Observability

edit

The client does not provide a default logger, but instead it offers an event emitter interface to hook into internal events, such as request and response.

Correlating those events can be hard, especially if your applications have a large codebase with many events happening at the same time.

To help you with this, the client offers you a correlation id system and other features. Let’s see them in action.

Events

edit

The client is an event emitter, this means that you can listen for its event and add additional logic to your code, without need to change the client internals or your normal usage. You can find the events names by access the events key of the client.

const { events } = require('@elastic/elasticsearch')
console.log(events)

The event emitter functionality can be useful if you want to log every request, response and error that is happening during the use of the client.

const logger = require('my-logger')()
const { Client } = require('@elastic/elasticsearch')
const client = new Client({
  cloud: { id: '<cloud-id>' },
  auth: { apiKey: 'base64EncodedKey' }
})

client.diagnostic.on('response', (err, result) => {
  if (err) {
    logger.error(err)
  } else {
    logger.info(result)
  }
})

The client emits the following events:

serialization

Emitted before starting serialization and compression. If you want to measure this phase duration, you should measure the time elapsed between this event and request.

client.diagnostic.on('serialization', (err, result) => {
  console.log(err, result)
})

request

Emitted before sending the actual request to Elasticsearch (emitted multiple times in case of retries).

client.diagnostic.on('request', (err, result) => {
  console.log(err, result)
})

deserialization

Emitted before starting deserialization and decompression. If you want to measure this phase duration, you should measure the time elapsed between this event and response. (This event might not be emitted in certain situations).

client.diagnostic.on('deserialization', (err, result) => {
  console.log(err, result)
})

response

Emitted once Elasticsearch response has been received and parsed.

client.diagnostic.on('response', (err, result) => {
  console.log(err, result)
})

sniff

Emitted when the client ends a sniffing request.

client.diagnostic.on('sniff', (err, result) => {
  console.log(err, result)
})

resurrect

Emitted if the client is able to resurrect a dead node.

client.diagnostic.on('resurrect', (err, result) => {
  console.log(err, result)
})

The values of result in serialization, request, deserialization, response and sniff are:

body: any;
statusCode: number | null;
headers: anyObject | null;
warnings: string[] | null;
meta: {
  context: any;
  name: string;
  request: {
    params: TransportRequestParams;
    options: TransportRequestOptions;
    id: any;
  };
  connection: Connection;
  attempts: number;
  aborted: boolean;
  sniff?: {
    hosts: any[];
    reason: string;
  };
};

While the result value in resurrect is:

strategy: string;
isAlive: boolean;
connection: Connection;
name: string;
request: {
  id: any;
};
Events order
edit

The event order is described in the following graph, in some edge cases, the order is not guaranteed. You can find in test/acceptance/events-order.test.js how the order changes based on the situation.

serialization
  │
  │ (serialization and compression happens between those two events)
  │
  └─▶ request
        │
        │ (actual time spent over the wire)
        │
        └─▶ deserialization
              │
              │ (deserialization and decompression happens between those two events)
              │
              └─▶ response

Correlation id

edit

Correlating events can be hard, especially if there are many events at the same time. The client offers you an automatic (and configurable) system to help you handle this problem.

const { Client } = require('@elastic/elasticsearch')
const client = new Client({
  cloud: { id: '<cloud-id>' },
  auth: { apiKey: 'base64EncodedKey' }
})

client.diagnostic.on('request', (err, result) => {
  const { id } = result.meta.request
  if (err) {
    console.log({ error: err, reqId: id })
  }
})

client.diagnostic.on('response', (err, result) => {
  const { id } = result.meta.request
  if (err) {
    console.log({ error: err, reqId: id })
  }
})

client.search({
  index: 'my-index',
  query: { match_all: {} }
}).then(console.log, console.log)

By default the id is an incremental integer, but you can configure it with the generateRequestId option:

const { Client } = require('@elastic/elasticsearch')
const client = new Client({
  cloud: { id: '<cloud-id>' },
  auth: { apiKey: 'base64EncodedKey' },
  // it takes two parameters, the request parameters and options
  generateRequestId: function (params, options) {
    // your id generation logic
    // must be syncronous
    return 'id'
  }
})

You can also specify a custom id per request:

client.search({
  index: 'my-index',
  query: { match_all: {} }
}, {
  id: 'custom-id'
}).then(console.log, console.log)

Context object

edit

Sometimes, you might need to make some custom data available in your events, you can do that via the context option of a request:

const { Client } = require('@elastic/elasticsearch')
const client = new Client({
  cloud: { id: '<cloud-id>' },
  auth: { apiKey: 'base64EncodedKey' }
})

client.diagnostic.on('request', (err, result) => {
  const { id } = result.meta.request
  const { context } = result.meta
  if (err) {
    console.log({ error: err, reqId: id, context })
  }
})

client.diagnostic.on('response', (err, result) => {
  const { id } = result.meta.request
  const { winter } = result.meta.context
  if (err) {
    console.log({ error: err, reqId: id, winter })
  }
})

client.search({
  index: 'my-index',
  query: { match_all: {} }
}, {
  context: { winter: 'is coming' }
}).then(console.log, console.log)

The context object can also be configured as a global option in the client configuration. If you provide both, the two context objects will be shallow merged, and the API level object will take precedence.

const { Client } = require('@elastic/elasticsearch')
const client = new Client({
  cloud: { id: '<cloud-id>' },
  auth: { apiKey: 'base64EncodedKey' },
  context: { winter: 'is coming' }
})

client.diagnostic.on('request', (err, result) => {
  const { id } = result.meta.request
  const { context } = result.meta
  if (err) {
    console.log({ error: err, reqId: id, context })
  }
})

client.diagnostic.on('response', (err, result) => {
  const { id } = result.meta.request
  const { winter } = result.meta.context
  if (err) {
    console.log({ error: err, reqId: id, winter })
  }
})

client.search({
  index: 'my-index',
  query: { match_all: {} }
}, {
  context: { winter: 'has come' }
}).then(console.log, console.log)

Client name

edit

If you are using multiple instances of the client or if you are using multiple child clients (which is the recommended way to have multiple instances of the client), you might need to recognize which client you are using. The name options help you in this regard.

const { Client } = require('@elastic/elasticsearch')
const client = new Client({
  cloud: { id: '<cloud-id>' },
  auth: { apiKey: 'base64EncodedKey' },
  name: 'parent-client' // default to 'elasticsearch-js'
})

const child = client.child({
  name: 'child-client'
})

console.log(client.name, child.name)

client.diagnostic.on('request', (err, result) => {
  const { id } = result.meta.request
  const { name } = result.meta
  if (err) {
    console.log({ error: err, reqId: id, name })
  }
})

client.diagnostic.on('response', (err, result) => {
  const { id } = result.meta.request
  const { name } = result.meta
  if (err) {
    console.log({ error: err, reqId: id, name })
  }
})

client.search({
  index: 'my-index',
  query: { match_all: {} }
}).then(console.log, console.log)

child.search({
  index: 'my-index',
  query: { match_all: {} }
}).then(console.log, console.log)

X-Opaque-Id support

edit

To improve observability, the client offers an easy way to configure the X-Opaque-Id header. If you set the X-Opaque-Id in a specific request, this allows you to discover this identifier in the deprecation logs, helps you with identifying search slow log origin as well as identifying running tasks.

The X-Opaque-Id should be configured in each request, for doing that you can use the opaqueId option, as you can see in the following example. The resulting header will be { 'X-Opaque-Id': 'my-search' }.

const { Client } = require('@elastic/elasticsearch')
const client = new Client({
  cloud: { id: '<cloud-id>' },
  auth: { apiKey: 'base64EncodedKey' }
})

client.search({
  index: 'my-index',
  body: { foo: 'bar' }
}, {
  opaqueId: 'my-search'
}).then(console.log, console.log)

Sometimes it may be useful to prefix all the X-Opaque-Id headers with a specific string, in case you need to identify a specific client or server. For doing this, the client offers a top-level configuration option: opaqueIdPrefix. In the following example, the resulting header will be { 'X-Opaque-Id': 'proxy-client::my-search' }.

const { Client } = require('@elastic/elasticsearch')
const client = new Client({
  cloud: { id: '<cloud-id>' },
  auth: { apiKey: 'base64EncodedKey' },
  opaqueIdPrefix: 'proxy-client::'
})

client.search({
  index: 'my-index',
  body: { foo: 'bar' }
}, {
  opaqueId: 'my-search'
}).then(console.log, console.log)