Trusence Every claim has a source
Last updated 15 October 2026 Search Türkçe
← All stories
Infrastructure

Grafana Tempo 3.1 adds query-based trace redaction and sampling-aware metrics

The update makes secure Kafka setups easier and gives operators a faster way to redact traces without listing IDs.

Grafana Labs has shipped Tempo 3.1, which extends the Kafka-based 3.0 architecture with practical fixes for secured brokers and new controls for trace redaction and TraceQL metrics. The release adds TLS, mTLS, several SASL options, rack-aware reads, and configurable producer compression, which makes Kafka backends such as Azure Event Hubs workable and can cut cross-zone traffic. Tempo 3.1 also lets operators redact data with a TraceQL query, preview the match count first, and narrow the affected time window, though older workers will still redact every matching span if they do not understand the window. For analytics, it introduces an experimental sampling-aware hint for TraceQL metrics, arithmetic between query results and scalars, and a faster span-only fetch path that is now enabled by default for vParquet5 blocks. Grafana also added trace diffing, span pruning on trace-by-ID reads, and a few metrics-generator improvements aimed at lower allocation and better OpenTelemetry compatibility.

Why it matters

For teams running Tempo at scale, the release changes how they handle secured Kafka backends and sensitive trace data. It also gives them new ways to query traces and metrics with less manual setup, while the default span-only fetch path should reduce the work needed for vParquet5 reads. The new diff and pruning endpoints add more focused trace inspection and cleanup options for trace-by-ID workflows.

Sources

  • InfoQ