ci(#606): Gitea-compatible trx upload + deterministic fixes for the CI-load test flakes

- v2-ci.yml: upload-artifact@v4 -> @v3 + continue-on-error (act_runner reports as GHES and
  v4 refuses GHES, failing every run); runner-capacity diagnostic step; AKKA_TEST_TIMEFACTOR=3
  on the test step.
- Runtime/ControlPlane harnesses: akka.test.timefactor from AKKA_TEST_TIMEFACTOR
  (TestTimeFactor), single-expect-default 15s; ControlPlane gains PresenceBudget.
- Leaked spinner threads: the infinite-loop timeout tests (Calculation, Host.IntegrationTests
  RoslynVirtualTagEvaluator, 3x Core.Scripting) pinned a core for the rest of the test process
  (abandoned, never stopped). Now a bounded 5 s wall-clock spin; outer compile-inclusive bound 30 s.
- Real test races fixed: ScriptedAlarmEngine push-during-drain (Inactive is also the start
  state + persist-before-update => wait for the Cleared save); DriverReconnectE2e (both nodes
  host the driver; Created[] is last-writer-wins and ForceReconnect is a no-op while
  Connecting); ControlPlaneStarterOrder (lost-race heal waited for a 60 s refresh tick).
- Fixed sleeps -> poll-for-condition: ReconnectSupervisor LastError, Modbus deadband steps
  (wait for observed reads), Modbus dispose hygiene precondition, FOCAS probe-loop log.
- Short presence budgets -> PresenceBudget in the Runtime/ControlPlane tests seen failing on
  Gitea; Calculation Settle 10 -> 30 s. No assertion removed, nothing skipped.
- infra/README.md §10 CI; stale TreatWarningsAsErrors comment in Directory.Build.props.

Claude-Session: https://claude.ai/code/session_01Ef2ozxS7QPanGT8uoATbTe
This commit is contained in:
Joseph Doherty
2026-10-08 19:44:31 -04:00
parent 8801631739
commit 55ae230fa4
36 changed files with 449 additions and 128 deletions
@@ -1183,6 +1183,23 @@ public sealed class ScriptedAlarmEngineTests
throw new TimeoutException("Condition did not become true in time");
}
/// <summary>
/// Waits for a blocked-Activated-save scenario to finish its follow-up pass: the Cleared transition
/// has been persisted (the second SaveAsync) AND applied in memory.
/// </summary>
/// <remarks>
/// Waiting for <c>Active == Inactive</c> alone is ambiguous here — Inactive is also the state the
/// alarm STARTED in, and the engine persists BEFORE it updates memory, so in the instant after the
/// gate is released the in-memory state still reads the pre-activation Inactive. The wait then
/// returned before the follow-up pass ran and the evaluation count read 1 (#606, a Gitea CI failure
/// in 137 ms). <see cref="CountingAlarmStateStore.SaveCount"/> counts at SaveAsync entry, so
/// "two saves started AND Inactive in memory" is only true after the Cleared update.
/// </remarks>
private static Task WaitForClearedFollowUpAsync(ScriptedAlarmEngine eng, CountingAlarmStateStore store) =>
WaitForAsync(
() => store.SaveCount >= 2 && eng.GetState("A")!.Active == AlarmActiveState.Inactive,
timeoutMs: 3000);
// -------------------------------------------------------------------------
// Test helpers
// -------------------------------------------------------------------------
@@ -1486,7 +1503,7 @@ public sealed class ScriptedAlarmEngineTests
up.Push("Temp", 50);
saveGate.SetResult();
await WaitForAsync(() => eng.GetState("A")!.Active == AlarmActiveState.Inactive, timeoutMs: 3000);
await WaitForClearedFollowUpAsync(eng, store);
probe.Count.ShouldBeLessThanOrEqualTo(2,
"11 pushes over one blocked evaluation must produce the in-flight evaluation plus at " +
@@ -1529,7 +1546,9 @@ public sealed class ScriptedAlarmEngineTests
await eng.LoadAsync([
Alarm("A", """ctx.Logger.Information("EVALPROBE"); return (int)ctx.GetTag("Temp").Value > 100;"""),
], TestContext.Current.CancellationToken);
// LoadAsync evaluates AND saves once (startup recovery) — count both from here.
probe.Reset();
store.ResetCount();
var saveGate = new TaskCompletionSource();
store.BlockNextSave = saveGate;
@@ -1542,7 +1561,7 @@ public sealed class ScriptedAlarmEngineTests
up.Push("Temp", 40);
saveGate.SetResult();
await WaitForAsync(() => eng.GetState("A")!.Active == AlarmActiveState.Inactive, timeoutMs: 3000);
await WaitForClearedFollowUpAsync(eng, store);
probe.Count.ShouldBe(2, "the in-flight pass plus exactly one follow-up pass for the late push");
}