@uswitch/koa-zipkin
v1.11.13
Published
🕵️♀️ A koa.js middleware to add Zipkin tracing to requests
Downloads
5,545
Maintainers
Keywords
Readme
Overview
This package is a Koa library to do a lot of
the heavy lifting to enable zipkin
tracing around your service and
of upstream requests within your app.
It tries to encapsulate some of the run time mentality of how you build a service with the compile time integration of `zipkin.js** libraries and instrumentations.
N.B. - This library is opinionated. It provides;
- Instrumentation for
fetch
- A HTTP / Client side logger designed to work with
@uswitch/koa-signal
Usage
Using a fetch client without a cache
This is the simplest way to get zipkin tracing working with a koa application and sending to a Zipkin server.
// Define a module `zipkin.js`
import { middleware, fetch, Logger, Tracer } from '@uswitch/koa-zipkin'
import fetch from 'node-fetch'
const { middleware, fetch, Logger, Tracer } = koaZipkin
const { NODE_ENV, ZIPKIN_HOST, APP_NAME = 'example-service' } = process.env
const logger = Logger({ local: NODE_ENV === 'development', endpoint: ZIPKIN_HOST })
const tracer = Tracer(APP_NAME, logger)
export const zipkin = middleware({ tracer })
export const zipkinFetch = fetch({ local: APP_NAME, client: fetch, tracer })
// ----------------------------------------
// Then in your `koa` server you can just use
import { zipkin, zipkinFetch } from './zipkin'
app.use(zipkin)
app.use((ctx, next) => zipkinFetch({ remote: 'my-upstream' }, 'http://my-upstream'))
Using axios with a cache
Similarly to above, there are some benefits to using axios
with its
cache adapter. You get added logging of cache hits and misses as well
as the zipkin tracing. However, this needs a few extra bits like
adding the X-Cache-Enabled:true
header to your responses.
// Define a module `zipkin.js`
import koaZipkin from '@uswitch/koa-zipkin'
import axios from 'axios'
import cache from 'axios-cache-adapter'
const { middleware, fetch, Logger, Tracer } = koaZipkin
const { USE_CACHE = true, NODE_ENV, APP_NAME = 'example-uf-service' } = process.env
const { adapter } = cache.setupCache({ maxAge: USE_CACHE ? 10 * 1000 : 0 })
const client = axios.create({ adapter })
/* Mark the intent of using a cache */
client.interceptors.response.use(
(res) => {
res.headers['X-Cache-Enabled'] = true
return res
},
Promise.reject
)
const tracer = Tracer(APP_NAME, Logger({ local: NODE_ENV === 'development' }))
export const zipkin = middleware({ tracer })
export const zipkinFetch = fetch({ local: APP_NAME, client, tracer })
Enabling experimental ContextCLS feature to support tracing through code with promises / async / await
Please see https://github.com/openzipkin/zipkin-js/tree/master/packages/zipkin-context-cls#a-note-on-cls-context-and-promises, and note that by default - asynchronous code is not supported by ContextCLS. If you see fetches appearing in Zipkin with a different Trace ID, this could be the reason..
To enable asynchronous ContextCLS, create the Tracer object as follows:
const tracer = Tracer(APP_NAME, logger, true)
Please also note there are known performance implications of enabling this feature on Node versions below 16.3.0