Feature · Collector

Self-hostable collector for on-network capture.

Single Go binary inside your network. Local batching, on-disk buffering, redaction. Fans out to Trefur and any backends you already run, in parallel.

How it works.

The trefur collector is one Go binary. Run it on a host inside your network, in a Kubernetes sidecar, or as a systemd unit on a VM. Your SDKs point at it with the standard OTLP endpoints — /v1/traces, /v1/metrics, /v1/logs.

The collector batches locally, buffers to disk when the upstream drops, and replays cleanly when the network comes back. Redaction runs in the collector before egress — drop fields, mask substrings, hash identifiers — so sensitive content never leaves your perimeter unredacted.

The collector is fan-out friendly. Send traces to Trefur and any other OTLP backend in parallel. Keep your existing pipeline; add agent-aware observability without ripping anything out.

What you can do with it.

Single Go binary. No dependencies. Runs on Linux, macOS, Windows, ARM, x86.
Receives OTLP/HTTP and OTLP/gRPC at /v1/traces, /v1/metrics, /v1/logs.
Local batching with configurable size and flush interval.
On-disk buffering keeps data when the network drops; replays when it comes back.
Configurable redaction — drop fields, regex-mask values, hash identifiers before egress.
Fans out to Trefur and any other OTLP backend in parallel, including Datadog, Honeycomb, Grafana.

A minimal collector config.

trefur-collector.yaml
receivers:
otlp:
protocols: [http, grpc]
processors:
batch:
timeout: 2s
redact:
drop_fields: ['user.email', 'user.phone']
mask_regex: ['(?i)api[_-]?key=\w+']
exporters:
trefur: { endpoint: https://ingest.trefur.com }
datadog: { endpoint: https://otlp.datadoghq.com }
pipelines:
traces: receivers=[otlp] -> processors=[redact, batch] -> exporters=[trefur, datadog]

Runs on.

Linux
macOS
Windows
Docker
Kubernetes
systemd
OTLP/HTTP
OTLP/gRPC

Capture on your network. Send only what you mean to.

Free tier. The collector binary is free to download and run.