Back to posts

Grafana Alloy Migration — Replace Promtail with Alloy

Read the full guide on docs.beyondyou.my.id
observabilitygrafanaalloypromtaillokikubernetes

Grafana Alloy is the unified telemetry collector that replaces Promtail, Tempo, and Mimir agents. This guide covers a practical migration path from Promtail to Alloy in Kubernetes, including config conversion and validation.

The migration is straightforward but has pitfalls — Promtail pipeline stages don’t always map 1:1 to Alloy components, and log labeling differences can cause silent data loss. This guide covers the common traps.

Key Takeaways

  • Alloy is a single binary replacing Promtail, Tempo Agent, and Mimir Agent
  • Promtail pipeline stages map directly to Alloy’s loki.process component
  • Rollback is simple — keep Promtail DaemonSet disabled but ready during validation

What’s Covered

SectionDescription
Config MigrationConvert Promtail YAML pipelines to Alloy River config syntax
DeploymentReplace Promtail DaemonSet with Alloy DaemonSet in Kubernetes
Pipeline MappingStage-by-stage conversion: labels, regex, drops, and transformations
ValidationVerify log labels, pipeline behavior, and Loki ingestion correctness
ObservabilityMonitor Alloy itself via Prometheus metrics and self-scraping
RollbackStep-by-step procedure to revert to Promtail if issues arise

Migration Phases

PhaseAction
1Deploy Alloy alongside Promtail (dual-write)
2Validate log labels and pipeline output in Loki
3Switch traffic to Alloy and disable Promtail
4Remove Promtail DaemonSet after validation period

Read the full guide: Grafana Alloy Migration — Replace Promtail with Alloy → — includes Promtail-to-Alloy config mapping, Helm values, and production rollout strategy.