Description
The pfTop documentation and dashboard describe a 611-series hard ceiling, but the bounded inventories cap only one current scrape. After the five-minute expiry, a disjoint top set can replace every address and port label, so retained Prometheus storage can accumulate many more distinct series and endpoint-identifying values. The opt-in feature remains bounded in process, but the published capacity and privacy contract is incomplete.
Acceptance Criteria
- #1 Documentation and dashboard wording distinguish the simultaneous active-series ceiling from historical distinct-series churn over a retention window
- #2 The opt-in warning names source, destination, gateway and port labels as endpoint-identifying telemetry and gives operators a concrete retention/cardinality planning caution
- #3 No claim is made about a live deployment or retention volume that was not measured
Definition of Done
- #1 just check
- #2 just gen (if any generated artifact changed) and the diff committed
Implementation Plan
Correct the hand-written pfTop capacity/privacy prose and the dashboard panel text without changing the bounded-inventory behavior, regenerate dashboard artifacts once, and validate the source plus generated forms.
Final Summary
Fixed by afda1a4c, with the regenerated dashboard line landing in b360e86b. The docs and dashboard now distinguish simultaneous active series from retained identity churn and name the endpoint-identifying labels and retention implications without claiming a measured live volume. just dashboard reported coverage 1087/1087 and log streams 10/10; just docs-check and integrated just check passed. Tests and CodeRabbit were intentionally skipped for prose/declarative dashboard wording.