CLDSRV-573: Catch prom cluster timeout

Fix crashes of primary because of prom-client 5s timeout.
Mostly to happen at startup when workers are not ready.
Should also fix error write EPIPE in workers by preventing
primary to crash.

(cherry picked from commit bdb4f23)

lib/utilities/monitoringHandler.js

@@ -24,7 +24,15 @@ async function routeHandler(req, res, cb) {
     if (req.method !== 'GET') {
         return cb(errors.BadRequest, []);
     }
-    const metrics = await aggregatorRegistry.clusterMetrics();
+    let metrics;
+    try {
+        // Catch timeout on IPC between worker and primary
+        // prom-client has a 5s hardcoded timeout
+        metrics = await aggregatorRegistry.clusterMetrics();
+    } catch (err) {
+        return cb(err, { message: err.toString() });
+    }
+
     const contentLen = Buffer.byteLength(metrics, 'utf8');
     res.writeHead(200, {
         'Content-Length': contentLen,
@@ -48,7 +56,11 @@ function monitoringHandler(clientIP, req, res, log) {
     function monitoringEndHandler(err, results) {
         writeResponse(res, err, results, error => {
             if (error) {
-                return log.end().warn('monitoring error', { err: error });
+                return log.end().warn('monitoring error', { err: {
+                    ...error,
+                    // For ArsenalError message is in description
+                    message: error.description || error.message,
+                } });
             }
             return log.end();
         });