Logger: OtelHttp

The OtelHttpExporter sends structured logs and spans to an OpenTelemetry-compatible collector via HTTP. It’s TriFrost’s production-grade exporter when you want real observability, including distributed tracing, performance timelines, and full-context logs.

It supports:

OTLP-over-HTTP export of logs and spans
Batch buffering, retry with backoff, and payload scrubbing
Compatible with tools like: SigNoz, Uptrace, Grafana Tempo, ...

The TriFrost website for example uses a combination of JsonExporter for Cloudflare observability and OTEL through UpTrace.

📥 Import and Use

import {App, OtelHttpExporter} from '@trifrost/core';

new App({
  ...
  tracing: {
    exporters: ({env}) => [
      new OtelHttpExporter({
        logEndpoint: env.OTEL_LOGS,
        spanEndpoint: env.OTEL_SPANS,
        headers: {
          Authorization: `Bearer ${env.OTEL_TOKEN}`
        }
      }),
    ],
  },
  ...
})

Note: 💡 By default, logs and spans are flushed at the end of each request — you do not need to manually call flush().

⚙️ Available Options

logEndpoint: string
OTEL collector HTTP endpoint for logs.
spanEndpoint: string
Optional. If not provided, logEndpoint is reused for spans.
headers: Record<string, string>
Optional headers to include in every OTLP request. (usually for authorization purposes)
maxBatchSize: number
How many logs/spans to batch before flushing.
default: 20
maxBufferSize: number
Maximum retained buffer size if transport fails.
default: 10 000
maxRetries: number
Number of retries for failed deliveries.
default: 3
omit: TriFrostLogScramblerValue[]
Fields to redact before export.
default: OMIT_PRESETS.default

👉 See Scrambling Hygiene for how redaction works

🚚 What gets sent?

Logs -> OTEL resourceLogs

Each ctx.logger.log(...) call is sent as an OTLP log record:

{
  "timeUnixNano": 1717689600000000000,
  "severityText": "INFO",
  "body": {"stringValue": "User logged in"},
  "attributes": [
    {"key": "trace_id", "value": {"stringValue": "abc..."}},
    {"key": "ctx.userId", "value": {"stringValue": "123"}},
    {"key": "data.status", "value": {"stringValue": "success"}}
  ]
}

Spans -> OTEL resourceSpans

Spans created via ctx.logger.span(...), decorators, or middleware tracing are serialized as full OTLP spans:

{
  "name": "fetchUser",
  "traceId": "...",
  "spanId": "...",
  "startTimeUnixNano": 1717689600000000000,
  "endTimeUnixNano": 1717689600123456000,
  "attributes": [
    {"key": "db.query", "value": {"stringValue": "SELECT * FROM users WHERE ..."}}
  ]
}

Retry and Buffering

Batches are retried up to maxRetries times on network failure
Retries use jittered exponential backoff
If all retries fail, logs/spans are requeued if within buffer limit, otherwise they're dropped to prevent memory pressure

Note: Failed sends log a console warning — you may wish to hook this in production.

Examples

Minimal

new OtelHttpExporter({
  logEndpoint: 'https://collector.mycompany.com/v1/logs'
});

With custom headers

new OtelHttpExporter({
  logEndpoint: 'https://otel.example.com/v1/logs',
  headers: {
    'X-Project-ID': 'abc123',
    'Authorization': `Bearer ${process.env.OTEL_TOKEN}`
  }
});

Custom retry/batch settings

new OtelHttpExporter({
  logEndpoint: 'https://myotel/logs',
  maxBatchSize: 50,
  maxBufferSize: 5000,
  maxRetries: 5
});

SigNoz (EU Region)

new OtelHttpExporter({
  logEndpoint: 'https://ingest.eu.signoz.cloud/v1/logs',
  spanEndpoint: 'https://ingest.eu.signoz.cloud/v1/traces',
  headers: {
    'signoz-access-token': env.SIGNOZ_API_TOKEN
  }
});

Note: You'll find this token in your SigNoz project settings under API Tokens. Make sure you match the ingest region (eu, us, etc).

👉 See SigNoz OTEL Setup for more info

Uptrace

new OtelHttpExporter({
  logEndpoint: 'https://otlp.uptrace.dev/v1/logs',
  spanEndpoint: 'https://otlp.uptrace.dev/v1/traces',
  headers: {
    'uptrace-dsn': env.UPTRACE_DSN
  }
});

Note: The uptrace-dsn is a secure connection string from your Uptrace project — it contains project ID, token, and endpoint info.

👉 See Uptrace DSN Guide for more info

Best Practices

✅ Use in production when full tracing and telemetry are required
✅ Tune maxBatchSize and maxBufferSize based on expected traffic
✅ Set meaningful redaction via omit
💡 You can use this exporter in tandem with other exporters such as JsonExporter
💡 Want to go fancy? Make use of TriFrost's isDevMode function to refine between dev and prod
❌ Don’t rely on it for logs in CI — prefer Console/Json exporters

TriFrost Docs