrl: healthUrl, ok: false, error: e.code || e.message });
+    }
+  }
+  return { ok: false, activeUrl: _lastGoodMcpUrl || null, checked: results, candidates: MCP_URLS };
+}
+
+function buildMcpUrls() {
+  var raw = [];
+  if (process.env.CLINICAL_ASSISTANT_MCP_URLS) raw = raw.concat(process.env.CLINICAL_ASSISTANT_MCP_URLS.split(','));
+  raw.push(process.env.CLINICAL_ASSISTANT_MCP_URL || process.env.MCP_SERVER_URL || '');
+  raw.push('http://mcp:8000/mcp');
+  raw.push('http://mcp-server-mcp-1:8000/mcp');
+  raw.push('http://127.0.0.1:8100/mcp');
+  var out = [];
+  raw.forEach(function(url) {
+    url = String(url || '').trim();
+    if (!url) return;
+    if (!/\/mcp\/?$/.test(url)) url = url.replace(/\/$/, '') + '/mcp';
+    if (out.indexOf(url) === -1) out.push(url);
+  });
+  return out;
+}
+
+function sleep(ms) {
+  return new Promise(function(resolve) { setTimeout(resolve, ms); });
+}
+
 function parseMcpResponse(body) {
   var text = String(body || '').trim();
   if (!text) return {};