prom-client
Advanced tools
The prom-client npm package is a client for the Prometheus monitoring tool, which allows you to collect and expose metrics from your Node.js applications. It supports various metric types, custom labels, and can be integrated with the default Node.js metrics and other systems.
Counter
Counters are a metric that only goes up (e.g., page views, total number of requests).
const { Counter } = require('prom-client');
const httpRequestsTotal = new Counter({
name: 'http_requests_total',
help: 'Total number of HTTP requests'
});
httpRequestsTotal.inc(); // Increment by 1
httpRequestsTotal.inc(5); // Increment by a given valueGauge
Gauges are a metric that can go up or down (e.g., memory usage, number of active users).
const { Gauge } = require('prom-client');
const memoryUsageGauge = new Gauge({
name: 'memory_usage_bytes',
help: 'Memory usage in bytes'
});
memoryUsageGauge.set(process.memoryUsage().heapUsed); // Set to current heap usedHistogram
Histograms track the size and number of events in buckets (e.g., request durations or response sizes).
const { Histogram } = require('prom-client');
const httpRequestDuration = new Histogram({
name: 'http_request_duration_seconds',
help: 'Duration of HTTP requests in seconds',
buckets: [0.1, 0.2, 0.5, 1, 5]
});
httpRequestDuration.observe(0.4); // Observe a value in secondsSummary
Summaries calculate percentiles of observed values (e.g., long-tail request durations).
const { Summary } = require('prom-client');
const httpRequestDurationSummary = new Summary({
name: 'http_request_duration_summary',
help: 'Summary of HTTP request durations',
percentiles: [0.5, 0.9, 0.99]
});
httpRequestDurationSummary.observe(0.3); // Observe a value in secondsDefault Metrics
Collect default metrics from Node.js runtime and system, such as CPU usage, event loop lag, etc.
const { collectDefaultMetrics } = require('prom-client');
collectDefaultMetrics();This package is similar to prom-client in that it provides a way to collect and expose metrics for monitoring purposes. However, it is not specifically tied to Prometheus and can be used with other monitoring tools.
This package is a Prometheus monitoring and metrics exporter for Node.js applications. It is similar to prom-client but is built on top of the appmetrics library, which provides additional insights into the Node.js runtime.
This package is an Express middleware that automatically records request durations and exposes them in Prometheus format. It is similar to prom-client but is more focused on Express applications and provides out-of-the-box middleware for easy integration.
A prometheus client for Node.js that supports histogram, summaries, gauges and counters.
See example folder for a sample usage. The library does not bundle any web
framework. To expose the metrics, respond to Prometheus's scrape requests with
the result of await registry.metrics().
cluster moduleNode.js's cluster module spawns multiple processes and hands off socket
connections to those workers. Returning metrics from a worker's local registry
will only reveal that individual worker's metrics, which is generally
undesirable. To solve this, you can aggregate all of the workers' metrics in the
master process. See example/cluster.js for an example.
Default metrics use sensible aggregation methods. (Note, however, that the event
loop lag mean and percentiles are averaged, which is not perfectly accurate.)
Custom metrics are summed across workers by default. To use a different
aggregation method, set the aggregator property in the metric config to one of
'sum', 'first', 'min', 'max', 'average' or 'omit'. (See lib/metrics/version.js
for an example.)
If you need to expose metrics about an individual worker, you can include a
value that is unique to the worker (such as the worker ID or process ID) in a
label. (See example/server.js for an example using
worker_${cluster.worker.id} as a label value.)
Metrics are aggregated from the global registry by default. To use a different
registry, call
client.AggregatorRegistry.setRegistries(registryOrArrayOfRegistries) from the
worker processes.
There are some default metrics recommended by Prometheus
itself.
To collect these, call collectDefaultMetrics. In addition, some
Node.js-specific metrics are included, such as event loop lag, active handles,
GC and Node.js version. See lib/metrics for a list of all
metrics.
NOTE: Some of the metrics, concerning File Descriptors and Memory, are only available on Linux.
collectDefaultMetrics optionally accepts a config object with following entries:
prefix an optional prefix for metric names. Default: no prefix.register to which registry the metrics should be registered. Default: the global default registry.gcDurationBuckets with custom buckets for GC duration histogram. Default buckets of GC duration histogram are [0.001, 0.01, 0.1, 1, 2, 5] (in seconds).eventLoopMonitoringPrecision with sampling rate in milliseconds. Must be greater than zero. Default: 10.To register metrics to another registry, pass it in as register:
const client = require('prom-client');
const collectDefaultMetrics = client.collectDefaultMetrics;
const Registry = client.Registry;
const register = new Registry();
collectDefaultMetrics({ register });
To use custom buckets for GC duration histogram, pass it in as gcDurationBuckets:
const client = require('prom-client');
const collectDefaultMetrics = client.collectDefaultMetrics;
collectDefaultMetrics({ gcDurationBuckets: [0.1, 0.2, 0.3] });
To prefix metric names with your own arbitrary string, pass in a prefix:
const client = require('prom-client');
const collectDefaultMetrics = client.collectDefaultMetrics;
const prefix = 'my_application_';
collectDefaultMetrics({ prefix });
To apply generic labels to all default metrics, pass an object to the labels property (useful if you're working in a clustered environment):
const client = require('prom-client');
const collectDefaultMetrics = client.collectDefaultMetrics;
collectDefaultMetrics({
labels: { NODE_APP_INSTANCE: process.env.NODE_APP_INSTANCE },
});
You can get the full list of metrics by inspecting
client.collectDefaultMetrics.metricsList.
Default metrics are collected on scrape of metrics endpoint, not on an interval.
const client = require('prom-client');
const collectDefaultMetrics = client.collectDefaultMetrics;
collectDefaultMetrics();
All metric types have two mandatory parameters: name and help. Refer to
https://prometheus.io/docs/practices/naming/ for guidance on naming metrics.
For metrics based on point-in-time observations (e.g. current memory usage, as
opposed to HTTP request durations observed continuously in a histogram), you
should provide a collect() function, which will be invoked when Prometheus
scrapes your metrics endpoint. collect() can either be synchronous or return a
promise. See Gauge below for an example. (Note that you should not update
metric values in a setInterval callback; do so in this collect function
instead.)
See Labels for information on how to configure labels for all metric types.
Counters go up, and reset when the process restarts.
const client = require('prom-client');
const counter = new client.Counter({
name: 'metric_name',
help: 'metric_help',
});
counter.inc(); // Increment by 1
counter.inc(10); // Increment by 10
Gauges are similar to Counters but a Gauge's value can be decreased.
const client = require('prom-client');
const gauge = new client.Gauge({ name: 'metric_name', help: 'metric_help' });
gauge.set(10); // Set to 10
gauge.inc(); // Increment 1
gauge.inc(10); // Increment 10
gauge.dec(); // Decrement by 1
gauge.dec(10); // Decrement by 10
If the gauge is used for a point-in-time observation, you should provide a
collect function:
const client = require('prom-client');
new client.Gauge({
name: 'metric_name',
help: 'metric_help',
collect() {
// Invoked when the registry collects its metrics' values.
// This can be synchronous or it can return a promise/be an async function.
this.set(/* the current value */);
},
});
// Async version:
const client = require('prom-client');
new client.Gauge({
name: 'metric_name',
help: 'metric_help',
async collect() {
// Invoked when the registry collects its metrics' values.
const currentValue = await somethingAsync();
this.set(currentValue);
},
});
Note that you should not use arrow functions for collect because arrow
functions will not have the correct value for this.
// Set value to current time in seconds:
gauge.setToCurrentTime();
// Record durations:
const end = gauge.startTimer();
http.get('url', res => {
end();
});
Histograms track sizes and frequency of events.
The defaults buckets are intended to cover usual web/RPC requests, but they can be overridden. (See also Bucket Generators.)
const client = require('prom-client');
new client.Histogram({
name: 'metric_name',
help: 'metric_help',
buckets: [0.1, 5, 15, 50, 100, 500],
});
const client = require('prom-client');
const histogram = new client.Histogram({
name: 'metric_name',
help: 'metric_help',
});
histogram.observe(10); // Observe value in histogram
const end = histogram.startTimer();
xhrRequest(function (err, res) {
const seconds = end(); // Observes and returns the value to xhrRequests duration in seconds
});
Summaries calculate percentiles of observed values.
The default percentiles are: 0.01, 0.05, 0.5, 0.9, 0.95, 0.99, 0.999. But they
can be overridden by specifying a percentiles array. (See also
Bucket Generators.)
const client = require('prom-client');
new client.Summary({
name: 'metric_name',
help: 'metric_help',
percentiles: [0.01, 0.1, 0.9, 0.99],
});
To enable the sliding window functionality for summaries you need to add
maxAgeSeconds and ageBuckets to the config like this:
const client = require('prom-client');
new client.Summary({
name: 'metric_name',
help: 'metric_help',
maxAgeSeconds: 600,
ageBuckets: 5,
pruneAgedBuckets: false,
});
The maxAgeSeconds will tell how old a bucket can be before it is reset and
ageBuckets configures how many buckets we will have in our sliding window for
the summary. If pruneAgedBuckets is false (default), the metric value will
always be present, even when empty (its percentile values will be 0). Set
pruneAgedBuckets to true if you don't want to export it when it is empty.
const client = require('prom-client');
const summary = new client.Summary({
name: 'metric_name',
help: 'metric_help',
});
summary.observe(10);
const end = summary.startTimer();
xhrRequest(function (err, res) {
end(); // Observes the value to xhrRequests duration in seconds
});
All metrics can take a labelNames property in the configuration object. All
label names that the metric support needs to be declared here. There are two
ways to add values to the labels:
const client = require('prom-client');
const gauge = new client.Gauge({
name: 'metric_name',
help: 'metric_help',
labelNames: ['method', 'statusCode'],
});
// 1st version: Set value to 100 with "method" set to "GET" and "statusCode" to "200"
gauge.set({ method: 'GET', statusCode: '200' }, 100);
// 2nd version: Same effect as above
gauge.labels({ method: 'GET', statusCode: '200' }).set(100);
// 3rd version: And again the same effect as above
gauge.labels('GET', '200').set(100);
It is also possible to use timers with labels, both before and after the timer is created:
const end = startTimer({ method: 'GET' }); // Set method to GET, we don't know statusCode yet
xhrRequest(function (err, res) {
if (err) {
end({ statusCode: '500' }); // Sets value to xhrRequest duration in seconds with statusCode 500
} else {
end({ statusCode: '200' }); // Sets value to xhrRequest duration in seconds with statusCode 200
}
});
Metrics with labels can not be exported before they have been observed at least once since the possible label values are not known before they're observed.
For histograms, this can be solved by explicitly zeroing all expected label values:
const histogram = new client.Histogram({
name: 'metric_name',
help: 'metric_help',
buckets: [0.1, 5, 15, 50, 100, 500],
labels: ['method'],
});
histogram.zero({ method: 'GET' });
histogram.zero({ method: 'POST' });
Typescript can also enforce label names using as const
import * as client from 'prom-client';
const counter = new client.Counter({
name: 'metric_name',
help: 'metric_help',
// add `as const` here to enforce label names
labelNames: ['method'] as const,
});
// Ok
counter.inc({ method: 1 });
// this is an error since `'methods'` is not a valid `labelName`
// @ts-expect-error
counter.inc({ methods: 1 });
Static labels may be applied to every metric emitted by a registry:
const client = require('prom-client');
const defaultLabels = { serviceName: 'api-v1' };
client.register.setDefaultLabels(defaultLabels);
This will output metrics in the following way:
# HELP process_resident_memory_bytes Resident memory size in bytes.
# TYPE process_resident_memory_bytes gauge
process_resident_memory_bytes{serviceName="api-v1"} 33853440 1498510040309
Default labels will be overridden if there is a name conflict.
register.clear() will clear default labels.
The exemplars defined in the OpenMetrics specification can be enabled on Counter
and Histogram metric types. The default metrics have support for OpenTelemetry,
they will populate the exemplars with the labels {traceId, spanId} and their
corresponding values.
The format for inc() and observe() calls are different if exemplars are
enabled. They get a single object with the format
{labels, value, exemplarLabels}.
When using exemplars, the registry used for metrics should be set to OpenMetrics type (including the global or default registry if no registries are specified).
The library supports both the old Prometheus format and the OpenMetrics format. The format can be set per registry. For default metrics:
const Prometheus = require('prom-client');
Prometheus.register.setContentType(
Prometheus.Registry.OPENMETRICS_CONTENT_TYPE,
);
Currently available registry types are defined by the content types:
PROMETHEUS_CONTENT_TYPE - version 0.0.4 of the original Prometheus metrics, this is currently the default registry type.
OPENMETRICS_CONTENT_TYPE - defaults to version 1.0.0 of the OpenMetrics standard.
The HTTP Content-Type string for each registry type is exposed both at module
level (prometheusContentType and openMetricsContentType) and as static
properties on the Registry object.
The contentType constant exposed by the module returns the default content
type when creating a new registry, currently defaults to Prometheus type.
By default, metrics are automatically registered to the global registry (located
at require('prom-client').register). You can prevent this by specifying
registers: [] in the metric constructor configuration.
Using non-global registries requires creating a Registry instance and passing it
inside registers in the metric configuration object. Alternatively you can
pass an empty registers array and register it manually.
Registry has a merge function that enables you to expose multiple registries
on the same endpoint. If the same metric name exists in both registries, an
error will be thrown.
Merging registries of different types is undefined. The user needs to make sure all used registries have the same type (Prometheus or OpenMetrics versions).
const client = require('prom-client');
const registry = new client.Registry();
const counter = new client.Counter({
name: 'metric_name',
help: 'metric_help',
registers: [registry], // specify a non-default registry
});
const histogram = new client.Histogram({
name: 'metric_name',
help: 'metric_help',
registers: [], // don't automatically register this metric
});
registry.registerMetric(histogram); // register metric manually
counter.inc();
const mergedRegistries = client.Registry.merge([registry, client.register]);
If you want to use multiple or non-default registries with the Node.js cluster
module, you will need to set the registry/registries to aggregate from:
const AggregatorRegistry = client.AggregatorRegistry;
AggregatorRegistry.setRegistries(registry);
// or for multiple registries:
AggregatorRegistry.setRegistries([registry1, registry2]);
You can get all metrics by running await register.metrics(), which will return
a string in the Prometheus exposition format.
If you need to output a single metric in the Prometheus exposition format, you
can use await register.getSingleMetricAsString(*name of metric*), which will
return a string for Prometheus to consume.
If you need to get a reference to a previously registered metric, you can use
register.getSingleMetric(*name of metric*).
You can remove all metrics by calling register.clear(). You can also remove a
single metric by calling register.removeSingleMetric(*name of metric*).
If you need to reset all metrics, you can use register.resetMetrics(). The
metrics will remain present in the register and can be used without the need to
instantiate them again, like you would need to do after register.clear().
You can get aggregated metrics for all workers in a Node.js cluster with
await register.clusterMetrics(). This method returns a promise that resolves
with a metrics string suitable for Prometheus to consume.
const metrics = await register.clusterMetrics();
// - or -
register
.clusterMetrics()
.then(metrics => {
/* ... */
})
.catch(err => {
/* ... */
});
It is possible to push metrics via a Pushgateway.
const client = require('prom-client');
let gateway = new client.Pushgateway('http://127.0.0.1:9091');
gateway.pushAdd({ jobName: 'test' })
.then(({resp, body}) => {
/* ... */
})
.catch(err => {
/* ... */
})); //Add metric and overwrite old ones
gateway.push({ jobName: 'test' })
.then(({resp, body}) => {
/* ... */
})
.catch(err => {
/* ... */
})); //Overwrite all metrics (use PUT)
gateway.delete({ jobName: 'test' })
.then(({resp, body}) => {
/* ... */
})
.catch(err => {
/* ... */
})); //Delete all metrics for jobName
//All gateway requests can have groupings on it
gateway.pushAdd({ jobName: 'test', groupings: { key: 'value' } })
.then(({resp, body}) => {
/* ... */
})
.catch(err => {
/* ... */
}));
// It's possible to extend the Pushgateway with request options from nodes core
// http/https library. In particular, you might want to provide an agent so that
// TCP connections are reused.
gateway = new client.Pushgateway('http://127.0.0.1:9091', {
timeout: 5000, //Set the request timeout to 5000ms
agent: new http.Agent({
keepAlive: true,
keepAliveMsec: 10000,
maxSockets: 5,
}),
});
Some gateways such as Gravel Gateway do not support grouping by job name, exposing a plain /metrics endpoint instead of /metrics/job/<jobName>. It's possible to configure a gateway instance to not require a jobName in the options argument.
gravelGateway = new client.Pushgateway('http://127.0.0.1:9091', {
timeout: 5000,
requireJobName: false,
});
gravelGateway.pushAdd();
For convenience, there are two bucket generator functions - linear and exponential.
const client = require('prom-client');
new client.Histogram({
name: 'metric_name',
help: 'metric_help',
buckets: client.linearBuckets(0, 10, 20), //Create 20 buckets, starting on 0 and a width of 10
});
new client.Histogram({
name: 'metric_name',
help: 'metric_help',
buckets: client.exponentialBuckets(1, 2, 5), //Create 5 buckets, starting on 1 and with a factor of 2
});
To avoid native dependencies in this module, GC statistics for bytes reclaimed in each GC sweep are kept in a separate module: https://github.com/SimenB/node-prometheus-gc-stats. (Note that that metric may no longer be accurate now that v8 uses parallel garbage collection.)
[15.1.3] - 2024-06-27
FAQs
Client for prometheus
The npm package prom-client receives a total of 385,129 weekly downloads. As such, prom-client popularity was classified as popular.
We found that prom-client demonstrated a healthy version release cadence and project activity because the last version was released less than a year ago. It has 0 open source maintainers collaborating on the project.
Did you know?
Socket for GitHub automatically highlights issues in each pull request and monitors the health of all your open source dependencies. Discover the contents of your packages and block harmful activity before you install or update your dependencies.
Security News
Cloudflare is expanding Node.js compatibility for Workers and Pages, enabling developers to use more npm packages through a hybrid approach that combines native code and polyfills for Node.js APIs.
Security News
The Python Software Foundation has expanded its CNA scope to include the Pallets Projects, enabling faster, more reliable CVE tracking for critical frameworks used in Python applications.
Security News
Elastic’s return to open source with the AGPL license has been met with skepticism, as many developers see it as a strategic move rather than a genuine effort to restore user trust and freedoms.