v2.2.4

Open Telemetry API	Core	Instrumentations
1.4.1	1.14.0	0.35.1

Changes

July 1, 2023

• fix: pin OpenTelemetry dependencies to avoid multiple versions of @opentelemetry/instrumentation packages being loaded at the same time #745

v2.2.3

Open Telemetry API	Core	Instrumentations
1.3.0	1.10.1	0.35.1

Changes

June 27, 2023

• Fixed logLevel configuration option for start function throwing an error #741
• Added Docker image for the OpenTelemetry Operator for Kubernetes #740

v2.2.2

Open Telemetry API	Core	Instrumentations
1.3.0	1.10.1	0.35.1

Changes

May 3, 2023

• Internal: added frame count to profiling data. #726

v2.2.1

Open Telemetry API	Core	Instrumentations
1.3.0	1.10.1	0.35.1

Changes

April 5, 2023

• Fixed @opentelemetry/instrumentation-mongoose not being loaded. #715

v2.2.0

Open Telemetry API	Core	Instrumentations
1.3.0	1.10.1	0.35.1

Changes

March 22, 2023

• Fixed SPLUNK_REALM environment variable taking precedence over endpoint supplied programmatically. endpoint now correctly overrides the endpoint created via SPLUNK_REALM and when both are set logs a warning. #668
• Empty environment variables are now considered as not defined. #693
• New configuration option: SPLUNK_DEBUG_METRICS_ENABLED / metrics.debugMetricsEnabled. #700 When set, extra set of internal troubleshooting metrics are produced. This should only be enabled to assist debugging. Defaults to false. Currently debug metrics for the CPU and memory profiler are produced, each being a histogram:
  • splunk.profiler.cpu.start.duration
  • splunk.profiler.cpu.stop.duration
  • splunk.profiler.cpu.process.duration
  • splunk.profiler.heap.collect.duration
  • splunk.profiler.heap.process.duration
• Upgrade to OpenTelemetry 1.10.1 / 0.35.1. The full changes can be seen at OpenTelemetry JS releases:
  • 1.9.0
  • 1.9.1
  • 1.10.0
  • 0.35.0
  • 0.35.1

v2.1.0

Open Telemetry API	Core	Instrumentations
1.3.0	1.8.0	0.34.0

Changes

1st of December, 2022

• Deduce the service name from package.json if it is not explicitly configured. #625
• Fix console metric exporter omitting datapoint specific attributes. #626

v2.0.0

Open Telemetry API	Core	Instrumentations
1.3.0	1.8.0	0.34.0

Changes

22nd of November, 2022

For a list of major changes and features in 2.0.0 see the notes for 2.0.0-rc1.

Additional changes in this release:

• Upgrade to OpenTelemetry JS 1.8.0 / 0.34.0. #612
• Remove @opentelemetry/instrumentation-aws-lambda from the bundled packages as there exists a separate lambda instrumentation and due to the package not being compatible with @opentelemetry/[email protected]. #612.
• @opentelemetry/api is now a peer dependency and the required version has been bumped to 1.3.0.
• OTEL_LOG_LEVEL now also sets up the logging pipeline, thus diagnostic logging can now be enabled just by enabling it
  via the environment variable. The supported log level values are none, verbose, debug, info, warn, error.
  The logging pipeline can additionally be enabled by setting logLevel configuration option. #605.
• process.command, process.command_line and process.runtime.description resource attributes have been removed from the automatic process detection. #613
• OTEL_TRACES_EXPORTER now only supports otlp, console or both (e.g. OTEL_TRACES_EXPORTER=otlp,console).
  #599
• Add support for OTEL_EXPORTER_OTLP_PROTOCOL, OTEL_EXPORTER_OTLP_TRACES_PROTOCOL, OTEL_EXPORTER_OTLP_METRICS_PROTOCOL. The supported values are grpc (the default) or http/protobuf. #599 #614

v2.0.0-rc2

Open Telemetry API	Core	Instrumentations
1.2.0	1.7.0	0.33.0

Changes

• Omit setting the default endpoint for metrics, as OpenTelemetry OTLP metrics exporters already have their own default configuration #592

v2.0.0-rc1

Open Telemetry API	Core	Instrumentations
1.2.0	1.7.0	0.33.0

Changes

• Deprecate startTracing, startMetrics, startProfiling functions

  There is a new function called start, which can be used to start all 3 signals:

  const { start } = require('@splunk/otel');
  
  start({
    serviceName: 'my-node-service',
    metrics: true, // Sets up OpenTelemetry metrics pipeline for custom metrics
    profiling: true, // Enables CPU profiling
  });

  If you have been only been using tracing, then startTracing can be replaced like this:

  const { start } = require('@splunk/otel');
  
  start({
    serviceName: 'my-node-service',
    endpoint: 'http://collector:4317',
  });

  Signal-specific configuration options can still be passed in:

  const { start } = require('@splunk/otel');
  
  start({
    serviceName: 'my-node-service',
    tracing: {
      instrumentations: [
        // Custom instrumentation list
      ],
    },
    metrics: {
      runtimeMetricsEnabled: true,
    },
    profiling: {
      memoryProfilingEnabled: true,
    }
  });

  For all possible options see Advanced Configuration.

  The deprecated functions are still available, but using them will log a deprecation message.

• Replace SignalFx metrics with OpenTelemetry metrics

  SignalFx metrics SDK has been removed and replaced with OpenTelemetry Metrics SDKs.
  The internal SignalFx client is no longer available to users, if you have been using custom metrics with the SignalFx client provided by Splunk OpenTelemetry JS distribution, see the migration guide for how to start using custom metrics via OpenTelemetry.

  Runtime metric names are now using OpenTelemetry conventions, the following is a list of changed metric names:

  SignalFx (no longer available)	OpenTelemetry
  nodejs.memory.heap.total	process.runtime.nodejs.memory.heap.total
  nodejs.memory.heap.used	process.runtime.nodejs.memory.heap.used
  nodejs.memory.rss	process.runtime.nodejs.memory.rss
  nodejs.memory.gc.size	process.runtime.nodejs.memory.gc.size
  nodejs.memory.gc.pause	process.runtime.nodejs.memory.gc.pause
  nodejs.memory.gc.count	process.runtime.nodejs.memory.gc.count
  nodejs.event_loop.lag.max	process.runtime.nodejs.event_loop.lag.max
  nodejs.event_loop.lag.min	process.runtime.nodejs.event_loop.lag.min

• Bundle instrumentations with the distribution

  It is no longer necessary to add instrumentation packages to your package.json file, unless you intend to use instrumentations not bundled in the distribution.

  For the list of instrumentations bundled by default see Default instrumentation packages.

• Remove Jaeger exporter

  It is no longer possible to use the jaeger-thrift-splunk value for OTEL_TRACES_EXPORTER to send traces via Jaeger Thrift over HTTP.

  The default exporting format is still OTLP over gRPC, however it is now possible to use OTLP over HTTP by setting OTLP_TRACES_EXPORTER environment variable to otlp-splunk.

  If you want to keep using Jaeger exporter, you can use the @opentelemetry/exporter-jaeger package by specifying a custom span exporter for the tracing configuration:

  const { start } = require('@splunk/otel');
  const { JaegerExporter } = require('@opentelemetry/exporter-jaeger');
  
  start({
    serviceName: 'my-node-service',
    tracing: {
      spanExporterFactory: (options) => {
        return new JaegerExporter({
          serviceName: options.serviceName,
          // Additional config
        })
      }
    },
  });

• Update OpenTelemetry SDKs

  Minimum Node.js version has been bumped to 14 due to OpenTelemetry SDKs dropping support for Node.js 12.

v1.4.1

Open Telemetry API	Core	Instrumentations
1.1.0	1.2.0	0.28.0

Changes

• (patch) chore: upgrade to signalfx 7.5.0 by [email protected]
• (patch) fix: don't produce otel diagnostic error when profiling is started by [email protected]