Using remote_write facility (see https://prometheus.io/docs/prometheus/latest/configuration/configuration/#remote_write) to send metrics to remote Prometheus from NodeJS app.
Pretty much anything from https://prometheus.io/docs/operating/integrations/#remote-endpoints-and-storage should be supported, but tested only with Grafana Cloud, Grafana Mimir Prometheus, AWS AMP and VictoriaMetrics.
(List copied from https://github.com/prometheus/docs/blob/main/content/docs/operating/integrations.md)
import { pushTimeseries, pushMetrics } from "prometheus-remote-write";
// Just push some metrics metrics
await pushMetrics(
{
queue_depth_total: 100,
},
{
url: process.env.GRAFANA_PUSH_URL || "http://localhost:9201",
labels: { service: "queue-worker" },
}
);
// Follows remote_write payload format (see https://github.com/prometheus/prometheus/blob/main/prompb/types.proto)
await pushTimeseries(
{
labels: {
// Name should conform to https://prometheus.io/docs/practices/naming/
__name__: "queue_depth_total",
instance: "dev.example.com",
service: "SQS",
},
samples: [
{
value: 150,
timestamp: Date.now(),
},
],
},
config
);
// Full config - only url is required
const config = {
// Remote url
url: "http://localhost:9201",
// Auth settings
auth: {
username: "...",
password: "...",
},
// Optional prometheus protocol descripton .proto/.json
proto: undefined,
// Override default console.name(...log) used
console: undefined,
// Be verbose
verbose: false,
timing: false,
// Override used node-fetch
fetch: undefined,
// Additional labels to apply to each timeseries, i.e. [{ service: "SQS" }]
labels: undefined,
// Additional HTTP headers to send with each request
headers: undefined
};