@elastic/opentelemetry-node 1.7.0

Changelog

Features and enhancements

• Add support for TLS/mTLS configuration of the OpAMP client (used for central configuration). Three new environment variables can be used: ELASTIC_OTEL_OPAMP_CERTIFICATE, ELASTIC_OTEL_OPAMP_CLIENT_CERTIFICATE, and ELASTIC_OTEL_OPAMP_CLIENT_KEY. See the Configure central configuration doc section for details. #1044

• Improve the "preamble" log message at startup to include some details on envvars set that are relevant to the EDOT Node.js config. This uses an allowlist of non-sensitive envvars to avoid logging sensitive details. #1018

• Add support for the opamp_polling_interval setting in Central Configuration. #1128

  While this is supported in the EDOT Node.js SDK, it will only be present in
  the "Agent Configuration" UI of Kibana version 9.3 and later.

Fixes

• Update @opentelemetry/instrumentation-openai dependency to version 0.7.0 to resolve a possible issue with TypeScript compilation. opentelemetry-js-contrib#3240

---

README | Full Release Notes | Breaking Changes

@elastic/opamp-client-node 0.4.0

Changelog

• BREAKING CHANGE: The heartbeatIntervalSeconds option to createOpAMPClient
  used to clamp the given value to [100ms, 1d]. Starting in this version,
  a value less than 100ms will be ignored, and the default value will be
  used. The reason for this is to ignore a possible accidental error case where
  zero or a negative number is provided, resulting in a too-fast 100ms
  interval. (The 100ms lower bound really only exists for faster testing. It
  is not a reasonable value for production.)

• Add opampClient.setHeartbeatIntervalSeconds(num) and
  .resetHeartbeatIntervalSeconds() methods for dynamically changing the
  heartbeat interval used by the OpAMP client. Values less than 100ms are
  ignored (with a log.warn) and values greater than 1d are clamped to 1d.
  #1128

---

README | Full Changelog

@elastic/opamp-client-node 0.3.0

Changelog

• Add TLS and mTLS support. #1044

  const client = createOpAMPClient({
      // ...
      connect: {
          ca: fs.readFileSync(path.join(CERTS_DIR, 'ca.crt')),
          cert: fs.readFileSync(path.join(CERTS_DIR, 'client.crt')),
          key: fs.readFileSync(path.join(CERTS_DIR, 'client.key')),
      }
  });

• Be more defensive in handling the heartbeatIntervalSeconds option: clamp to
  [100ms, 1d] and use the default (30s) for invalid values. The previous
  behaviour was to throw on invalid values, and to not have a max value.

---

README | Full Changelog

@elastic/mockopampserver 0.5.0

Changelog

• Add TLS and mTLS support. #1044

  The mockopampserver CLI now has these new options for setting TLS data, and
  for requesting client certs (for mTLS):

  --cacert=ARG             Override the built-in-to-Node.js trusted CA
                           certificates with the content of this file, in PEM
                           format. Enables HTTPS.
  --cert=ARG               Path to file with TLS certificate chains, in PEM
                           format, to use. Enables HTTPS.
  --key=ARG                Path to server TLS private keys, in PEM format.
                           Enables HTTPS.
  --request-client-cert    Tells the server to request TLS certs from
                           connecting clients. Enables HTTPS.
  

  The code interface also has equivalent new options:

  const opampServer = new MockOpAMPServer({
      // ...
      ca: fs.readFileSync("..."),
      cert: fs.readFileSync("..."),
      key: fs.readFileSync("..."),
      requestCert: true
  });

---

README | Full Changelog

@elastic/opentelemetry-node 1.6.0

Changelog

Features and enhancements

• Add support for ELASTIC_OTEL_OPAMP_HEADERS environment variable for setting headers for OpAMP requests. This typically would be used to specify authorization for a given ELASTIC_OTEL_OPAMP_ENDPOINT, e.g.: export ELASTIC_OTEL_OPAMP_HEADERS="Authorization=ApiKey sekrit". #1069

• HTTP exporters now accept TLS configuration via the env vars OTEL_EXPORTER_OTLP_CERTIFICATE, OTEL_EXPORTER_OTLP_CLIENT_KEY and OTEL_EXPORTER_OTLP_CLIENT_CERTIFICATE. The signal specific vars (OTEL_EXPORTER_OTLP_TRACES_CERTIFICATE, ...) are also taken into account.

• Prepend EDOT Node.js product identifier in the user-agent for all HTTP and GRPC exporters. (#1104)

• Switch back to the upstream resource detector @opentelemetry/resource-detector-gcp which covers more environments thanks to Google's contribution.

• Add support for dynamic configuration of the sampling rate of the default logger via the sampling_rate variable in Central Configuration. #1070

  While this is supported in the EDOT Node.js SDK, it will only be present in
  the "Agent Configuration" UI of Kibana version 9.3 and later.

Fixes

• Update to latest @opentelemetry/* dependencies (#1104), which includes a couple of fixes:
  • a bug in @opentelemetry/[email protected] prevented to set the user-agent header in metadata. A change in the exporter configuration fixed it.
  • the context manager and propagation setup is the first thing at the start of the SDK. This prevents detectors from leaking telemetry since they get the right context.

---

README | Full Release Notes | Breaking Changes

@elastic/mockopampserver 0.4.1

Changelog

• Fix an edge case where the server could crash on receiving a AgentToServer
  with an instance_uid that could not be stringified to a valid UUID.

• Bump the OpAMP protobuf definitions to v0.14.0 (from v0.12.0).
  https://github.com/open-telemetry/opamp-spec/blob/main/CHANGELOG.md#v0140

---

README | Full Changelog

@elastic/opentelemetry-node 1.5.0

Changelog

Chores

• Update to the latest upstream OpenTelemetry JS dependencies. (#1062)

• Switch to trusted publishing (https://docs.npmjs.com/trusted-publishers) as a security precaution against supply-chain attacks.

---

README | Full Release Notes | Breaking Changes

@elastic/opentelemetry-node 1.4.0

Changelog

Features and enhancements

• The Elastic-authored @elastic/opentelemetry-instrumentation-openai
  instrumentation has been upstreamed to OpenTelemetry.
  EDOT Node.js now uses the @opentelemetry/instrumentation-openai package
  to instrument openai. The newer package supports instrumenting openai@5 --
  the current major version.
  #1015

  • If you were using @elastic/opentelemetry-instrumentation-openai with the
    OTEL_NODE_ENABLED_INSTRUMENTATIONS or OTEL_NODE_ENABLED_INSTRUMENTATIONS
    environment variables, you should switch to using openai. The old value
    will still work until the next major release.

Fixes

• Update to latest @opentelemetry/* dependencies (#1027), which includes a fix for a bug in @opentelemetry/[email protected] that could break bundling.

---

README | Full Release Notes | Breaking Changes

@elastic/opentelemetry-node 1.3.0

Changelog

Features and enhancements

• Added @opentelemetry/instrumentation-oracledb to the default set of instrumentations.

• New Central Configuration settings. Typically these settings are only useful
  for temporary debugging of telemetry.

  • send_traces: A boolean to disable/enable sending of trace telemetry (i.e. spans).
  • send_metrics: The same, for the metrics signal.
  • send_logs: The same, for the logs signal.

  While these are supported in EDOT Node.js, they will only be present in
  "Agent Configuration" UI of Kibana version 9.2 and later.
  #928

• New ELASTIC_OTEL_CONTEXT_PROPAGATION_ONLY configuration environment variable.
  Set this to true to disable sending of spans, but otherwise continue to
  do context propagation. This can be useful in limited conditions to support
  propagating trace-context through a service to downstream services for
  distributed tracing, but not collect spans from the service. (Note that this
  typically results in incomplete or broken traces in Kibana trace viewer.)
  #928

Fixes

Chores

• OTLP export requests (HTTP flavors only) will include an identifier for EDOT Node.js in the User-Agent header.
  #982

---

README | Full Release Notes | Breaking Changes

@elastic/mockotlpserver 0.10.0

Changelog

• feat: Add a new onRequest callback option to MockOTLPServer that allows
  capturing some info for each incoming request. The callback will be called
  for each incoming OTLP request. The callback data will include:

    {
        transport: <string>,    // 'http' or 'grpc'
        method?: <string>,      // HTTP method if transport is 'http'
        path: <string>,         // HTTP or gRPC request path
        headers: <map>,         // HTTP headers if transport is 'http'
        metadata: <Metadata>,   // gRPC headers if transport is 'grpc'
    }
  

---

README | Full Changelog