Migration guide

From a self-hosted observability stack

Teams running their own combination of TSDB + log store + tracing backend + visualizer + alertmanager.

Estimated time: ~1–2 weeks for a mid-size team; one weekend for a small one

Why teams switch

  • Five services to upgrade independently — one binary instead
  • On-call paged for the monitoring stack itself, not the product
  • Storage rebalancing and retention math is somebody’s job
  • Auth, SSO, multi-tenancy bolted on by hand

What transfers as-is

  • PromQL expressions — paste them in as-is
  • Existing scrape configs via a thin shim agent
  • Dashboards via JSON import (panels map 1:1 for the common cases)
  • Alert rules — the YAML format is compatible

What to adapt

  • Recording rules — Unimoni stores them in the same DB; same syntax
  • Long-tail dashboard plugins may not have a 1:1 widget — file an issue, we’ll add

Step by step

  • 1.Stand up Unimoni alongside the existing stack (one binary, docker-compose)
  • 2.Point one agent at it; verify the same series shows up
  • 3.Import dashboards via /api/v1/dashboards/import
  • 4.Cut over a non-critical team for a week — keep the old stack live
  • 5.When confidence is there, point all agents and decommission