Skip to content
This repository has been archived by the owner on Aug 4, 2023. It is now read-only.

Latest commit

 

History

History
505 lines (388 loc) · 21.9 KB

README.md

File metadata and controls

505 lines (388 loc) · 21.9 KB
Raw

Note: This repository will no longer be updated. This APM HTTP client code was moved into the elastic/apm-agent-nodejs repository in 2023-07. The code now lives at: https://github.com/elastic/apm-agent-nodejs/tree/main/lib/apm-client/http-apm-client. Please open any issues, questions, PRs for the APM HTTP client there.

elastic-apm-http-client

A low-level HTTP client for communicating with the Elastic APM intake API version 2. For support for version 1, use version 5.x of this module.

This module is meant for building other modules that needs to communicate with Elastic APM.

If you are looking to use Elastic APM in your app or website, you'd most likely want to check out the official Elastic APM agent for Node.js instead.

Installation

npm install elastic-apm-http-client --save

Example Usage

const Client = require('elastic-apm-http-client')

const client = new Client({
  serviceName: 'My App',
  agentName: 'my-nodejs-agent',
  agentVersion: require('./package.json').version,
  userAgent: 'My Custom Elastic APM Agent'
})

const span = {
  name: 'SELECT FROM users',
  duration: 42,
  start: 0,
  type: 'db.mysql.query'
}

client.sendSpan(span)

API

new Client(options)

Construct a new client object. Data given to the client will be converted to ndjson, compressed using gzip, and streamed to the APM Server.

Arguments:

  • options - An object containing config options (see below). All options are optional, except those marked "(required)".

Data sent to the APM Server as part of the metadata object. See also the "Cloud & Extra Metadata" section below.

  • agentName - (required) The APM agent name
  • agentVersion - (required) The APM agent version
  • agentActivationMethod - An enum string (spec) that identifies the way this agent was activated/started
  • serviceName - (required) The name of the service being instrumented
  • serviceNodeName - Unique name of the service being instrumented
  • serviceVersion - The version of the service being instrumented
  • frameworkName - If the service being instrumented is running a specific framework, use this config option to log its name
  • frameworkVersion - If the service being instrumented is running a specific framework, use this config option to log its version
  • configuredHostname - A user-configured hostname, if any, e.g. from the ELASTIC_APM_HOSTNAME envvar. See https://github.com/elastic/apm/blob/main/specs/agents/metadata.md#hostname.
  • environment - Environment name (default: process.env.NODE_ENV || 'development')
  • containerId - Docker container id, if not given will be parsed from /proc/self/cgroup
  • kubernetesNodeName - Kubernetes node name
  • kubernetesNamespace - Kubernetes namespace
  • kubernetesPodName - Kubernetes pod name, if not given will be the hostname
  • kubernetesPodUID - Kubernetes pod id, if not given will be parsed from /proc/self/cgroup
  • globalLabels - An object of key/value pairs to use to label all data reported (only applied when using APM Server 7.1+)

HTTP client configuration:

  • userAgent - (required) The HTTP user agent that your module should identify itself as
  • secretToken - The Elastic APM intake API secret token
  • apiKey - Elastic APM API key
  • serverUrl - The APM Server URL (default: http://127.0.0.1:8200)
  • headers - An object containing extra HTTP headers that should be used when making HTTP requests to he APM Server
  • rejectUnauthorized - Set to false if the client shouldn't verify the APM Server TLS certificates (default: true)
  • serverCaCert - The CA certificate used to verify the APM Server's TLS certificate, and has the same requirements as the ca option of tls.createSecureContext.
  • serverTimeout - HTTP request timeout in milliseconds. If no data is sent or received on the socket for this amount of time, the request will be aborted. It's not recommended to set a serverTimeout lower than the time config option. That might result in healthy requests being aborted prematurely. (default: 15000 ms)
  • keepAlive - If set the false the client will not reuse sockets between requests (default: true)
  • keepAliveMsecs - When using the keepAlive option, specifies the initial delay for TCP Keep-Alive packets. Ignored when the keepAlive option is false or undefined (default: 1000 ms)
  • maxSockets - Maximum number of sockets to allow per host (default: Infinity)
  • maxFreeSockets - Maximum number of sockets to leave open in a free state. Only relevant if keepAlive is set to true (default: 256)
  • freeSocketTimeout - A number of milliseconds of inactivity on a free (kept-alive) socket after which to timeout and recycle the socket. Set this to a value less than the HTTP Keep-Alive timeout of the APM server to avoid ECONNRESET exceptions. This defaults to 4000ms to be less than the node.js HTTP server default of 5s (useful when using a Node.js-based mock APM server) and the Go lang Dialer KeepAlive default of 15s (when talking to the Elastic APM Lambda extension). (default: 4000)

Cloud & Extra Metadata Configuration. Zero or one of the following three options may be used.

  • cloudMetadataFetcher - An object with a getCloudMetadata(cb) method for fetching metadata related to the current cloud environment. The callback is of the form function (err, cloudMetadata) and the returned cloudMetadata will be set on metadata.cloud for intake requests to APM Server. If provided, this client will not begin any intake requests until the callback is called.
  • expectExtraMetadata - A boolean option to indicate that the client should not allow any intake requests to begin until cloud.setExtraMetadata(...) has been called. It is the responsibility of the caller to call cloud.setExtraMetadata(). If not, then the Client will never perform an intake request.
  • extraMetadata - An object with extra metadata to merge into the metadata object created from the individual fields above.

APM Agent Configuration via Kibana:

  • centralConfig - Whether or not the client should poll the APM Server regularly for new agent configuration. If set to true, the config event will be emitted when there's an update to an agent config option (default: false). Requires APM Server v7.3 or later and that the APM Server is configured with kibana.enabled: true.

Streaming configuration:

  • size - The maxiumum compressed body size (in bytes) of each HTTP request to the APM Server. An overshoot of up to the size of the internal zlib buffer should be expected as the buffer is flushed after this limit is reached. The default zlib buffer size is 16kB. (default: 768000 bytes)
  • time - The maxiumum number of milliseconds a streaming HTTP request to the APM Server can be ongoing before it's ended. Set to -1 to disable (default: 10000 ms)
  • bufferWindowTime - Objects written in quick succession are buffered and grouped into larger clusters that can be processed as a whole. This config option controls the maximum time that buffer can live before it's flushed (counted in milliseconds). Set to -1 for no buffering (default: 20 ms)
  • bufferWindowSize - Objects written in quick succession are buffered and grouped into larger clusters that can be processed as a whole. This config option controls the maximum size of that buffer (counted in number of objects). Set to -1 for no max size (default: 50 objects)
  • maxQueueSize - The maximum number of buffered events (transactions, spans, errors, metricsets). Events are buffered when the agent can't keep up with sending them to the APM Server or if the APM server is down. If the queue is full, events are rejected which means transactions, spans, etc. will be lost. This guards the application from consuming unbounded memory, possibly overusing CPU (spent on serializing events), and possibly crashing in case the APM server is unavailable for a long period of time. A lower value will decrease the heap overhead of the agent, while a higher value makes it less likely to lose events in case of a temporary spike in throughput. (default: 1024)
  • intakeResTimeout - The time (in milliseconds) by which a response from the APM Server events intake API is expected after all the event data for that request has been sent. This allows a smaller timeout than serverTimeout to handle an APM server that is accepting connections but is slow to respond. (default: 10000 ms)
  • intakeResTimeoutOnEnd - The same as intakeResTimeout, but used when the client has ended, hence for the possible last request to APM server. This is typically a lower value to not hang an ending process that is waiting for that APM server request to complete. (default: 1000 ms)

Data sanitizing configuration:

  • truncateKeywordsAt - Maximum size in unicode characters for strings stored as Elasticsearch keywords. Strings larger than this will be trucated (default: 1024)
  • truncateLongFieldsAt - The maximum size in unicode characters for a specific set of long string fields. String values above this length will be truncated. Default: 10000. This applies to the following fields:
    • transaction.context.request.body, error.context.request.body
    • transaction.context.message.body, span.context.message.body, error.context.message.body
    • span.context.db.statement
    • error.exception.message (unless truncateErrorMessagesAt is specified)
    • error.log.message (unless truncateErrorMessagesAt is specified)
  • truncateStringsAt - The maximum size in unicode characters for strings. String values above this length will be truncated (default: 1024)
  • truncateErrorMessagesAt - DEPRECATED: prefer truncateLongFieldsAt. The maximum size in unicode characters for error messages. Messages above this length will be truncated. Set to -1 to disable truncation. This applies to the following properties: error.exception.message and error.log.message. (default: 2048)

Other options:

  • logger - A pino logger to use for trace and debug-level logging.
  • payloadLogFile - Specify a file path to which a copy of all data sent to the APM Server should be written. The data will be in ndjson format and will be uncompressed. Note that using this option can impact performance.
  • apmServerVersion - A string version to assume is the version of the APM Server at serverUrl. This option is typically only used for testing. Normally this client will fetch the APM Server version at startup via a GET / request. Setting this option avoids that request.

Event: config

Emitted every time a change to the agent config is pulled from the APM Server. The listener is passed the updated config options as a key/value object.

Each key is the lowercase version of the environment variable, without the ELASTIC_APM_ prefix, e.g. transaction_sample_rate instead of ELASTIC_APM_TRANSACTION_SAMPLE_RATE.

If no central configuration is set up for the given serviceName / environment when the client is started, this event will be emitted once with an empty object. This will also happen after central configuration for the given serviceName / environment is deleted.

Event: close

The close event is emitted when the client and any of its underlying resources have been closed. The event indicates that no more events will be emitted, and no more data can be sent by the client.

Event: error

Emitted if an error occurs. The listener callback is passed a single Error argument when called.

Event: finish

The finish event is emitted after the client.end() method has been called, and all data has been flushed to the underlying system.

Event: request-error

Emitted if an error occurs while communicating with the APM Server. The listener callback is passed a single Error argument when called.

The request to the APM Server that caused the error is terminated and the data included in that request is lost. This is normally only important to consider for requests to the Intake API.

If a non-2xx response was received from the APM Server, the status code will be available on error.code.

For requests to the Intake API where the response is a structured error message, the error object will have the following properties:

  • error.accepted - An integer indicating how many events was accepted as part of the failed request. If 100 events was sent to the APM Server as part of the request, and the error reports only 98 as accepted, it means that two events either wasn't received or couldn't be processed for some reason
  • error.errors - An array of error messages. Each element in the array is an object containing a message property (String) and an optional document property (String). If the document property is given it will contain the failed event as it was received by the APM Server

If the response contained an error body that could not be parsed by the client, the raw body will be available on error.response.

The client is not closed when the request-error event is emitted.

client.sent

An integer indicating the number of events (spans, transactions, errors, or metricsets) sent by the client. An event is considered sent when the HTTP request used to transmit it has ended. Note that errors in requests to APM server may mean this value is not the same as the number of events accepted by the APM server.

client.config(options)

Update the configuration given to the Client constructor. All configuration options can be updated except:

  • size
  • time
  • keepAlive
  • keepAliveMsecs
  • maxSockets
  • maxFreeSockets
  • centralConfig

client.supportsKeepingUnsampledTransaction()

This method returns a boolean indicating whether the remote APM Server (per the configured serverUrl) is of a version that requires unsampled transactions to be sent.

This defaults to true if the remote APM server version is not known -- either because the background fetch of the APM Server version hasn't yet completed, or the version could not be fetched.

client.supportsActivationMethodField()

This method returns a boolean indicating whether the remote APM Server (per the configured serverUrl) is of a version that supports the metadata.service.agent.activation_method field. This is true for APM server versions >=8.7.1. It defaults to true if the APM server version is not (yet) known.

client.addMetadataFilter(fn)

Add a filter function for the "metadata" object sent to APM server. This will be called once at client creation, and possibly again later if client.config() is called to reconfigure the client or client.addMetadataFilter(fn) is called to add additional filters.

Here is an example of a filter that removes the metadata.process.argv field:

apm.addMetadataFilter(function dropArgv(md) {
  if (md.process && md.process.argv) {
    delete md.process.argv
  }
  return md
})

It is up to the user to ensure the returned object conforms to the metadata schema, otherwise APM data injest will be broken. An example of that (when used with the Node.js APM agent) is this in the application's log:

[2021-04-14T22:28:35.419Z] ERROR (elastic-apm-node): APM Server transport error (400): Unexpected APM Server response
APM Server accepted 0 events in the last request
Error: validation error: 'metadata' required
  Document: {"metadata":null}

See the APM Agent addMetadataFilter documentation for further details.

client.setExtraMetadata([metadata])

Add extra metadata to be included in the "metadata" object sent to APM Server in intake requests. The given metadata object is merged into the metadata determined from the client configuration.

The reason this exists is to allow some metadata to be provided asynchronously, especially in combination with the expectExtraMetadata configuration option to ensure that event data is not sent to APM Server until this extra metadata is provided. For example, in an AWS Lambda function some metadata is not available until the first function invocation -- which is some async time after Client creation.

client.lambdaStart()

Tells the client that a Lambda function invocation has started. See Notes on Lambda Usage below.

client.lambdaShouldRegisterTransactions()

This returns a boolean indicating if the APM agent -- when running in a Lambda environment -- should bother calling client.lambdaRegisterTransaction(...). This can help the APM agent avoid some processing gathering transaction data.

Typically the reason this would return false is when the Lambda extension is too old to support registering transactions.

client.lambdaRegisterTransaction(transaction, awsRequestId)

Tells the Lambda Extension about the ongoing transaction, so that data can be used to report the transaction in certain error cases -- e.g. a Lambda handler timeout. See Notes on Lambda Usage below.

Arguments:

  • transaction - A transaction object that can be serialized to JSON.
  • awsRequestId - The AWS request ID for this invocation. This is a UUID available on the Lambda context object.

client.sendSpan(span[, callback])

Send a span to the APM Server.

Arguments:

  • span - A span object that can be serialized to JSON
  • callback - Callback is called when the span have been flushed to the underlying system

client.sendTransaction(transaction[, callback])

Send a transaction to the APM Server.

Arguments:

  • transaction - A transaction object that can be serialized to JSON
  • callback - Callback is called when the transaction have been flushed to the underlying system

client.sendError(error[, callback])

Send a error to the APM Server.

Arguments:

  • error - A error object that can be serialized to JSON
  • callback - Callback is called when the error have been flushed to the underlying system

client.sendMetricSet(metricset[, callback])

Send a metricset to the APM Server.

Arguments:

  • error - A error object that can be serialized to JSON
  • callback - Callback is called when the metricset have been flushed to the underlying system

client.flush([opts,] [callback])

Flush the internal buffer and end the current HTTP request to the APM Server. If no HTTP request is in process nothing happens. In an AWS Lambda environment this will also initiate a quicker shutdown of the intake request, because the APM agent always flushes at the end of a Lambda handler.

Arguments:

  • opts:
    • opts.lambdaEnd - An optional boolean to indicate if this is the final flush at the end of the Lambda function invocation. The client will do some extra handling if this is the case. See notes in client.lambdaStart() above.
  • callback - Callback is called when the internal buffer has been flushed and the HTTP request ended. If no HTTP request is in progress the callback is called in the next tick.

client.end([callback])

Calling the client.end() method signals that no more data will be sent to the client. If the internal buffer contains any data, this is flushed before ending.

Arguments:

  • callback - If provided, the optional callback function is attached as a listener for the 'finish' event

client.destroy()

Destroy the client. After this call, the client has ended and subsequent calls to sendSpan(), sendTransaction(), sendError(), flush(), or end() will result in an error.

Notes on Lambda usage

To properly handle data flushing for instrumented Lambda functions this Client should be used as follows in a Lambda environment.

  1. Ensure that metadata is set before any of the following calls. Typically in Lambda this is done by (a) configuring the client with expectExtraMetadata and (b) calling setExtraMetadata() at the start of the first invocation.

  2. When a Lambda invocation starts, client.lambdaStart() must be called. The Client prevents intake requests to APM Server when in a Lambda environment when a function invocation is not active. This is to ensure that an intake request does not accidentally span a period when a Lambda VM is frozen, which can lead to timeouts and lost APM data.

  3. When the transaction for this Lambda invocation has been created, await client.lambdaRegisterTransaction(<transaction>, <awsRequestId>) should be called. This is used to pass transaction details to the Lambda Extension so a transaction can be reported in certain failure modes (e.g. a Lambda handler timeout).

    client.lambdaShouldRegisterTransactions() can be used to avoid gathering data for this call.

  4. When a Lambda invocation finishes, client.flush({lambdaEnd: true}, cb) must be called.

    The lambdaEnd: true tells the Client to (a) mark the lambda as inactive so a subsequent intake request is not started until the next invocation, and (b) signal the Elastic AWS Lambda Extension that this invocation is done. The user's Lambda handler should not finish until cb is called. This ensures that the extension receives tracing data and the end signal before the Lambda Runtime freezes the VM.

License

MIT