Skip to content

SYSTEM Cited by 1 source

Slack Peekaboo

What it is

Peekaboo is Slack's real-time EC2 fleet inventory system, built as part of Shipyard. It replaces Chef Server as the source of truth for fleet state by tapping directly into cloud events and instance metadata, providing near-real-time visibility across environments.

Architecture

Built on three AWS primitives:

  • AWS EventBridge — consumes EC2 lifecycle events (launch, terminate, state changes).
  • OpenSearch — stores and indexes instance metadata for fast querying.
  • Lambda — event processing and transformation layer.

Capabilities

  • UI to explore the fleet visually.
  • API for programmatic integrations (used by The Reaper for age tracking).
  • CLI for quick command-line fleet checks.

Peekaboo tracks instances from both Shipyard and non-Shipyard deployments, providing a complete view of the entire EC2 fleet in one place.

Why not Chef Server?

The legacy model relied on Chef Server as the fleet inventory, but this had limitations: Chef convergence lag meant inventory was stale, and Chef-managed instances were the only ones tracked. Peekaboo's event-driven approach provides better telemetry, broader coverage, and fresher state.

Caveats

  • No specific latency numbers for "near real-time" freshness.
  • No fleet-scale numbers (total instances tracked).
  • API surface and data model not disclosed.
  • Retention and lifecycle of historical instance records not specified.

Seen in

Last updated · 585 distilled / 1,765 read