Changes

A reverse-chronological feed of everything that changed in the cluster, with a time-window selector and a diff viewer on each entry.

What it is for

Most incidents are preceded by a change. When something starts failing, this tab answers “what happened just before?” without asking you to reconstruct it from three other systems.

Each entry expands into a structured diff — the object before, the object after, and the fields that actually differ. Not a commit message, and not a summary of the change: the change itself.

Windows

The feed is windowed (last 24 hours, and wider) and paginated in blocks of 30. Narrow the window first — a cluster of any size generates a lot of legitimate change.

Correlating with an incident

The natural path is Risk or Reliability → a degraded workload → its change history. A workload’s own change history is also reachable directly from its drawer.

From the CLI

rtz changes --since 24              # cluster-wide, newest first
rtz changes payments/checkout       # one workload's history