Exercise a real one-target ManagedClientGroup through timeout, Draining, a gated retained reply, LateReply classification, local close, and authoritative TargetFinished publication. Synchronize release after EchoLoss and a scheduler-handled no-op update so the ordering is deterministic rather than delay-dependent.
Add a silent-server control that verifies completion only after the final-drain deadline, and assert bounded join, queued-event draining through disconnection, target counters, terminal outcome equality, and group aggregates in both cases.
Harden the analogous single-client late-drain test with a viable one-probe run window and packet-shape checks. No production final-drain behavior change was required.
Add a scheduler-owned saturating peer-close counter shared with the group session. Update it before best-effort TargetFinished publication so bounded subscriber overflow cannot hide the live shutdown condition, while retaining the existing events, bounded history, and aggregate accounting.
Make grouped CLI and TUI supervision consult the durable count on each existing activity loop and funnel both count and event observations through one-shot stop policy. Preserve interruption suppression, queued-event draining, structured failures, drop reporting, and peer-close diagnostic precedence.
Cover scheduler-only accounting, non-peer exclusions, receiver and removal races, deterministic capacity-one DropOldest eviction with an active target, frontend one-shot policy, and the existing process-level peer-close and interruption behavior.
After publishing the authenticated reply classification and session-close events, mark the target PeerClosed while the receiver still holds its lock. Wake the scheduler to publish and account for the single authoritative terminal outcome.
Add controlled-order regressions proving later scheduler sends and target removal cannot overwrite peer-close provenance, counters remain intact, and ordinary replies stay active.
Nils Andreas Svee