Adds CliRunner.ListAreaIdsByNamePrefixAsync so Playwright tests can locate
and delete areas they created through the UI without needing the id returned
from a CLI create call. Mirrors ListTemplateIdsByNamePrefixAsync. Round-trip
test verifies create → list-by-prefix → delete against the live cluster.
Also extends AddAttributeAsync with an optional dataSourceReference parameter
so the fixture attribute appears in both _bindingDataSourceAttrs (bindings UI)
and _overrideAttrs (overrides UI) on the InstanceConfigure page.
In Disable_Instance_ShowsOutcomeToast and Enable_Instance_ShowsOutcomeToast, the
precondition CLI calls (DeployInstanceAsync / DisableInstanceAsync) were between
CreateInstanceAsync and the try block. A throw there would skip the finally
DeleteInstanceAsync, leaking the instance. Moved those calls to be the first
statements inside try so cleanup always runs once the instance exists.
Adds CreateEditDelete_Site_RoundTrips [SkippableFact] to SiteCrudTests.
Exercises the full create → edit → delete UI flow against the live cluster,
with CliRunner best-effort teardown so no zztest-* sites leak on mid-test failure.
Guards against the Akka singleton Ask hang regression: asserts all nine KPI
tiles on /monitoring/health resolve to numeric values and never show the
em-dash degrade placeholder (—). Covers Notification Outbox, Audit, and
Site Call tile groups. Selector disambiguation: Outbox tiles are div.card,
Site Call tiles are button.card — prevents strict-mode collisions on the
shared "Stuck" and "Parked" labels.
Captures the 2026-06-05 coverage audit's gaps and the approved approach for
closing them: ephemeral CLI-provisioned fixtures with outcome-tolerant asserts
for the mutating suites (deploy lifecycle, retry/discard, transport import),
UI CRUD round-trips, nav render hardening, a Health KPI load test, and a
standardized skip-and-log policy. Next: writing-plans turns this into tasks.
Per-probe health-check child scopes were disposing the AddTransient-bridged
ActorSystem (IDisposable), terminating the live cluster node ~4s after boot and
leaving every singleton-proxy Ask to hang the full 30s QueryTimeout — the central
report pages (/notifications, /site-calls, /monitoring/health) loaded in ~30s.
Bridge it as a singleton via a new lazy AkkaHostedService.GetOrCreateActorSystem()
so child-scope disposal never touches it. Verified: 0 post-startup terminates,
healthy active/standby, report pages ~0.05s, Playwright 68 passed / 0 failed.
Joseph Doherty