@uc-engg/mycroft

Monitoring Client for NodeJS

• Easy to create store where user can measure custom metrics.
• Default metrics of node can be measured
• Default labels can be set for all metrics
• Supported custom metrics : Counter, Gauge and Histogram

Functions

• createStore :

  • Used to create store which is further used to register custom metrics
  • Params
    • store name
    • defaultLabels : labels that are added to every metrics
    • isCollectNodeMetrics : to add node server metrics to given store

• registerMetric :

  • Used to register metric with given store
  • Params : store name for given metric and metric properties

• exportMetrics :

  • Used to get all the metrics in prometheus exposition format with contentType
  • Params : store name for which metrics are required

Usage

• Create store (with default labels and node metrics)

  Mycroft.createStore({
    storeName: 'store-name',
    defaultLabels: defaultLabels,
    isCollectNodeMetrics: true
  });
  

• Register custom counter metric for a give store

  Mycroft.registerMetric.counter('store_1', {  
    name: 'counter-metric', 
    labelNames: ['label_1'], 
    help: 'help message' 
  });
  

• Measure custom metric for a given store

  Mycroft.incMetric('store_1', 'counter_metric', {
    label_1: 'label_1'
  });
  

• Export metrics for a given store

  Mycroft.exportMetrics('store_1');
  

  returns

  {
    metrics : 'all metrics in prometheus exposition format'
    contentType: 'content type information'
  }
  
  

Instrumenting

Three types of metric are offered: Counter, Gauge and Histogram.

Counter

Use it to monitor metrics which gets incremented by only one and further analysis can be performed based on the value. Example => number of errors occurred for http service request

Mycroft.registerMetric.counter('store_1', {  
  name: 'http_request_error_metric', 
  labelNames: ['service', 'route', 'error_type'], 
  help: 'Count of errors' 
});

Mycroft.incMetric('store_1', 'http_request_error_metric', {
  service: 'service_1',
  route: '/functionCall',
  error_type: 'null pointer exception'
});

Gauge

Use it to monitor metrics which can vary at any time. Example => how much time it took to complete a request. If you want the granular details like 95% or 99% percentile use Histogram.

Mycroft.registerMetric.gauge('store_1', {  
  name: 'http_request_lag_metric', 
  labelNames: ['service', 'route'], 
  help: 'Time took to complete service in ms' 
});

Mycroft.setMetric('store_1', 'http_request_lag_duration_millisecond', {
  service: 'service_1',
  route: '/functionCall'
}, 1000);

Histogram

Use it to track the size and number of events in buckets. This allows for aggregative calculation of quantiles. Caution : Using histogram can make the cardinality very high as buckets are stored as one label in metric itself.

Mycroft.registerMetric.histogram('store_1', {  
  name: 'http_request_lag_metric', 
  labelNames: ['service', 'route'], 
  help: 'Time took to complete service in ms',
  buckets: [50, 100, 200, 400, 600, 1000, 5000, 10000, 30000, 60000] 
});

Mycroft.setMetric('store_1', 'http_request_lag_duration_millisecond', {
  service: 'service_1',
  route: '/functionCall'
}, 1000);