Agent API

edit

You can access agent API after initializing the agent:

var apm = require('@elastic/apm-rum').init(...)

apm.init([config])

edit

Initializes the agent with the given configuration and returns itself. Under the hood init does the following

  • Registers a global error listener to track JavaScript errors on the page.
  • Adds a onload event listener to collect the page load metrics.

When the agent is inactive, both error and onload listeners will not be registered on the page

Both XHR and Fetch API are patched as soon as the agent script is executed on the page and does not get changed even if the agent is inactive. The reason is to allow users to initialize the agent asynchronously on the page.

apm.setUserContext()

edit
apm.setUserContext(context)

Call this method to enrich collected performance data and errors with information about the user.

The given context argument must be an object and can contain the following properties (all optional):

  • id - The users ID
  • username - The users username
  • email - The users e-mail

The provided user context is stored under context.user in Elasticsearch on both errors and transactions.

It’s possible to call this function multiple times within the scope of the same active transaction. For each call, the properties of the context argument are shallow merged with the context previously given.

apm.setCustomContext()

edit
apm.setCustomContext(context)

Call this to enrich collected errors and transactions with any information that you think will help you debug performance issues or errors.

The provided custom context is stored under context.custom in Elasticsearch on both errors and transactions.

It’s possible to call this function multiple times within the scope of the same active transaction. For each call, the properties of the context argument are shallow merged with the context previously given.

The given context argument must be an object and can contain any property that can be JSON encoded.

Before using custom context, ensure you understand the different types of metadata that are available.

apm.addTags()

edit

[4.1.x] Deprecated in 4.1.x. Tags have been replaced with labels , Please use apm.addLabels() instead

apm.addLabels()

edit
apm.addLabels({ [name]: value })

Set labels on transactions and errors. Starting with APM Server 7.6+, the labels are added to spans as well.

Labels are key/value pairs that are indexed by Elasticsearch and therefore searchable (as opposed to data set via apm.setCustomContext()). You can set multiple labels.

Before using custom labels, ensure you understand the different types of metadata that are available.

Arguments:

  • name - Any string. All periods (.), asterisks (*), and double quotation marks (") will be replaced by underscores (_), as those characters have special meaning in Elasticsearch
  • value - Any string. If a non-string data type is given, it’s converted to a string before being sent to the APM Server.

Avoid defining too many user-specified labels. Defining too many unique fields in an index is a condition that can lead to a mapping explosion.

apm.addFilter()

edit

A filter can be used to modify the APM payload before it is sent to the apm-server. This can be useful in for example redacting sensitive information from the payload:

apm.addFilter(function (payload) {
  if (payload.errors) {
    payload.errors.forEach(function (error) {
      error.exception.message = error.exception.message.replace('secret', '[REDACTED]')
    })
  }
  if (payload.transactions) {
    payload.transactions.forEach(function (tr) {
      tr.spans.forEach(function (span) {
        if (span.context && span.context.http && span.context.http.url) {
          var url = new URL(span.context.http.url)
          if (url.searchParams.get('token')) {
            url.searchParams.set('token', 'REDACTED')
          }
          span.context.http.url = url.toString()
        }
      })
    })
  }
  // Make sure to return the payload
  return payload
})

The payload will be dropped if one of the filters return a falsy value.

apm.startTransaction()

edit
apm.startTransaction(name, type, options)

Starts and returns a new transaction.

Arguments:

  • name - The name of the transaction (string). Defaults to Unknown
  • type - The type of the transaction (string). Defaults to custom
  • options - Options to modify the created transaction (object). This argument is optional. The following options are supported:

    • managed - Controls whether the transaction is managed by the agent or not. Defaults to false.

Use this method to create a custom transaction.

By default, custom transactions are not managed by the agent, however, you can start a managed transaction by passing { managed: true } as the options argument.

There are some differences between managed and unmanaged transactions:

  • For managed transactions, the agent keeps track of the relevant tasks during the lifetime of the transaction and automatically ends it once all of the tasks are finished. Unmanaged transactions need to be ended manually by calling the end method.
  • Managed transactions include information captured via our auto-instrumentations (e.g. XHR spans). See Supported Technologies for a list of instrumentations.
  • There can only be one managed transaction at any given time — starting a second managed transaction will end the previous one. There are no limits for unmanaged transactions.

This method returns undefined if apm is disabled or if active flag is set to false in the config.

apm.startSpan()

edit
var span = apm.startSpan(name, type)

Starts and returns a new span on the current transaction.

Arguments:

  • name - The name of the span (string). Defaults to Unknown
  • type - The type of the span (string). Defaults to custom

This method returns undefined if apm is disabled or if active flag is set to false in the config.

apm.setInitialPageLoadName()

edit
apm.setInitialPageLoadName(name)

Arguments:

  • name - The name of the page-load transaction (string).

Use this method to set the name of the page-load transaction that is sent automatically on page load event. See the custom initial page load transaction names documentation for strategies on using this method.

apm.getCurrentTransaction()

edit
apm.getCurrentTransaction()

Use this method to get the current active transaction. If there is no active transaction it will return undefined.

apm.captureError()

edit
apm.captureError(error)

Arguments:

  • error - An instance of Error.

Use this method to manually send an error to APM Server:

apm.captureError(new Error('<error-message>'))

apm.observe()

edit
apm.observe(name, callback)

Arguments:

  • name - The name of the event.
  • callback - A callback function to execute once the event is fired.

Use this method to listen for RUM agent internal events.

The following events are supported for the transaction lifecycle:

  • transaction:start event is fired on every transaction start.
  • transaction:end event is fired on transaciton end and before it is added to the queue to be sent to APM Server.

The callback function for these events receives the corresponding transaction object as its only argument. The transaction object can be modified through methods and properties documented in Transaction API:

apm.observe('transaction:start', function (transaction) {
  if (transaction.type === 'custom') {
    transaction.name = window.document.title
    transaction.addLabels({ 'custom-label': 'custom-value' })
  }
})