added code library

Author: JaimeSalas

06-monitoring/00-prometheus/.solutions/batch/batch.js

@@ -1,16 +1,23 @@
+const { counter, gauge } = require('./custom-metrics');
+
 const getRandom = (start, end) => Math.floor((Math.random() * end) + start);
 
 const randomBatchProcess = () => {
   const jobs = getRandom(50, 500);
   const failed = getRandom(1, 50);
   const processed = jobs - failed;
-  return { jobs, failed, processed };
+  const active = getRandom(1, 100);
+  return { active, failed, processed };
 };
 
 module.exports.batch = () => {
   const intervalId = setInterval(() => {
-    const { jobs, failed, processed } = randomBatchProcess();
-    console.log(jobs, failed, processed);
+    const { active, failed, processed } = randomBatchProcess();
+
+    counter.labels('200').inc(processed);
+    counter.labels('500').inc(failed);
+
+    gauge.set(active);
   }, 5_000);
 
   return intervalId;

06-monitoring/00-prometheus/.solutions/batch/custom-metrics.js

@@ -0,0 +1,18 @@
+const promClient = require('prom-client');
+
+// Reference: https://github.com/siimon/prom-client/blob/master/example/counter.js
+const counter = new promClient.Counter({
+  name: 'worker_jobs_total',
+  help: 'Total jobs handle',
+  labelNames: ['code']
+});
+
+// Reference: https://github.com/siimon/prom-client/blob/master/example/gauge.js
+const gauge = new promClient.Gauge({
+  name: 'worker_jobs_active',
+  help: 'Worker jobs in process'
+});
+
+
+module.exports.counter = counter;
+module.exports.gauge = gauge;

06-monitoring/00-prometheus/.solutions/batch/index.js

@@ -1,13 +1,19 @@
 const express = require('express');
+const promClient = require('prom-client');
 const { port } = require('./config');
 const { batch } = require('./batch');
 
+const { register, collectDefaultMetrics } = promClient;
+
+collectDefaultMetrics({ prefix: 'default_' });
 const intervalId = batch();
 const app = express();
 
 app.get('/metrics', async (_, res) => {
   try {
-    res.send('Not implemented yet');
+    // res.send('Not implemented yet');
+    res.set('Content-Type', register.contentType);
+    res.end(await register.metrics());
   } catch (ex) {
     res.status(500).end(ex);
   }

06-monitoring/00-prometheus/.solutions/batch/package-lock.json

@@ -11,6 +11,7 @@
   "author": "",
   "license": "ISC",
   "dependencies": {
-    "express": "^4.17.2"
+    "express": "^4.17.2",
+    "prom-client": "^14.0.1"
   }
-}
+}

06-monitoring/00-prometheus/.solutions/batch/package.json

@@ -8,4 +8,148 @@ El ejemplo que vamos a crear se trata de un `batch process` ficticio que irá va
 
 ## 1. Instalamos el cliente de Prometheus
 
-En el caso de Nodejs el cliente de Prometheus es [prom-client](https://github.com/siimon/prom-client)
+En el caso de Nodejs el cliente de Prometheus es [prom-client](https://github.com/siimon/prom-client)
+
+```bash
+ npm i prom-client
+```
+
+## 2. Creamos nuestras custom metrics
+
+```js
+// Reference: https://github.com/siimon/prom-client/blob/master/example/counter.js
+const counter = new promClient.Counter({
+  name: 'worker_jobs_total',
+  help: 'Total jobs handle',
+  labelNames: ['code']
+});
+
+// Reference: https://github.com/siimon/prom-client/blob/master/example/gauge.js
+const gauge = new promClient.Gauge({
+  name: 'worker_jobs_active',
+  help: 'Worker jobs in process'
+});
+
+
+module.exports.counter = counter;
+module.exports.gauge = gauge;
+```
+
+## 3. Actualizamos la 'lógica' de procesos
+
+Actualizar `batch.js`
+
+```js
+/*diff*/
+const { counter, gauge } = require('./custom-metrics');
+/*diff*/
+
+const getRandom = (start, end) => Math.floor((Math.random() * end) + start);
+
+const randomBatchProcess = () => {
+  const jobs = getRandom(50, 500);
+  const failed = getRandom(1, 50);
+  const processed = jobs - failed;
+  const active = getRandom(1, 100);
+  return { active, failed, processed };
+};
+
+module.exports.batch = () => {
+  const intervalId = setInterval(() => {
+    const { active, failed, processed } = randomBatchProcess();
+    /*diff*/
+    counter.labels('200').inc(processed);
+    counter.labels('500').inc(failed);
+
+    gauge.set(active);
+    /*diff*/
+  }, 5_000);
+
+  return intervalId;
+};
+```
+
+## 4. Recuperamos las métricas
+
+Lo primero que vamos a hacer es recuperar las métricas que expone por defecto `prom-client`
+
+Actualizamos `index.js`
+
+```diff
+const express = require('express');
++const promClient = require('prom-client');
+const { port } = require('./config');
+const { batch } = require('./batch');
++
++const { register, collectDefaultMetrics } = promClient;
++
++collectDefaultMetrics({ prefix: 'default_' });
+const intervalId = batch();
+const app = express();
+
+app.get('/metrics', async (_, res) => {
+  try {
+    res.send('Not implemented yet');
+  } catch (ex) {
+    res.status(500).end(ex);
+  }
+});
+
+app.post('/stop', (_, res) => {
+  clearInterval(intervalId);
+  res.send('batch process stopped');
+});
+
+app.listen(port, () => {
+  console.log(`Listening at ${port}`);
+});
+
+```
+
+Por último tenemos que indicar a `prom-client`, cuando debemos volcar las métricas:
+
+```diff
+app.get('/metrics', async (_, res) => {
+  try {
+-   res.send('Not implemented yet');
++   res.set('Content-Type', register.contentType);
++   res.end(await register.metrics());
+  } catch (ex) {
+    res.status(500).end(ex);
+  }
+});
+```
+
+Si ejecutamos nuestra aplicación ahora podremos ver como las métricas son expuestas en el formato `Prometheus`
+
+```bash
+npm start
+```
+
+Y desde otra terminal, podemos verificar la salida con `curl localhost:3000/metrics`
+
+
+```ini
+# HELP worker_jobs_total Total jobs handle
+# TYPE worker_jobs_total counter
+worker_jobs_total{code="200"} 739
+worker_jobs_total{code="500"} 44
+
+# HELP worker_jobs_active Worker jobs in process
+# TYPE worker_jobs_active gauge
+worker_jobs_active 52
+
+# HELP default_process_cpu_user_seconds_total Total user CPU time spent in seconds.
+# TYPE default_process_cpu_user_seconds_total counter
+default_process_cpu_user_seconds_total 0.185732
+
+# HELP default_process_cpu_system_seconds_total Total system CPU time spent in seconds.
+# TYPE default_process_cpu_system_seconds_total counter
+default_process_cpu_system_seconds_total 0.027831
+
+# HELP default_process_cpu_seconds_total Total user and system CPU time spent in seconds.
+# TYPE default_process_cpu_seconds_total counter
+default_process_cpu_seconds_total 0.213563
+
+#........ 
+```