fix(alarms): re-assert non-normal scripted-alarm conditions after a (re)load (#487)

A deploy that triggers a FULL address-space rebuild clears `_alarmConditions`
and re-materialises every condition node fresh — inactive, acked, confirmed,
unshelved. The engine reloads its persisted state and re-derives Active from the
predicate, so it ends up correct; but there is no transition to report, so
`LoadAsync` yields `EmissionKind.None` and `OnEngineEmission` drops it. Nothing
writes the node.

Live-reproduced on the docker-dev rig against the pre-fix image: an alarm that
had fired and cleared but was still UNACKNOWLEDGED came back from a rebuild
reporting Acknowledged with Retain=false — so it disappeared from
ConditionRefresh entirely, while the engine still held Unacknowledged. An
operator's outstanding alarm silently vanishes from the alarm list on deploy.

`ScriptedAlarmHostActor.ReassertConditionNodes` closes it. After each load it
reads the new `ScriptedAlarmEngine.GetProjections()` and sends one
`AlarmStateUpdate` per alarm NOT in the Part 9 no-event position.

Load-bearing properties:

- Node-only. It sends `AlarmStateUpdate` and nothing else — never the `alerts`
  topic, never the telemetry hub. `alerts` is the historization path, so a row
  per alarm per deploy would append a duplicate historian/AVEVA record every
  time anyone deploys. This is why it reads a projection rather than re-emitting:
  `ScriptedAlarmProjection` carries no `EmissionKind`, so there is nothing for
  the emission machinery — or a future refactor — to mistake for a transition.
  The issue's sketch said `GetState(alarmId)` + `LoadedAlarmIds`; that alone
  would have clobbered the node's severity and message, so the projection also
  carries the definition's severity, the resolved message template, and the
  last-observed worst-input quality.
- Only non-normal alarms. One in the no-event position already matches what the
  materialise built, so a never-fired alarm stays byte-identical to a deploy
  without this pass (including its Message, which the materialise seeds with the
  display name), and an all-normal deploy writes nothing at all.
- Timestamp is the condition's own `LastTransitionUtc`, not the deploy instant.
- No duplicate Part 9 events: `WriteAlarmCondition`'s delta-gate compares against
  the node's CURRENT state, so a re-assert onto a freshly materialised node is a
  genuine delta (fires once — correct, the rebuild reset what clients see) while
  one onto a surgically-preserved node is suppressed.

Tests: 6 new host tests + 6 new engine tests. Falsifiability checked by
disabling the call — 4 of the 6 host tests go red; the other 2 are absence
guards against an over-broad fix and pass either way by design.

Live gate (docker-dev, central-2, pre-fix image vs. rebuilt image):
- pre-fix: condition absent from ConditionRefresh after a rebuild;
- post-fix: present, Unacknowledged, Retain=True, stamped with its own
  LastTransitionUtc rather than the restart instant;
- /alerts stayed empty across two re-asserts, with the panel proven live by a
  real ACTIVATED transition immediately afterwards — no duplicate history.
This commit is contained in:
Joseph Doherty
2026-07-26 09:06:39 -04:00
parent 28c2866710
commit 549c656489
6 changed files with 575 additions and 14 deletions
@@ -334,6 +334,48 @@ public sealed class ScriptedAlarmEngine : IDisposable
public IReadOnlyCollection<AlarmConditionState> GetAllStates()
=> _alarms.Values.Select(a => a.Condition).ToArray();
/// <summary>
/// #487 — the currently-held condition of every loaded alarm, packaged with the definition's
/// severity, the rendered message template and the last-observed input quality, i.e. everything
/// a caller needs to <b>project</b> the alarm onto an OPC UA condition node without waiting for
/// the next transition.
/// </summary>
/// <remarks>
/// <para>
/// <b>This is a read, not an emission.</b> It deliberately returns a distinct type rather
/// than a <see cref="ScriptedAlarmEvent"/>, and it never touches <see cref="OnEvent"/>. A
/// still-active alarm produces <see cref="EmissionKind.None"/> at load (there is no
/// transition to report), so the transition stream cannot carry the current state — that is
/// exactly why the caller needs this. Routing these through the emission path instead would
/// append a duplicate historian / <c>alerts</c> row on every deploy, so the shape here is
/// intentionally one the emission machinery cannot consume.
/// </para>
/// <para>
/// <b>Synchronization.</b> Reads <c>_alarms</c> and <c>_valueCache</c> — both
/// <see cref="ConcurrentDictionary{TKey, TValue}"/> — without taking <c>_evalGate</c>, the
/// same posture as <see cref="GetState"/> / <see cref="GetAllStates"/>. Each projection is
/// individually coherent; the set as a whole is not an atomic snapshot across a concurrent
/// re-evaluation. That is sufficient for the node-projection use: an evaluation racing this
/// read publishes its own transition through the normal path.
/// </para>
/// </remarks>
/// <returns>One projection per loaded alarm; empty before the first <see cref="LoadAsync"/>.</returns>
public IReadOnlyList<ScriptedAlarmProjection> GetProjections()
{
var projections = new List<ScriptedAlarmProjection>(_alarms.Count);
foreach (var (alarmId, state) in _alarms)
{
projections.Add(new ScriptedAlarmProjection(
AlarmId: alarmId,
Severity: state.Definition.Severity,
Message: MessageTemplate.Resolve(state.Definition.MessageTemplate, TryLookup),
Condition: state.Condition,
WorstInputStatusCode: LastWorstStatus(alarmId)));
}
return projections;
}
/// <summary>Acknowledges the specified alarm on behalf of the given user.</summary>
/// <param name="alarmId">The alarm identifier.</param>
/// <param name="user">The user performing the acknowledgment.</param>
@@ -974,6 +1016,26 @@ public sealed record ScriptedAlarmEvent(
// the top-2 severity bits. Default 0u == Good keeps every existing constructor call unchanged.
uint WorstInputStatusCode = 0u);
/// <summary>
/// #487 — a loaded alarm's current condition, ready to be projected onto its OPC UA node.
/// Produced by <see cref="ScriptedAlarmEngine.GetProjections"/> on demand; it is <b>not</b> an
/// emission and never travels the <see cref="ScriptedAlarmEngine.OnEvent"/> path, so it can never
/// become an <c>alerts</c> row or a historian record. Deliberately a separate type from
/// <see cref="ScriptedAlarmEvent"/> — it carries no <see cref="EmissionKind"/>, so there is nothing
/// for a future refactor to mistake for a transition.
/// </summary>
/// <param name="AlarmId">The alarm identifier (also its OPC UA condition NodeId).</param>
/// <param name="Severity">The definition's severity bucket.</param>
/// <param name="Message">The message template resolved against the current value cache.</param>
/// <param name="Condition">The Part 9 condition state the engine currently holds.</param>
/// <param name="WorstInputStatusCode">The last-observed worst OPC UA StatusCode across the alarm's inputs.</param>
public sealed record ScriptedAlarmProjection(
string AlarmId,
AlarmSeverity Severity,
string Message,
AlarmConditionState Condition,
uint WorstInputStatusCode);
/// <summary>
/// Upstream source abstraction — intentionally identical shape to the virtual-tag
/// engine's so Stream G can compose them behind one driver bridge.
@@ -288,8 +288,101 @@ public sealed class ScriptedAlarmHostActor : ReceiveActor
// so a re-Apply with a fresh union simply supersedes the old one — no explicit unregister needed.
_mux?.Tell(new DependencyMuxActor.RegisterInterest(msg.DepRefs, Self));
_log.Debug("ScriptedAlarmHost: loaded; registered mux interest for {Count} dep refs", msg.DepRefs.Count);
ReassertConditionNodes();
}
/// <summary>
/// #487 — re-project every loaded alarm that is NOT in the Part 9 "no-event" position back onto
/// its condition node, right after a (re)load.
///
/// <para>
/// <b>The bug this closes.</b> A deploy that triggers a FULL address-space rebuild clears
/// <c>_alarmConditions</c> and re-materialises every condition node fresh — inactive, acked,
/// confirmed, unshelved. The engine, meanwhile, restores each alarm's persisted state and
/// re-derives Active from the predicate, so a still-active alarm ends the reload
/// <i>correctly Active in the engine</i> but produces <see cref="EmissionKind.None"/> (there
/// is no transition to report) — and <see cref="OnEngineEmission"/> drops None. Nothing then
/// writes the node, so an active alarm with static dependencies reads NORMAL to every OPC UA
/// client until its next real transition, which may never come. Same shape as the VirtualTag
/// redeploy-reset the <c>ReassertValue</c> fix closed.
/// </para>
///
/// <para>
/// <b>Node-only, by construction.</b> This sends <see cref="OpcUaPublishActor.AlarmStateUpdate"/>
/// and nothing else — it never publishes to the <c>alerts</c> topic and never touches the
/// telemetry hub. That is the whole reason it reads <see cref="ScriptedAlarmEngine.GetProjections"/>
/// (a plain read returning <see cref="ScriptedAlarmProjection"/>) rather than re-emitting: an
/// <c>alerts</c> row per alarm per deploy would append a duplicate historian / AVEVA record
/// every time anyone deploys — the alarm analogue of the VirtualTag M1 historian issue.
/// </para>
///
/// <para>
/// <b>Ordering.</b> <c>DriverHostActor</c> Tells <c>RebuildAddressSpace</c> to the publish
/// actor BEFORE it Tells <see cref="ApplyScriptedAlarms"/> here, and this runs later still —
/// after an <c>await</c>ed <see cref="ScriptedAlarmEngine.LoadAsync"/> pipes back. Both Tells
/// enqueue synchronously into the publish actor's local mailbox, so the re-materialise is
/// already queued ahead of these writes: the node exists by the time they are handled.
/// </para>
///
/// <para>
/// <b>Why only non-normal alarms.</b> An alarm sitting in the no-event position already
/// matches what the materialise just built, so writing it would be a pure no-op on state —
/// but it would still overwrite the node's Message (the materialise seeds it with the alarm's
/// display name; a projection carries the resolved template). Skipping them keeps a
/// never-fired alarm byte-identical to a deploy without this pass, and keeps the write count
/// at zero on the overwhelmingly common all-normal deploy.
/// </para>
///
/// <para>
/// <b>Duplicate events.</b> None. <c>WriteAlarmCondition</c>'s delta-gate compares the
/// incoming snapshot against the node's CURRENT state, so a re-assert onto a freshly
/// materialised (normal) node is a genuine delta and fires one Part 9 event — correct, since
/// the rebuild reset what clients see — while a re-assert onto a surgically-preserved node
/// that already holds the same state is suppressed. This write is ungated by redundancy role,
/// like every other node write here: the Secondary keeps its address space warm for failover.
/// </para>
/// </summary>
private void ReassertConditionNodes()
{
var reasserted = 0;
foreach (var projection in _engine.GetProjections())
{
if (IsInNoEventPosition(projection.Condition))
{
continue;
}
_publishActor.Tell(new OpcUaPublishActor.AlarmStateUpdate(
AlarmNodeId: projection.AlarmId,
State: ToSnapshot(projection),
// The instant the state we are re-asserting actually came about — NOT the deploy time.
// A client reading Time/ReceiveTime after a rebuild should still see when the alarm went
// active, not when the address space happened to be rebuilt.
TimestampUtc: projection.Condition.LastTransitionUtc,
Realm: AddressSpaceRealm.Uns));
reasserted++;
}
if (reasserted > 0)
{
_log.Info("ScriptedAlarmHost: re-asserted {Count} non-normal condition node(s) after (re)load", reasserted);
}
}
/// <summary>Whether <paramref name="condition"/> sits in the Part 9 "no-event" position — exactly the
/// state <see cref="AlarmConditionState.Fresh"/> produces and a freshly materialised condition node
/// already carries. Anything else (active, unacknowledged, unconfirmed, disabled, or shelved) is state
/// a rebuild would silently lose, so it is what <see cref="ReassertConditionNodes"/> re-projects.</summary>
/// <param name="condition">The engine's current condition state.</param>
/// <returns><c>true</c> when the condition carries nothing a rebuild could lose.</returns>
private static bool IsInNoEventPosition(AlarmConditionState condition)
=> condition.Enabled == AlarmEnabledState.Enabled
&& condition.Active == AlarmActiveState.Inactive
&& condition.Acked == AlarmAckedState.Acknowledged
&& condition.Confirmed == AlarmConfirmedState.Confirmed
&& condition.Shelving.Kind == ShelvingKind.Unshelved;
private void OnDependencyChanged(VirtualTagActor.DependencyValueChanged msg)
{
// Feed the live value into the upstream the engine subscribes from. #478 — carry the source
@@ -594,18 +687,38 @@ public sealed class ScriptedAlarmHostActor : ReceiveActor
/// Severity is the OPC UA 1..1000 value <see cref="SeverityToInt"/> derives from the coarse engine
/// bucket, cast to the <c>ushort</c> the SDK <c>SetSeverity</c> expects. Shelving's 3-way Core kind
/// maps 1:1 onto the Commons <see cref="AlarmShelvingKind"/>.</summary>
private static AlarmConditionSnapshot ToSnapshot(ScriptedAlarmEvent e) => new(
Active: e.Condition.Active == AlarmActiveState.Active,
Acknowledged: e.Condition.Acked == AlarmAckedState.Acknowledged,
Confirmed: e.Condition.Confirmed == AlarmConfirmedState.Confirmed,
Enabled: e.Condition.Enabled == AlarmEnabledState.Enabled,
Shelving: MapShelving(e.Condition.Shelving.Kind),
Severity: (ushort)SeverityToInt(e.Severity),
Message: e.Message,
private static AlarmConditionSnapshot ToSnapshot(ScriptedAlarmEvent e)
=> ToSnapshot(e.Condition, e.Severity, e.Message, e.WorstInputStatusCode);
/// <summary>#487 — the same projection for a <see cref="ScriptedAlarmProjection"/>: a state READ used
/// to restore a condition node after a rebuild, never an emission. Shares
/// <see cref="ToSnapshot(AlarmConditionState, AlarmSeverity, string, uint)"/> with the transition path
/// so a re-asserted node is byte-identical to what the alarm's own transition would have written.</summary>
/// <param name="p">The engine projection to map.</param>
/// <returns>The Commons snapshot the SDK sink projects onto the condition node.</returns>
private static AlarmConditionSnapshot ToSnapshot(ScriptedAlarmProjection p)
=> ToSnapshot(p.Condition, p.Severity, p.Message, p.WorstInputStatusCode);
/// <summary>The single Core → Commons condition projection, shared by the emission and re-assert
/// paths.</summary>
/// <param name="condition">The Part 9 condition state.</param>
/// <param name="severity">The engine's coarse severity bucket.</param>
/// <param name="message">The resolved condition message.</param>
/// <param name="worstInputStatusCode">The worst OPC UA StatusCode across the script's input tags.</param>
/// <returns>The Commons snapshot the SDK sink projects onto the condition node.</returns>
private static AlarmConditionSnapshot ToSnapshot(
AlarmConditionState condition, AlarmSeverity severity, string message, uint worstInputStatusCode) => new(
Active: condition.Active == AlarmActiveState.Active,
Acknowledged: condition.Acked == AlarmAckedState.Acknowledged,
Confirmed: condition.Confirmed == AlarmConfirmedState.Confirmed,
Enabled: condition.Enabled == AlarmEnabledState.Enabled,
Shelving: MapShelving(condition.Shelving.Kind),
Severity: (ushort)SeverityToInt(severity),
Message: message,
// #478 — the condition's Quality is the worst quality across the script's input tags at evaluation
// time (carried on the event by the engine). A transition fired while an input is Uncertain projects
// Uncertain here so the full-snapshot write doesn't clobber quality back to Good.
Quality: QualityFromStatus(e.WorstInputStatusCode));
Quality: QualityFromStatus(worstInputStatusCode));
/// <summary>Maps the Core <see cref="ShelvingKind"/> onto the Commons <see cref="AlarmShelvingKind"/>
/// mirror (the Commons assembly can't see the Core enum).</summary>