refactor: rename ScadaLink → ZB.MOM.WW.ScadaBridge (code + projects + namespaces)
Solution + 23 src projects + 26 test projects renamed; folders, csproj, namespaces, and ScadaLinkDbContext/ScadaBridgeDbContext class updated. ActorSystem "scadalink" → "scadabridge", Akka seed-node URLs migrated. SQL roles/logins, LDAP domains, CLI command name, and CLI config dir (~/.scadalink → ~/.scadabridge) also renamed. Build green; 5 Host.Tests fail awaiting SQL login rename in next commit. Pre-existing StaleTagMonitor timing flakes unchanged. Rename script committed at tools/rename-to-scadabridge.sh.
This commit is contained in:
Joseph Doherty
@@ -0,0 +1,673 @@
|
||||
using Akka.Actor;
|
||||
using Microsoft.CodeAnalysis.Scripting;
|
||||
using Microsoft.Extensions.Logging;
|
||||
using ZB.MOM.WW.ScadaBridge.Commons.Messages.Streaming;
|
||||
using ZB.MOM.WW.ScadaBridge.Commons.Types.Enums;
|
||||
using ZB.MOM.WW.ScadaBridge.Commons.Types.Flattening;
|
||||
using ZB.MOM.WW.ScadaBridge.HealthMonitoring;
|
||||
using ZB.MOM.WW.ScadaBridge.SiteRuntime.Scripts;
|
||||
using System.Globalization;
|
||||
using System.Text.Json;
|
||||
|
||||
namespace ZB.MOM.WW.ScadaBridge.SiteRuntime.Actors;
|
||||
|
||||
/// <summary>
|
||||
/// WP-16: Alarm Actor — coordinator actor, child of Instance Actor, peer to Script Actors.
|
||||
/// Subscribes to attribute change notifications from Instance Actor.
|
||||
///
|
||||
/// Evaluates alarm conditions:
|
||||
/// - ValueMatch: attribute equals a specific value
|
||||
/// - RangeViolation: attribute outside min/max range
|
||||
/// - RateOfChange: attribute rate exceeds threshold (configurable window, default per-second)
|
||||
///
|
||||
/// State (active/normal) is in memory only, NOT persisted.
|
||||
/// On restart: starts normal, re-evaluates from incoming values.
|
||||
///
|
||||
/// WP-21: AlarmExecutionActor CAN call Instance.CallScript() (ask to sibling Script Actor).
|
||||
/// Instance scripts CANNOT call alarm on-trigger scripts (no Instance.CallAlarmScript API).
|
||||
///
|
||||
/// Supervision: Resume on exception; AlarmExecutionActor stopped on exception.
|
||||
/// </summary>
|
||||
public class AlarmActor : ReceiveActor
|
||||
{
|
||||
private readonly string _alarmName;
|
||||
private readonly string _instanceName;
|
||||
private readonly IActorRef _instanceActor;
|
||||
private readonly SharedScriptLibrary _sharedScriptLibrary;
|
||||
private readonly SiteRuntimeOptions _options;
|
||||
private readonly ILogger _logger;
|
||||
private readonly ISiteHealthCollector? _healthCollector;
|
||||
|
||||
private AlarmState _currentState = AlarmState.Normal;
|
||||
/// <summary>
|
||||
/// Always <see cref="AlarmLevel.None"/> for binary trigger types. For
|
||||
/// <see cref="AlarmTriggerType.HiLo"/> this is the source of truth — the
|
||||
/// state machine transitions when the computed level changes.
|
||||
/// </summary>
|
||||
private AlarmLevel _currentLevel = AlarmLevel.None;
|
||||
private readonly AlarmTriggerType _triggerType;
|
||||
private readonly AlarmEvalConfig _evalConfig;
|
||||
private readonly int _priority;
|
||||
private readonly string? _onTriggerScriptName;
|
||||
private readonly Script<object?>? _onTriggerCompiledScript;
|
||||
|
||||
// Expression trigger: compiled expression + the attribute snapshot it
|
||||
// evaluates against. This field is the single home for the compiled
|
||||
// expression on the hot path.
|
||||
private readonly Script<object?>? _compiledTriggerExpression;
|
||||
private readonly Dictionary<string, object?> _attributeSnapshot = new();
|
||||
|
||||
/// <summary>
|
||||
/// SiteRuntime-017: the exact dictionary instance this actor was seeded from
|
||||
/// at construction. The Instance Actor must pass a private snapshot here, not
|
||||
/// its live <c>_attributes</c> field. Exposed for regression coverage of that
|
||||
/// isolation contract.
|
||||
/// </summary>
|
||||
internal IReadOnlyDictionary<string, object?>? SeedAttributesReference { get; }
|
||||
|
||||
// Rate of change tracking
|
||||
private readonly Queue<(DateTimeOffset Timestamp, double Value)> _rateOfChangeWindow = new();
|
||||
private readonly TimeSpan _rateOfChangeWindowDuration;
|
||||
|
||||
private int _executionCounter;
|
||||
|
||||
/// <summary>Initializes a new <see cref="AlarmActor"/> and configures message handlers for the alarm.</summary>
|
||||
/// <param name="alarmName">The canonical name of this alarm.</param>
|
||||
/// <param name="instanceName">The name of the owning instance.</param>
|
||||
/// <param name="instanceActor">Reference to the parent instance actor used for attribute access and script calls.</param>
|
||||
/// <param name="alarmConfig">The resolved alarm configuration including trigger type, priority, and script references.</param>
|
||||
/// <param name="onTriggerCompiledScript">Pre-compiled on-trigger script, or <c>null</c> if no script is defined.</param>
|
||||
/// <param name="sharedScriptLibrary">Shared script library providing common utilities to executed scripts.</param>
|
||||
/// <param name="options">Site runtime configuration options.</param>
|
||||
/// <param name="logger">Logger for alarm diagnostics.</param>
|
||||
/// <param name="compiledTriggerExpression">Pre-compiled trigger expression, or <c>null</c> for non-expression triggers.</param>
|
||||
/// <param name="initialAttributes">Seed attribute snapshot so static attributes evaluate correctly at startup.</param>
|
||||
/// <param name="healthCollector">Optional health collector for surfacing alarm execution metrics.</param>
|
||||
public AlarmActor(
|
||||
string alarmName,
|
||||
string instanceName,
|
||||
IActorRef instanceActor,
|
||||
ResolvedAlarm alarmConfig,
|
||||
Script<object?>? onTriggerCompiledScript,
|
||||
SharedScriptLibrary sharedScriptLibrary,
|
||||
SiteRuntimeOptions options,
|
||||
ILogger logger,
|
||||
Script<object?>? compiledTriggerExpression = null,
|
||||
IReadOnlyDictionary<string, object?>? initialAttributes = null,
|
||||
ISiteHealthCollector? healthCollector = null)
|
||||
{
|
||||
_alarmName = alarmName;
|
||||
_instanceName = instanceName;
|
||||
_instanceActor = instanceActor;
|
||||
_sharedScriptLibrary = sharedScriptLibrary;
|
||||
_options = options;
|
||||
_logger = logger;
|
||||
_healthCollector = healthCollector;
|
||||
_priority = alarmConfig.PriorityLevel;
|
||||
_onTriggerScriptName = alarmConfig.OnTriggerScriptCanonicalName;
|
||||
_onTriggerCompiledScript = onTriggerCompiledScript;
|
||||
_compiledTriggerExpression = compiledTriggerExpression;
|
||||
|
||||
// Seed the trigger-expression attribute snapshot from the instance's
|
||||
// initial attribute set so static attributes (which never re-emit an
|
||||
// AttributeValueChanged after deploy) evaluate correctly at startup.
|
||||
SeedAttributesReference = initialAttributes;
|
||||
if (initialAttributes != null)
|
||||
{
|
||||
foreach (var kvp in initialAttributes)
|
||||
_attributeSnapshot[kvp.Key] = kvp.Value;
|
||||
}
|
||||
|
||||
// Parse trigger type
|
||||
_triggerType = Enum.TryParse<AlarmTriggerType>(alarmConfig.TriggerType, true, out var tt)
|
||||
? tt : AlarmTriggerType.ValueMatch;
|
||||
|
||||
_evalConfig = ParseEvalConfig(alarmConfig.TriggerConfiguration);
|
||||
_rateOfChangeWindowDuration = _evalConfig is RateOfChangeEvalConfig roc
|
||||
? roc.WindowDuration
|
||||
: TimeSpan.FromSeconds(1);
|
||||
|
||||
// Handle attribute value changes
|
||||
Receive<AttributeValueChanged>(HandleAttributeValueChanged);
|
||||
|
||||
// Handle alarm execution completion
|
||||
Receive<AlarmExecutionCompleted>(_ =>
|
||||
_logger.LogDebug("Alarm {Alarm} execution completed on {Instance}", _alarmName, _instanceName));
|
||||
}
|
||||
|
||||
/// <inheritdoc />
|
||||
protected override void PreStart()
|
||||
{
|
||||
base.PreStart();
|
||||
_logger.LogInformation(
|
||||
"AlarmActor {Alarm} started on instance {Instance}, trigger={TriggerType}",
|
||||
_alarmName, _instanceName, _triggerType);
|
||||
}
|
||||
|
||||
/// <inheritdoc />
|
||||
protected override SupervisorStrategy SupervisorStrategy()
|
||||
{
|
||||
return new OneForOneStrategy(
|
||||
maxNrOfRetries: -1,
|
||||
withinTimeRange: TimeSpan.FromMinutes(1),
|
||||
decider: Decider.From(ex =>
|
||||
{
|
||||
_logger.LogWarning(ex,
|
||||
"AlarmExecutionActor for {Alarm} on {Instance} failed, stopping",
|
||||
_alarmName, _instanceName);
|
||||
return Directive.Stop;
|
||||
}));
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// Evaluates alarm condition on attribute change. Alarm evaluation errors are logged,
|
||||
/// actor continues (does not crash).
|
||||
/// </summary>
|
||||
private void HandleAttributeValueChanged(AttributeValueChanged changed)
|
||||
{
|
||||
// Expression triggers evaluate against a snapshot of every attribute,
|
||||
// not a single monitored attribute. Keep the snapshot current for every
|
||||
// change before the IsMonitoredAttribute gate (which does not apply).
|
||||
if (_triggerType == AlarmTriggerType.Expression)
|
||||
{
|
||||
_attributeSnapshot[changed.AttributeName] = changed.Value;
|
||||
}
|
||||
else if (!IsMonitoredAttribute(changed.AttributeName))
|
||||
{
|
||||
// Only evaluate if this change is for an attribute we're monitoring
|
||||
return;
|
||||
}
|
||||
|
||||
try
|
||||
{
|
||||
if (_triggerType == AlarmTriggerType.HiLo)
|
||||
{
|
||||
HandleHiLoTransition(EvaluateHiLo(changed.Value));
|
||||
return;
|
||||
}
|
||||
|
||||
var isTriggered = _triggerType switch
|
||||
{
|
||||
AlarmTriggerType.ValueMatch => EvaluateValueMatch(changed.Value),
|
||||
AlarmTriggerType.RangeViolation => EvaluateRangeViolation(changed.Value),
|
||||
AlarmTriggerType.RateOfChange => EvaluateRateOfChange(changed.Value, changed.Timestamp),
|
||||
AlarmTriggerType.Expression => EvaluateExpression(),
|
||||
_ => false
|
||||
};
|
||||
|
||||
if (isTriggered && _currentState == AlarmState.Normal)
|
||||
{
|
||||
// Transition: Normal → Active
|
||||
_currentState = AlarmState.Active;
|
||||
_logger.LogInformation(
|
||||
"Alarm {Alarm} ACTIVATED on instance {Instance}",
|
||||
_alarmName, _instanceName);
|
||||
|
||||
// Notify Instance Actor of alarm state change
|
||||
var alarmChanged = new AlarmStateChanged(
|
||||
_instanceName, _alarmName, AlarmState.Active, _priority, DateTimeOffset.UtcNow);
|
||||
_instanceActor.Tell(alarmChanged);
|
||||
|
||||
// Spawn AlarmExecutionActor if on-trigger script defined
|
||||
if (_onTriggerCompiledScript != null)
|
||||
{
|
||||
SpawnAlarmExecution(AlarmLevel.None, _priority, string.Empty);
|
||||
}
|
||||
}
|
||||
else if (!isTriggered && _currentState == AlarmState.Active)
|
||||
{
|
||||
// Transition: Active → Normal (no script on clear)
|
||||
_currentState = AlarmState.Normal;
|
||||
_logger.LogInformation(
|
||||
"Alarm {Alarm} CLEARED on instance {Instance}",
|
||||
_alarmName, _instanceName);
|
||||
|
||||
var alarmChanged = new AlarmStateChanged(
|
||||
_instanceName, _alarmName, AlarmState.Normal, _priority, DateTimeOffset.UtcNow);
|
||||
_instanceActor.Tell(alarmChanged);
|
||||
}
|
||||
}
|
||||
catch (Exception ex)
|
||||
{
|
||||
_healthCollector?.IncrementAlarmError();
|
||||
// Alarm evaluation errors logged, actor continues
|
||||
_logger.LogError(ex,
|
||||
"Alarm {Alarm} evaluation error on {Instance}",
|
||||
_alarmName, _instanceName);
|
||||
}
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// HiLo state machine: emit an AlarmStateChanged whenever the evaluated
|
||||
/// level changes. Spawns the on-trigger script only on the Normal→Active
|
||||
/// edge (i.e., when entering an alarm band from the normal band) — not on
|
||||
/// level escalations like Hi→HiHi or Low→LowLow.
|
||||
/// </summary>
|
||||
private void HandleHiLoTransition(AlarmLevel newLevel)
|
||||
{
|
||||
if (newLevel == _currentLevel) return;
|
||||
|
||||
var previousLevel = _currentLevel;
|
||||
_currentLevel = newLevel;
|
||||
_currentState = newLevel == AlarmLevel.None ? AlarmState.Normal : AlarmState.Active;
|
||||
var priority = LevelPriority(newLevel);
|
||||
var message = LevelMessage(newLevel);
|
||||
|
||||
_logger.LogInformation(
|
||||
"Alarm {Alarm} on {Instance} transitioned {Prev} → {New} (priority={Priority})",
|
||||
_alarmName, _instanceName, previousLevel, newLevel, priority);
|
||||
|
||||
var alarmChanged = new AlarmStateChanged(
|
||||
_instanceName, _alarmName, _currentState, priority, DateTimeOffset.UtcNow)
|
||||
{
|
||||
Level = newLevel,
|
||||
Message = message
|
||||
};
|
||||
_instanceActor.Tell(alarmChanged);
|
||||
|
||||
if (previousLevel == AlarmLevel.None
|
||||
&& newLevel != AlarmLevel.None
|
||||
&& _onTriggerCompiledScript != null)
|
||||
{
|
||||
SpawnAlarmExecution(newLevel, priority, message);
|
||||
}
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// Returns the per-setpoint priority for the given level. Falls back to
|
||||
/// the alarm-level <see cref="_priority"/> when the HiLo config did not
|
||||
/// override the priority for that band, or for <see cref="AlarmLevel.None"/>.
|
||||
/// </summary>
|
||||
private int LevelPriority(AlarmLevel level)
|
||||
{
|
||||
if (_evalConfig is not HiLoEvalConfig hiLo) return _priority;
|
||||
return level switch
|
||||
{
|
||||
AlarmLevel.LowLow => hiLo.LoLoPriority ?? _priority,
|
||||
AlarmLevel.Low => hiLo.LoPriority ?? _priority,
|
||||
AlarmLevel.High => hiLo.HiPriority ?? _priority,
|
||||
AlarmLevel.HighHigh => hiLo.HiHiPriority ?? _priority,
|
||||
_ => _priority
|
||||
};
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// Per-band operator message. Empty string when no message is configured
|
||||
/// for the band, or for non-HiLo trigger types, or for the None level
|
||||
/// (alarm clear).
|
||||
/// </summary>
|
||||
private string LevelMessage(AlarmLevel level)
|
||||
{
|
||||
if (_evalConfig is not HiLoEvalConfig hiLo) return string.Empty;
|
||||
return level switch
|
||||
{
|
||||
AlarmLevel.LowLow => hiLo.LoLoMessage ?? string.Empty,
|
||||
AlarmLevel.Low => hiLo.LoMessage ?? string.Empty,
|
||||
AlarmLevel.High => hiLo.HiMessage ?? string.Empty,
|
||||
AlarmLevel.HighHigh => hiLo.HiHiMessage ?? string.Empty,
|
||||
_ => string.Empty
|
||||
};
|
||||
}
|
||||
|
||||
private bool IsMonitoredAttribute(string attributeName)
|
||||
{
|
||||
return _evalConfig.MonitoredAttributeName == attributeName;
|
||||
}
|
||||
|
||||
private bool EvaluateValueMatch(object? value)
|
||||
{
|
||||
if (_evalConfig is not ValueMatchEvalConfig config) return false;
|
||||
if (config.MatchValue == null) return value == null;
|
||||
|
||||
var valueStr = value?.ToString() ?? "";
|
||||
|
||||
// Support "!=X" for not-equal matching
|
||||
if (config.MatchValue.StartsWith("!="))
|
||||
{
|
||||
var expected = config.MatchValue[2..];
|
||||
return !string.Equals(valueStr, expected, StringComparison.Ordinal);
|
||||
}
|
||||
|
||||
return string.Equals(valueStr, config.MatchValue, StringComparison.Ordinal);
|
||||
}
|
||||
|
||||
private bool EvaluateRangeViolation(object? value)
|
||||
{
|
||||
if (_evalConfig is not RangeViolationEvalConfig config) return false;
|
||||
if (value == null) return false;
|
||||
|
||||
try
|
||||
{
|
||||
// InvariantCulture so string attribute values parse consistently
|
||||
// regardless of host locale (SiteRuntime-023).
|
||||
var numericValue = Convert.ToDouble(value, CultureInfo.InvariantCulture);
|
||||
return numericValue < config.Min || numericValue > config.Max;
|
||||
}
|
||||
catch
|
||||
{
|
||||
return false;
|
||||
}
|
||||
}
|
||||
|
||||
private bool EvaluateRateOfChange(object? value, DateTimeOffset timestamp)
|
||||
{
|
||||
if (_evalConfig is not RateOfChangeEvalConfig config) return false;
|
||||
if (value == null) return false;
|
||||
|
||||
try
|
||||
{
|
||||
// InvariantCulture so string attribute values parse consistently
|
||||
// regardless of host locale (SiteRuntime-023).
|
||||
var numericValue = Convert.ToDouble(value, CultureInfo.InvariantCulture);
|
||||
|
||||
// Add to window
|
||||
_rateOfChangeWindow.Enqueue((timestamp, numericValue));
|
||||
|
||||
// Remove old entries outside the window
|
||||
var cutoff = timestamp - _rateOfChangeWindowDuration;
|
||||
while (_rateOfChangeWindow.Count > 0 && _rateOfChangeWindow.Peek().Timestamp < cutoff)
|
||||
{
|
||||
_rateOfChangeWindow.Dequeue();
|
||||
}
|
||||
|
||||
if (_rateOfChangeWindow.Count < 2) return false;
|
||||
|
||||
var oldest = _rateOfChangeWindow.Peek();
|
||||
var timeDelta = (timestamp - oldest.Timestamp).TotalSeconds;
|
||||
if (timeDelta <= 0) return false;
|
||||
|
||||
var signedRate = (numericValue - oldest.Value) / timeDelta;
|
||||
return config.Direction switch
|
||||
{
|
||||
RateOfChangeDirection.Rising => signedRate > config.ThresholdPerSecond,
|
||||
RateOfChangeDirection.Falling => -signedRate > config.ThresholdPerSecond,
|
||||
_ => Math.Abs(signedRate) > config.ThresholdPerSecond
|
||||
};
|
||||
}
|
||||
catch
|
||||
{
|
||||
return false;
|
||||
}
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// Evaluates the compiled trigger expression against the current attribute
|
||||
/// snapshot, returning the resulting bool. This bool feeds the existing
|
||||
/// binary Normal↔Active state path — the alarm is active while true. A
|
||||
/// throwing, non-bool, or timed-out expression is treated as false (logged
|
||||
/// as an alarm error) so that the state machine still runs — an Active
|
||||
/// alarm correctly clears if the expression starts throwing.
|
||||
/// </summary>
|
||||
private bool EvaluateExpression()
|
||||
{
|
||||
if (_compiledTriggerExpression == null) return false;
|
||||
|
||||
try
|
||||
{
|
||||
var globals = new TriggerExpressionGlobals(_attributeSnapshot);
|
||||
// Bound evaluation with a short timeout. The CancellationToken
|
||||
// covers cooperative/async cases; a pathological CPU-bound
|
||||
// expression is not fully interruptible. Acceptable because
|
||||
// trigger expressions are authored by trusted Design-role users
|
||||
// and are compile-checked pre-deployment.
|
||||
using var cts = new CancellationTokenSource(TimeSpan.FromSeconds(2));
|
||||
var state = _compiledTriggerExpression
|
||||
.RunAsync(globals, cancellationToken: cts.Token)
|
||||
.GetAwaiter().GetResult();
|
||||
return state.ReturnValue is bool b && b;
|
||||
}
|
||||
catch (Exception ex)
|
||||
{
|
||||
// OperationCanceledException (timeout) falls through here too,
|
||||
// and is correctly treated as false.
|
||||
_healthCollector?.IncrementAlarmError();
|
||||
_logger.LogError(ex,
|
||||
"Alarm {Alarm} trigger expression evaluation failed on {Instance}; treated as false",
|
||||
_alarmName, _instanceName);
|
||||
return false;
|
||||
}
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// HiLo level evaluator: returns the most-severe matching band for the
|
||||
/// given value. Severity order checked from highest to lowest so that a
|
||||
/// value at exactly Hi==HiHi resolves to HighHigh. Unset setpoints (null)
|
||||
/// are skipped, allowing partial configs (e.g., HighHigh only).
|
||||
///
|
||||
/// Hysteresis: when the alarm is already in a level whose threshold the
|
||||
/// value would re-cross from inside, the threshold is relaxed by the
|
||||
/// configured deadband. This prevents flapping at the boundary — once at
|
||||
/// HighHigh with HiHi=100 and hiHiDeadband=5, the alarm stays HighHigh
|
||||
/// until the value drops below 95.
|
||||
/// </summary>
|
||||
private AlarmLevel EvaluateHiLo(object? value)
|
||||
{
|
||||
if (_evalConfig is not HiLoEvalConfig config) return AlarmLevel.None;
|
||||
if (value == null) return _currentLevel;
|
||||
|
||||
double numericValue;
|
||||
// InvariantCulture so string attribute values parse consistently
|
||||
// regardless of host locale (SiteRuntime-023).
|
||||
try { numericValue = Convert.ToDouble(value, CultureInfo.InvariantCulture); }
|
||||
catch { return _currentLevel; }
|
||||
|
||||
// When the current level is at-or-above HighHigh, relax the HiHi exit.
|
||||
// Same for the other directions.
|
||||
var hiHiThreshold = config.HiHi;
|
||||
if (hiHiThreshold is { } hh && _currentLevel == AlarmLevel.HighHigh)
|
||||
hiHiThreshold = hh - Math.Max(0, config.HiHiDeadband ?? 0);
|
||||
|
||||
var hiThreshold = config.Hi;
|
||||
if (hiThreshold is { } h && (_currentLevel == AlarmLevel.High || _currentLevel == AlarmLevel.HighHigh))
|
||||
hiThreshold = h - Math.Max(0, config.HiDeadband ?? 0);
|
||||
|
||||
var loLoThreshold = config.LoLo;
|
||||
if (loLoThreshold is { } ll && _currentLevel == AlarmLevel.LowLow)
|
||||
loLoThreshold = ll + Math.Max(0, config.LoLoDeadband ?? 0);
|
||||
|
||||
var loThreshold = config.Lo;
|
||||
if (loThreshold is { } l && (_currentLevel == AlarmLevel.Low || _currentLevel == AlarmLevel.LowLow))
|
||||
loThreshold = l + Math.Max(0, config.LoDeadband ?? 0);
|
||||
|
||||
if (hiHiThreshold is { } effHiHi && numericValue >= effHiHi) return AlarmLevel.HighHigh;
|
||||
if (hiThreshold is { } effHi && numericValue >= effHi) return AlarmLevel.High;
|
||||
if (loLoThreshold is { } effLoLo && numericValue <= effLoLo) return AlarmLevel.LowLow;
|
||||
if (loThreshold is { } effLo && numericValue <= effLo) return AlarmLevel.Low;
|
||||
return AlarmLevel.None;
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// Spawns an AlarmExecutionActor to run the on-trigger script.
|
||||
/// Passes the firing alarm's level/priority/message so the script can
|
||||
/// branch on severity via the <c>Alarm</c> global.
|
||||
/// </summary>
|
||||
private void SpawnAlarmExecution(AlarmLevel level, int priority, string message)
|
||||
{
|
||||
if (_onTriggerCompiledScript == null) return;
|
||||
|
||||
var executionId = $"{_alarmName}-alarm-exec-{_executionCounter++}";
|
||||
|
||||
// SiteRuntime-009: the on-trigger script body runs on the dedicated
|
||||
// ScriptExecutionScheduler, not the shared .NET thread pool.
|
||||
var props = Props.Create(() => new AlarmExecutionActor(
|
||||
_alarmName,
|
||||
_instanceName,
|
||||
level,
|
||||
priority,
|
||||
message,
|
||||
_onTriggerCompiledScript,
|
||||
_instanceActor,
|
||||
_sharedScriptLibrary,
|
||||
_options,
|
||||
_logger));
|
||||
|
||||
Context.ActorOf(props, executionId);
|
||||
}
|
||||
|
||||
private AlarmEvalConfig ParseEvalConfig(string? triggerConfigJson)
|
||||
{
|
||||
if (string.IsNullOrEmpty(triggerConfigJson))
|
||||
return new ValueMatchEvalConfig("", null);
|
||||
|
||||
try
|
||||
{
|
||||
var doc = JsonDocument.Parse(triggerConfigJson);
|
||||
var root = doc.RootElement;
|
||||
|
||||
// Support both "attributeName" and "attribute" keys
|
||||
var attr = root.TryGetProperty("attributeName", out var attrEl)
|
||||
? attrEl.GetString() ?? ""
|
||||
: root.TryGetProperty("attribute", out var attrEl2)
|
||||
? attrEl2.GetString() ?? ""
|
||||
: "";
|
||||
|
||||
return _triggerType switch
|
||||
{
|
||||
AlarmTriggerType.ValueMatch => new ValueMatchEvalConfig(
|
||||
attr,
|
||||
root.TryGetProperty("matchValue", out var mv) ? mv.GetString()
|
||||
: root.TryGetProperty("value", out var mv2) ? mv2.GetString()
|
||||
: null),
|
||||
|
||||
AlarmTriggerType.RangeViolation => new RangeViolationEvalConfig(
|
||||
attr,
|
||||
root.TryGetProperty("min", out var minEl) ? minEl.GetDouble()
|
||||
: root.TryGetProperty("low", out var lowEl) ? lowEl.GetDouble()
|
||||
: double.MinValue,
|
||||
root.TryGetProperty("max", out var maxEl) ? maxEl.GetDouble()
|
||||
: root.TryGetProperty("high", out var highEl) ? highEl.GetDouble()
|
||||
: double.MaxValue),
|
||||
|
||||
AlarmTriggerType.RateOfChange => new RateOfChangeEvalConfig(
|
||||
attr,
|
||||
root.TryGetProperty("thresholdPerSecond", out var tps) ? tps.GetDouble() : 10.0,
|
||||
root.TryGetProperty("windowSeconds", out var ws)
|
||||
? TimeSpan.FromSeconds(ws.GetDouble())
|
||||
: TimeSpan.FromSeconds(1),
|
||||
root.TryGetProperty("direction", out var dirEl)
|
||||
? ParseDirection(dirEl.GetString())
|
||||
: RateOfChangeDirection.Either),
|
||||
|
||||
AlarmTriggerType.HiLo => new HiLoEvalConfig(
|
||||
attr,
|
||||
LoLo: TryReadDouble(root, "loLo"),
|
||||
Lo: TryReadDouble(root, "lo"),
|
||||
Hi: TryReadDouble(root, "hi"),
|
||||
HiHi: TryReadDouble(root, "hiHi"),
|
||||
LoLoPriority: TryReadInt(root, "loLoPriority"),
|
||||
LoPriority: TryReadInt(root, "loPriority"),
|
||||
HiPriority: TryReadInt(root, "hiPriority"),
|
||||
HiHiPriority: TryReadInt(root, "hiHiPriority"),
|
||||
LoLoDeadband: TryReadDouble(root, "loLoDeadband"),
|
||||
LoDeadband: TryReadDouble(root, "loDeadband"),
|
||||
HiDeadband: TryReadDouble(root, "hiDeadband"),
|
||||
HiHiDeadband: TryReadDouble(root, "hiHiDeadband"),
|
||||
LoLoMessage: TryReadString(root, "loLoMessage"),
|
||||
LoMessage: TryReadString(root, "loMessage"),
|
||||
HiMessage: TryReadString(root, "hiMessage"),
|
||||
HiHiMessage: TryReadString(root, "hiHiMessage")),
|
||||
|
||||
// Expression triggers have no single monitored attribute; they
|
||||
// evaluate the compiled expression (passed into the actor and
|
||||
// cached in _compiledTriggerExpression) over the full attribute
|
||||
// snapshot. MonitoredAttributeName is unused.
|
||||
AlarmTriggerType.Expression => new ExpressionEvalConfig(
|
||||
"",
|
||||
TriggerExpressionGlobals.ExtractExpression(triggerConfigJson) ?? ""),
|
||||
|
||||
_ => new ValueMatchEvalConfig(attr, null)
|
||||
};
|
||||
}
|
||||
catch (Exception ex)
|
||||
{
|
||||
_logger.LogWarning(ex, "Failed to parse alarm trigger config for {Alarm}", _alarmName);
|
||||
return new ValueMatchEvalConfig("", null);
|
||||
}
|
||||
}
|
||||
|
||||
private static RateOfChangeDirection ParseDirection(string? raw) => raw?.ToLowerInvariant() switch
|
||||
{
|
||||
"rising" or "up" or "positive" => RateOfChangeDirection.Rising,
|
||||
"falling" or "down" or "negative" => RateOfChangeDirection.Falling,
|
||||
_ => RateOfChangeDirection.Either
|
||||
};
|
||||
|
||||
private static double? TryReadDouble(JsonElement el, string name)
|
||||
{
|
||||
if (!el.TryGetProperty(name, out var p)) return null;
|
||||
return p.ValueKind switch
|
||||
{
|
||||
JsonValueKind.Number => p.GetDouble(),
|
||||
JsonValueKind.String when double.TryParse(p.GetString(), System.Globalization.NumberStyles.Float, System.Globalization.CultureInfo.InvariantCulture, out var v) => v,
|
||||
_ => null
|
||||
};
|
||||
}
|
||||
|
||||
private static int? TryReadInt(JsonElement el, string name)
|
||||
{
|
||||
if (!el.TryGetProperty(name, out var p)) return null;
|
||||
return p.ValueKind switch
|
||||
{
|
||||
JsonValueKind.Number when p.TryGetInt32(out var i) => i,
|
||||
JsonValueKind.Number => (int)p.GetDouble(),
|
||||
JsonValueKind.String when int.TryParse(p.GetString(), System.Globalization.NumberStyles.Integer, System.Globalization.CultureInfo.InvariantCulture, out var v) => v,
|
||||
_ => null
|
||||
};
|
||||
}
|
||||
|
||||
private static string? TryReadString(JsonElement el, string name)
|
||||
{
|
||||
if (!el.TryGetProperty(name, out var p)) return null;
|
||||
return p.ValueKind == JsonValueKind.String ? p.GetString() : null;
|
||||
}
|
||||
|
||||
// ── Internal messages ──
|
||||
internal record AlarmExecutionCompleted(string AlarmName, bool Success);
|
||||
}
|
||||
|
||||
internal enum RateOfChangeDirection { Either, Rising, Falling }
|
||||
|
||||
// ── Alarm evaluation config types ──
|
||||
internal abstract record AlarmEvalConfig(string MonitoredAttributeName);
|
||||
internal record ValueMatchEvalConfig(string MonitoredAttributeName, string? MatchValue) : AlarmEvalConfig(MonitoredAttributeName);
|
||||
internal record RangeViolationEvalConfig(string MonitoredAttributeName, double Min, double Max) : AlarmEvalConfig(MonitoredAttributeName);
|
||||
internal record RateOfChangeEvalConfig(
|
||||
string MonitoredAttributeName,
|
||||
double ThresholdPerSecond,
|
||||
TimeSpan WindowDuration,
|
||||
RateOfChangeDirection Direction) : AlarmEvalConfig(MonitoredAttributeName);
|
||||
|
||||
/// <summary>
|
||||
/// Expression evaluation config: a read-only boolean C# expression evaluated
|
||||
/// over the full attribute snapshot. Has no single monitored attribute
|
||||
/// (<see cref="AlarmEvalConfig.MonitoredAttributeName"/> is empty). The
|
||||
/// compiled expression itself lives on the actor's <c>_compiledTriggerExpression</c>
|
||||
/// field, the single source for the hot path.
|
||||
/// </summary>
|
||||
internal record ExpressionEvalConfig(
|
||||
string MonitoredAttributeName,
|
||||
string Expression) : AlarmEvalConfig(MonitoredAttributeName);
|
||||
|
||||
/// <summary>
|
||||
/// HiLo evaluation config: any subset of the four setpoints may be set; null
|
||||
/// means "don't evaluate that band". Per-setpoint priorities override the
|
||||
/// alarm-level priority for AlarmStateChanged messages emitted for that band.
|
||||
/// </summary>
|
||||
internal record HiLoEvalConfig(
|
||||
string MonitoredAttributeName,
|
||||
double? LoLo,
|
||||
double? Lo,
|
||||
double? Hi,
|
||||
double? HiHi,
|
||||
int? LoLoPriority,
|
||||
int? LoPriority,
|
||||
int? HiPriority,
|
||||
int? HiHiPriority,
|
||||
double? LoLoDeadband = null,
|
||||
double? LoDeadband = null,
|
||||
double? HiDeadband = null,
|
||||
double? HiHiDeadband = null,
|
||||
string? LoLoMessage = null,
|
||||
string? LoMessage = null,
|
||||
string? HiMessage = null,
|
||||
string? HiHiMessage = null) : AlarmEvalConfig(MonitoredAttributeName);
|
||||
@@ -0,0 +1,128 @@
|
||||
using Akka.Actor;
|
||||
using Microsoft.CodeAnalysis.Scripting;
|
||||
using Microsoft.Extensions.Logging;
|
||||
using ZB.MOM.WW.ScadaBridge.Commons.Types;
|
||||
using ZB.MOM.WW.ScadaBridge.Commons.Types.Enums;
|
||||
using ZB.MOM.WW.ScadaBridge.Commons.Types.Scripts;
|
||||
using ZB.MOM.WW.ScadaBridge.SiteRuntime.Scripts;
|
||||
|
||||
namespace ZB.MOM.WW.ScadaBridge.SiteRuntime.Actors;
|
||||
|
||||
/// <summary>
|
||||
/// WP-16: Alarm Execution Actor -- short-lived child of Alarm Actor.
|
||||
/// Same pattern as ScriptExecutionActor.
|
||||
/// WP-21: CAN call Instance.CallScript() (ask to sibling Script Actor).
|
||||
/// Instance scripts CANNOT call alarm on-trigger scripts (no API for it).
|
||||
/// Supervision: Stop on unhandled exception.
|
||||
/// </summary>
|
||||
public class AlarmExecutionActor : ReceiveActor
|
||||
{
|
||||
/// <summary>Initializes a new <see cref="AlarmExecutionActor"/> and immediately schedules execution of the alarm on-trigger script.</summary>
|
||||
/// <param name="alarmName">The canonical name of the alarm that triggered.</param>
|
||||
/// <param name="instanceName">The name of the owning instance.</param>
|
||||
/// <param name="level">The alarm severity level at the time of triggering.</param>
|
||||
/// <param name="priority">The alarm priority value.</param>
|
||||
/// <param name="message">The alarm message to pass to the script.</param>
|
||||
/// <param name="compiledScript">The pre-compiled on-trigger script to execute.</param>
|
||||
/// <param name="instanceActor">Reference to the parent instance actor for attribute/script calls.</param>
|
||||
/// <param name="sharedScriptLibrary">Shared script library providing common utilities.</param>
|
||||
/// <param name="options">Site runtime configuration options, including the execution timeout.</param>
|
||||
/// <param name="logger">Logger for execution diagnostics.</param>
|
||||
public AlarmExecutionActor(
|
||||
string alarmName,
|
||||
string instanceName,
|
||||
AlarmLevel level,
|
||||
int priority,
|
||||
string message,
|
||||
Script<object?> compiledScript,
|
||||
IActorRef instanceActor,
|
||||
SharedScriptLibrary sharedScriptLibrary,
|
||||
SiteRuntimeOptions options,
|
||||
ILogger logger)
|
||||
{
|
||||
var self = Self;
|
||||
var parent = Context.Parent;
|
||||
|
||||
ExecuteAlarmScript(
|
||||
alarmName, instanceName, level, priority, message,
|
||||
compiledScript, instanceActor,
|
||||
sharedScriptLibrary, options, self, parent, logger);
|
||||
}
|
||||
|
||||
private static void ExecuteAlarmScript(
|
||||
string alarmName,
|
||||
string instanceName,
|
||||
AlarmLevel level,
|
||||
int priority,
|
||||
string message,
|
||||
Script<object?> compiledScript,
|
||||
IActorRef instanceActor,
|
||||
SharedScriptLibrary sharedScriptLibrary,
|
||||
SiteRuntimeOptions options,
|
||||
IActorRef self,
|
||||
IActorRef parent,
|
||||
ILogger logger)
|
||||
{
|
||||
var timeout = TimeSpan.FromSeconds(options.ScriptExecutionTimeoutSeconds);
|
||||
|
||||
// SiteRuntime-009: run the alarm on-trigger body on the dedicated
|
||||
// script-execution scheduler, not the shared .NET thread pool.
|
||||
var scheduler = ScriptExecutionScheduler.Shared(options);
|
||||
|
||||
_ = Task.Factory.StartNew(async () =>
|
||||
{
|
||||
using var cts = new CancellationTokenSource(timeout);
|
||||
try
|
||||
{
|
||||
// WP-21: AlarmExecutionActor can call Instance.CallScript()
|
||||
// via the ScriptRuntimeContext injected into globals
|
||||
var context = new ScriptRuntimeContext(
|
||||
instanceActor,
|
||||
self,
|
||||
sharedScriptLibrary,
|
||||
currentCallDepth: 0,
|
||||
options.MaxScriptCallDepth,
|
||||
timeout,
|
||||
instanceName,
|
||||
logger);
|
||||
|
||||
var globals = new ScriptGlobals
|
||||
{
|
||||
Instance = context,
|
||||
Parameters = new ScriptParameters(),
|
||||
CancellationToken = cts.Token,
|
||||
Alarm = new AlarmContext
|
||||
{
|
||||
Name = alarmName,
|
||||
Level = level,
|
||||
Priority = priority,
|
||||
Message = message
|
||||
}
|
||||
};
|
||||
|
||||
await compiledScript.RunAsync(globals, cts.Token);
|
||||
|
||||
parent.Tell(new AlarmActor.AlarmExecutionCompleted(alarmName, true));
|
||||
}
|
||||
catch (OperationCanceledException)
|
||||
{
|
||||
logger.LogWarning(
|
||||
"Alarm on-trigger script for {Alarm} on {Instance} timed out",
|
||||
alarmName, instanceName);
|
||||
parent.Tell(new AlarmActor.AlarmExecutionCompleted(alarmName, false));
|
||||
}
|
||||
catch (Exception ex)
|
||||
{
|
||||
// WP-32: Failures logged, alarm continues
|
||||
logger.LogError(ex,
|
||||
"Alarm on-trigger script for {Alarm} on {Instance} failed",
|
||||
alarmName, instanceName);
|
||||
parent.Tell(new AlarmActor.AlarmExecutionCompleted(alarmName, false));
|
||||
}
|
||||
finally
|
||||
{
|
||||
self.Tell(PoisonPill.Instance);
|
||||
}
|
||||
}, CancellationToken.None, TaskCreationOptions.DenyChildAttach, scheduler).Unwrap();
|
||||
}
|
||||
}
|
||||
File diff suppressed because it is too large
Load Diff
@@ -0,0 +1,800 @@
|
||||
using Akka.Actor;
|
||||
using Microsoft.CodeAnalysis.Scripting;
|
||||
using Microsoft.Extensions.Logging;
|
||||
using ZB.MOM.WW.ScadaBridge.Commons.Messages.DataConnection;
|
||||
using ZB.MOM.WW.ScadaBridge.Commons.Messages.DebugView;
|
||||
using ZB.MOM.WW.ScadaBridge.Commons.Messages.Instance;
|
||||
using ZB.MOM.WW.ScadaBridge.Commons.Messages.ScriptExecution;
|
||||
using ZB.MOM.WW.ScadaBridge.Commons.Messages.Streaming;
|
||||
using ZB.MOM.WW.ScadaBridge.Commons.Types.Enums;
|
||||
using ZB.MOM.WW.ScadaBridge.Commons.Types.Flattening;
|
||||
using ZB.MOM.WW.ScadaBridge.HealthMonitoring;
|
||||
using ZB.MOM.WW.ScadaBridge.SiteRuntime.Persistence;
|
||||
using ZB.MOM.WW.ScadaBridge.SiteRuntime.Scripts;
|
||||
using ZB.MOM.WW.ScadaBridge.SiteRuntime.Streaming;
|
||||
using System.Text.Json;
|
||||
|
||||
namespace ZB.MOM.WW.ScadaBridge.SiteRuntime.Actors;
|
||||
|
||||
/// <summary>
|
||||
/// Represents a single deployed instance at runtime. Holds the in-memory attribute state
|
||||
/// (loaded from FlattenedConfiguration + static overrides from SQLite).
|
||||
///
|
||||
/// The Instance Actor is the single source of truth for runtime instance state.
|
||||
/// WP-24: All state mutations are serialized through the actor mailbox.
|
||||
/// Multiple Script Execution Actors run concurrently; state mutations through this actor.
|
||||
///
|
||||
/// WP-15/16: Creates child Script Actors and Alarm Actors on startup.
|
||||
/// WP-22: Tell for tag value updates, attribute notifications, stream publishing.
|
||||
/// Ask for CallScript, debug snapshot.
|
||||
/// WP-25: Debug view backend — snapshot + stream subscription.
|
||||
/// </summary>
|
||||
public class InstanceActor : ReceiveActor
|
||||
{
|
||||
private readonly string _instanceUniqueName;
|
||||
private readonly SiteStorageService _storage;
|
||||
private readonly ScriptCompilationService _compilationService;
|
||||
private readonly SharedScriptLibrary _sharedScriptLibrary;
|
||||
private readonly SiteStreamManager? _streamManager;
|
||||
private readonly SiteRuntimeOptions _options;
|
||||
private readonly ILogger _logger;
|
||||
private readonly ISiteHealthCollector? _healthCollector;
|
||||
private readonly IServiceProvider? _serviceProvider;
|
||||
private readonly Dictionary<string, object?> _attributes = new();
|
||||
private readonly Dictionary<string, string> _attributeQualities = new();
|
||||
private readonly Dictionary<string, DateTimeOffset> _attributeTimestamps = new();
|
||||
private readonly Dictionary<string, AlarmState> _alarmStates = new();
|
||||
private readonly Dictionary<string, DateTimeOffset> _alarmTimestamps = new();
|
||||
private readonly Dictionary<string, int> _alarmPriorities = new();
|
||||
private readonly Dictionary<string, IActorRef> _scriptActors = new();
|
||||
private readonly Dictionary<string, IActorRef> _alarmActors = new();
|
||||
private FlattenedConfiguration? _configuration;
|
||||
|
||||
// DCL manager actor reference for subscribing to tag values
|
||||
private readonly IActorRef? _dclManager;
|
||||
// Maps each tag path to every attribute canonical name that references it.
|
||||
// A tag path can back more than one attribute (e.g. two composed modules
|
||||
// whose members reference the same PLC node), so a tag value update must
|
||||
// fan out to all of them — not just the last one registered.
|
||||
private readonly Dictionary<string, List<string>> _tagPathToAttributes = new();
|
||||
|
||||
/// <summary>
|
||||
/// Initializes the instance actor with its configuration and dependencies.
|
||||
/// </summary>
|
||||
/// <param name="instanceUniqueName">System-wide unique name identifying this instance.</param>
|
||||
/// <param name="configJson">JSON-serialized flattened configuration for this instance.</param>
|
||||
/// <param name="storage">Site storage service for loading and persisting static overrides.</param>
|
||||
/// <param name="compilationService">Service used to compile instance scripts.</param>
|
||||
/// <param name="sharedScriptLibrary">Library of shared scripts available to instance scripts.</param>
|
||||
/// <param name="streamManager">Optional site stream manager for publishing attribute/alarm changes.</param>
|
||||
/// <param name="options">Site runtime configuration options.</param>
|
||||
/// <param name="logger">Logger for this actor.</param>
|
||||
/// <param name="dclManager">Optional Data Connection Layer manager actor reference.</param>
|
||||
/// <param name="healthCollector">Optional health collector for reporting metrics.</param>
|
||||
/// <param name="serviceProvider">Optional DI service provider for script execution services.</param>
|
||||
public InstanceActor(
|
||||
string instanceUniqueName,
|
||||
string configJson,
|
||||
SiteStorageService storage,
|
||||
ScriptCompilationService compilationService,
|
||||
SharedScriptLibrary sharedScriptLibrary,
|
||||
SiteStreamManager? streamManager,
|
||||
SiteRuntimeOptions options,
|
||||
ILogger logger,
|
||||
IActorRef? dclManager = null,
|
||||
ISiteHealthCollector? healthCollector = null,
|
||||
IServiceProvider? serviceProvider = null)
|
||||
{
|
||||
_instanceUniqueName = instanceUniqueName;
|
||||
_storage = storage;
|
||||
_compilationService = compilationService;
|
||||
_sharedScriptLibrary = sharedScriptLibrary;
|
||||
_streamManager = streamManager;
|
||||
_options = options;
|
||||
_logger = logger;
|
||||
_dclManager = dclManager;
|
||||
_healthCollector = healthCollector;
|
||||
_serviceProvider = serviceProvider;
|
||||
|
||||
// Deserialize the flattened configuration
|
||||
_configuration = JsonSerializer.Deserialize<FlattenedConfiguration>(configJson);
|
||||
|
||||
// Load default attribute values from the flattened configuration
|
||||
// Data-sourced attributes start with Uncertain quality until the first DCL value arrives.
|
||||
// Static attributes start with Good quality.
|
||||
if (_configuration != null)
|
||||
{
|
||||
foreach (var attr in _configuration.Attributes)
|
||||
{
|
||||
_attributes[attr.CanonicalName] = attr.Value;
|
||||
_attributeQualities[attr.CanonicalName] =
|
||||
string.IsNullOrEmpty(attr.DataSourceReference) ? "Good" : "Uncertain";
|
||||
}
|
||||
}
|
||||
|
||||
// Handle attribute queries (Tell pattern -- sender gets response)
|
||||
Receive<GetAttributeRequest>(HandleGetAttribute);
|
||||
|
||||
// Handle static attribute writes
|
||||
Receive<SetStaticAttributeCommand>(HandleSetStaticAttribute);
|
||||
|
||||
// SiteRuntime-019: the disable/enable lifecycle is owned entirely by the
|
||||
// Deployment Manager — DeploymentManagerActor.HandleDisable/HandleEnable
|
||||
// stop or re-create the Instance Actor directly and reply to the caller.
|
||||
// DisableInstanceCommand / EnableInstanceCommand are never routed to the
|
||||
// Instance Actor, so no handlers are registered here. (The previous no-op
|
||||
// handlers were dead code that implied a non-existent instance-side
|
||||
// acknowledgement contract.)
|
||||
|
||||
// WP-15: Handle script call requests — route to appropriate Script Actor (Ask pattern)
|
||||
Receive<ScriptCallRequest>(HandleScriptCallRequest);
|
||||
|
||||
// WP-22/23: Handle attribute value changes from DCL (Tell pattern)
|
||||
Receive<AttributeValueChanged>(HandleAttributeValueChanged);
|
||||
|
||||
// Handle tag value updates from DCL — convert to AttributeValueChanged
|
||||
Receive<TagValueUpdate>(HandleTagValueUpdate);
|
||||
Receive<SubscribeTagsResponse>(_ => { }); // Ack from DCL subscribe — no action needed
|
||||
Receive<ConnectionQualityChanged>(HandleConnectionQualityChanged);
|
||||
|
||||
// WP-16: Handle alarm state changes from Alarm Actors (Tell pattern)
|
||||
Receive<AlarmStateChanged>(HandleAlarmStateChanged);
|
||||
|
||||
// WP-25: Debug view subscribe/unsubscribe (Ask pattern for snapshot)
|
||||
Receive<SubscribeDebugViewRequest>(HandleSubscribeDebugView);
|
||||
Receive<UnsubscribeDebugViewRequest>(HandleUnsubscribeDebugView);
|
||||
|
||||
// Debug snapshot (one-shot, no subscription)
|
||||
Receive<DebugSnapshotRequest>(HandleDebugSnapshot);
|
||||
|
||||
// Handle internal messages
|
||||
Receive<LoadOverridesResult>(HandleOverridesLoaded);
|
||||
}
|
||||
|
||||
/// <inheritdoc />
|
||||
protected override void PreStart()
|
||||
{
|
||||
base.PreStart();
|
||||
_logger.LogInformation("InstanceActor started for {Instance}", _instanceUniqueName);
|
||||
|
||||
// Asynchronously load static overrides from SQLite and pipe to self
|
||||
var self = Self;
|
||||
_storage.GetStaticOverridesAsync(_instanceUniqueName).ContinueWith(t =>
|
||||
{
|
||||
if (t.IsCompletedSuccessfully)
|
||||
return new LoadOverridesResult(t.Result, null);
|
||||
return new LoadOverridesResult(new Dictionary<string, string>(), t.Exception?.GetBaseException().Message);
|
||||
}).PipeTo(self);
|
||||
|
||||
// Create child Script Actors and Alarm Actors from configuration
|
||||
CreateChildActors();
|
||||
|
||||
// Subscribe to DCL for data-sourced attributes
|
||||
SubscribeToDcl();
|
||||
}
|
||||
|
||||
/// <inheritdoc />
|
||||
protected override SupervisorStrategy SupervisorStrategy()
|
||||
{
|
||||
return new OneForOneStrategy(
|
||||
maxNrOfRetries: -1,
|
||||
withinTimeRange: TimeSpan.FromMinutes(1),
|
||||
decider: Decider.From(ex =>
|
||||
{
|
||||
_logger.LogWarning(ex,
|
||||
"Child actor on instance {Instance} threw exception, resuming",
|
||||
_instanceUniqueName);
|
||||
return Directive.Resume;
|
||||
}));
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// Returns the current attribute value. Uses Tell pattern; sender gets the response.
|
||||
/// </summary>
|
||||
private void HandleGetAttribute(GetAttributeRequest request)
|
||||
{
|
||||
var found = _attributes.TryGetValue(request.AttributeName, out var value);
|
||||
_attributeQualities.TryGetValue(request.AttributeName, out var quality);
|
||||
Sender.Tell(new GetAttributeResponse(
|
||||
request.CorrelationId,
|
||||
_instanceUniqueName,
|
||||
request.AttributeName,
|
||||
value,
|
||||
found,
|
||||
quality ?? "Good",
|
||||
DateTimeOffset.UtcNow));
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// Handles an attribute write (<c>Instance.SetAttribute</c> / Inbound API).
|
||||
/// WP-24: State mutation serialized through this actor's mailbox.
|
||||
///
|
||||
/// The write is routed by the attribute's data binding:
|
||||
/// * Data-sourced attribute → forwards a <see cref="WriteTagRequest"/> to the
|
||||
/// DCL, which writes the physical device. The in-memory value is NOT
|
||||
/// optimistically updated and NO static override is persisted — the
|
||||
/// confirmed device value arrives later via the subscription. Success or
|
||||
/// failure of the device write is returned to the caller.
|
||||
/// * Static attribute → updates the in-memory value and persists the override
|
||||
/// to SQLite.
|
||||
///
|
||||
/// Either way the caller receives a <see cref="SetStaticAttributeResponse"/>.
|
||||
/// </summary>
|
||||
private void HandleSetStaticAttribute(SetStaticAttributeCommand command)
|
||||
{
|
||||
// Resolve the target attribute's data binding from the flattened config.
|
||||
var resolved = _configuration?.Attributes
|
||||
.FirstOrDefault(a => a.CanonicalName == command.AttributeName);
|
||||
|
||||
// SiteRuntime-025: reject writes targeting an attribute that does not exist
|
||||
// on the deployed instance. Without this check, an inbound API
|
||||
// SetAttribute("notARealAttr", ...) would pollute the in-memory
|
||||
// _attributes dictionary, publish a synthetic AttributeValueChanged to
|
||||
// debug-view subscribers, and persist a durable static-override row that
|
||||
// resurrects on every restart. The override row is also outside the
|
||||
// ClearStaticOverridesAsync window for unknown names. Refuse the write
|
||||
// and let the caller see the failure, mirroring the script trust model's
|
||||
// "scripts can only read/write attributes on their own instance" framing.
|
||||
if (resolved == null)
|
||||
{
|
||||
_logger.LogWarning(
|
||||
"SetAttribute rejected — attribute '{Attribute}' is not defined on instance '{Instance}'",
|
||||
command.AttributeName, _instanceUniqueName);
|
||||
Sender.Tell(new SetStaticAttributeResponse(
|
||||
command.CorrelationId,
|
||||
_instanceUniqueName,
|
||||
command.AttributeName,
|
||||
false,
|
||||
$"Unknown attribute '{command.AttributeName}'",
|
||||
DateTimeOffset.UtcNow));
|
||||
return;
|
||||
}
|
||||
|
||||
var isDataSourced =
|
||||
!string.IsNullOrEmpty(resolved.DataSourceReference)
|
||||
&& !string.IsNullOrEmpty(resolved.BoundDataConnectionName);
|
||||
|
||||
if (isDataSourced)
|
||||
{
|
||||
HandleSetDataAttribute(command, resolved);
|
||||
return;
|
||||
}
|
||||
|
||||
HandleSetStaticAttributeCore(command);
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// Static attribute write: updates in-memory state, publishes the change,
|
||||
/// persists the override to SQLite, and replies with success.
|
||||
/// </summary>
|
||||
private void HandleSetStaticAttributeCore(SetStaticAttributeCommand command)
|
||||
{
|
||||
_attributes[command.AttributeName] = command.Value;
|
||||
|
||||
// Publish attribute change to stream (WP-23) and notify children
|
||||
var changed = new AttributeValueChanged(
|
||||
_instanceUniqueName,
|
||||
command.AttributeName,
|
||||
command.AttributeName,
|
||||
command.Value,
|
||||
"Good",
|
||||
DateTimeOffset.UtcNow);
|
||||
|
||||
PublishAndNotifyChildren(changed);
|
||||
|
||||
// Persist asynchronously -- fire and forget since the actor is the source of truth.
|
||||
var instanceName = _instanceUniqueName;
|
||||
var attributeName = command.AttributeName;
|
||||
var logger = _logger;
|
||||
_storage.SetStaticOverrideAsync(_instanceUniqueName, command.AttributeName, command.Value)
|
||||
.ContinueWith(t =>
|
||||
{
|
||||
logger.LogWarning(
|
||||
t.Exception?.GetBaseException(),
|
||||
"Failed to persist static override for {Instance}.{Attribute}; in-memory state is authoritative",
|
||||
instanceName,
|
||||
attributeName);
|
||||
}, TaskContinuationOptions.OnlyOnFaulted);
|
||||
|
||||
Sender.Tell(new SetStaticAttributeResponse(
|
||||
command.CorrelationId, _instanceUniqueName, command.AttributeName,
|
||||
true, null, DateTimeOffset.UtcNow));
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// Data-sourced attribute write: forwards a write request to the DCL and pipes
|
||||
/// the device write result back to the caller. The in-memory value is left
|
||||
/// untouched (it is refreshed by the subscription when the device confirms);
|
||||
/// no static override is persisted for a data-sourced attribute.
|
||||
/// </summary>
|
||||
private void HandleSetDataAttribute(SetStaticAttributeCommand command, ResolvedAttribute resolved)
|
||||
{
|
||||
var caller = Sender;
|
||||
var correlationId = command.CorrelationId;
|
||||
var attributeName = command.AttributeName;
|
||||
var instanceName = _instanceUniqueName;
|
||||
|
||||
if (_dclManager == null)
|
||||
{
|
||||
_logger.LogWarning(
|
||||
"SetAttribute on data-sourced attribute {Instance}.{Attribute} cannot be routed — no DCL manager configured",
|
||||
instanceName, attributeName);
|
||||
caller.Tell(new SetStaticAttributeResponse(
|
||||
correlationId, instanceName, attributeName, false,
|
||||
"Data Connection Layer not available for write.", DateTimeOffset.UtcNow));
|
||||
return;
|
||||
}
|
||||
|
||||
var writeRequest = new WriteTagRequest(
|
||||
correlationId,
|
||||
resolved.BoundDataConnectionName!,
|
||||
resolved.DataSourceReference!,
|
||||
command.Value,
|
||||
DateTimeOffset.UtcNow);
|
||||
|
||||
// Ask the DCL and pipe the result back to the original caller. The DCL
|
||||
// returns the failure synchronously so the script can handle it.
|
||||
_dclManager.Ask<WriteTagResponse>(writeRequest, TimeSpan.FromSeconds(30))
|
||||
.ContinueWith(t =>
|
||||
{
|
||||
if (t.IsCompletedSuccessfully)
|
||||
return new SetStaticAttributeResponse(
|
||||
correlationId, instanceName, attributeName,
|
||||
t.Result.Success, t.Result.ErrorMessage, DateTimeOffset.UtcNow);
|
||||
|
||||
return new SetStaticAttributeResponse(
|
||||
correlationId, instanceName, attributeName, false,
|
||||
t.Exception?.GetBaseException().Message ?? "DCL write timed out",
|
||||
DateTimeOffset.UtcNow);
|
||||
}).PipeTo(caller);
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// WP-15: Routes script call requests to the appropriate Script Actor.
|
||||
/// Uses Ask pattern (WP-22).
|
||||
/// </summary>
|
||||
private void HandleScriptCallRequest(ScriptCallRequest request)
|
||||
{
|
||||
if (_scriptActors.TryGetValue(request.ScriptName, out var scriptActor))
|
||||
{
|
||||
// Forward the request to the Script Actor, preserving the original
|
||||
// sender. The whole record is forwarded unchanged, so any
|
||||
// ParentExecutionId (Audit Log #23) set by an inbound-API-routed
|
||||
// call is carried through to the Script Actor verbatim.
|
||||
scriptActor.Forward(request);
|
||||
}
|
||||
else
|
||||
{
|
||||
Sender.Tell(new ScriptCallResult(
|
||||
request.CorrelationId,
|
||||
false,
|
||||
null,
|
||||
$"Script '{request.ScriptName}' not found on instance '{_instanceUniqueName}'."));
|
||||
}
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// WP-22/23: Handles attribute value changes from DCL or static writes.
|
||||
/// Updates in-memory state, publishes to stream, and notifies children.
|
||||
/// </summary>
|
||||
private void HandleAttributeValueChanged(AttributeValueChanged changed)
|
||||
{
|
||||
// WP-24: State mutation serialized through this actor
|
||||
_attributes[changed.AttributeName] = changed.Value;
|
||||
_attributeQualities[changed.AttributeName] = changed.Quality;
|
||||
_attributeTimestamps[changed.AttributeName] = changed.Timestamp;
|
||||
|
||||
PublishAndNotifyChildren(changed);
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// Handles tag value updates from DCL. Maps the tag path back to the attribute
|
||||
/// canonical name and converts to an AttributeValueChanged for unified processing.
|
||||
/// </summary>
|
||||
private void HandleTagValueUpdate(TagValueUpdate update)
|
||||
{
|
||||
if (!_tagPathToAttributes.TryGetValue(update.TagPath, out var attrNames))
|
||||
return;
|
||||
|
||||
// Normalize array values to JSON strings so they survive Akka serialization
|
||||
var value = update.Value is Array
|
||||
? System.Text.Json.JsonSerializer.Serialize(update.Value, update.Value.GetType())
|
||||
: update.Value;
|
||||
|
||||
// One tag path may back several attributes — update every one of them.
|
||||
foreach (var attrName in attrNames)
|
||||
{
|
||||
var changed = new AttributeValueChanged(
|
||||
_instanceUniqueName, update.TagPath, attrName,
|
||||
value, update.Quality.ToString(), update.Timestamp);
|
||||
HandleAttributeValueChanged(changed);
|
||||
}
|
||||
}
|
||||
|
||||
private void HandleConnectionQualityChanged(ConnectionQualityChanged qualityChanged)
|
||||
{
|
||||
_logger.LogWarning("Connection {Connection} quality changed to {Quality} for instance {Instance}",
|
||||
qualityChanged.ConnectionName, qualityChanged.Quality, _instanceUniqueName);
|
||||
|
||||
if (_configuration == null) return;
|
||||
|
||||
// Mark all attributes bound to this connection with the new quality
|
||||
// and publish to the site stream so the debug view updates in real-time.
|
||||
// We intentionally do NOT notify script/alarm actors here — the value
|
||||
// hasn't changed, only the quality, and firing scripts/alarms would
|
||||
// cause spurious evaluations.
|
||||
var qualityStr = qualityChanged.Quality.ToString();
|
||||
foreach (var attr in _configuration.Attributes)
|
||||
{
|
||||
if (attr.BoundDataConnectionName == qualityChanged.ConnectionName &&
|
||||
!string.IsNullOrEmpty(attr.DataSourceReference))
|
||||
{
|
||||
_attributeQualities[attr.CanonicalName] = qualityStr;
|
||||
_attributeTimestamps[attr.CanonicalName] = qualityChanged.Timestamp;
|
||||
|
||||
// Publish quality change to stream (current value, new quality)
|
||||
_attributes.TryGetValue(attr.CanonicalName, out var currentValue);
|
||||
_streamManager?.PublishAttributeValueChanged(new AttributeValueChanged(
|
||||
_instanceUniqueName,
|
||||
attr.DataSourceReference,
|
||||
attr.CanonicalName,
|
||||
currentValue,
|
||||
qualityStr,
|
||||
qualityChanged.Timestamp));
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// Subscribes to DCL for all data-sourced attributes. Groups tag paths by connection
|
||||
/// name and sends SubscribeTagsRequest to the DCL manager.
|
||||
/// </summary>
|
||||
private void SubscribeToDcl()
|
||||
{
|
||||
if (_dclManager == null || _configuration == null) return;
|
||||
|
||||
// Group attributes by their bound connection name
|
||||
var byConnection = new Dictionary<string, List<string>>();
|
||||
foreach (var attr in _configuration.Attributes)
|
||||
{
|
||||
if (string.IsNullOrEmpty(attr.DataSourceReference) ||
|
||||
string.IsNullOrEmpty(attr.BoundDataConnectionName))
|
||||
continue;
|
||||
|
||||
// Record every attribute that references this tag path so a single
|
||||
// tag value update fans out to all of them.
|
||||
if (!_tagPathToAttributes.TryGetValue(attr.DataSourceReference, out var attrs))
|
||||
{
|
||||
attrs = new List<string>();
|
||||
_tagPathToAttributes[attr.DataSourceReference] = attrs;
|
||||
}
|
||||
attrs.Add(attr.CanonicalName);
|
||||
|
||||
if (!byConnection.TryGetValue(attr.BoundDataConnectionName, out var connTags))
|
||||
{
|
||||
connTags = new List<string>();
|
||||
byConnection[attr.BoundDataConnectionName] = connTags;
|
||||
}
|
||||
// Subscribe each distinct tag path once per connection — a tag shared
|
||||
// by several attributes still needs only one DCL subscription.
|
||||
if (!connTags.Contains(attr.DataSourceReference))
|
||||
connTags.Add(attr.DataSourceReference);
|
||||
}
|
||||
|
||||
// Send subscription requests to DCL for each connection
|
||||
foreach (var (connectionName, tagPaths) in byConnection)
|
||||
{
|
||||
var request = new SubscribeTagsRequest(
|
||||
Guid.NewGuid().ToString("N"),
|
||||
_instanceUniqueName,
|
||||
connectionName,
|
||||
tagPaths,
|
||||
DateTimeOffset.UtcNow);
|
||||
_dclManager.Tell(request, Self);
|
||||
_logger.LogInformation(
|
||||
"Instance {Instance} subscribed to {Count} tags on connection {Connection}",
|
||||
_instanceUniqueName, tagPaths.Count, connectionName);
|
||||
}
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// WP-16: Handles alarm state changes from Alarm Actors.
|
||||
/// Updates in-memory alarm state and publishes to stream.
|
||||
/// </summary>
|
||||
private void HandleAlarmStateChanged(AlarmStateChanged changed)
|
||||
{
|
||||
_alarmStates[changed.AlarmName] = changed.State;
|
||||
_alarmTimestamps[changed.AlarmName] = changed.Timestamp;
|
||||
|
||||
// WP-23: Publish to site-wide stream
|
||||
_streamManager?.PublishAlarmStateChanged(changed);
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// WP-25: Debug view subscribe — returns snapshot and begins streaming.
|
||||
/// </summary>
|
||||
private void HandleSubscribeDebugView(SubscribeDebugViewRequest request)
|
||||
{
|
||||
// Build snapshot from current state
|
||||
var now = DateTimeOffset.UtcNow;
|
||||
var attributeValues = _attributes.Select(kvp => new AttributeValueChanged(
|
||||
_instanceUniqueName,
|
||||
kvp.Key,
|
||||
kvp.Key,
|
||||
kvp.Value,
|
||||
_attributeQualities.GetValueOrDefault(kvp.Key, "Good"),
|
||||
_attributeTimestamps.GetValueOrDefault(kvp.Key, now))).ToList();
|
||||
|
||||
var alarmStates = _alarmActors.Keys.Select(name => new AlarmStateChanged(
|
||||
_instanceUniqueName,
|
||||
name,
|
||||
_alarmStates.GetValueOrDefault(name, AlarmState.Normal),
|
||||
_alarmPriorities.GetValueOrDefault(name, 0),
|
||||
_alarmTimestamps[name])).ToList();
|
||||
|
||||
var snapshot = new DebugViewSnapshot(
|
||||
_instanceUniqueName,
|
||||
attributeValues,
|
||||
alarmStates,
|
||||
DateTimeOffset.UtcNow);
|
||||
|
||||
Sender.Tell(snapshot);
|
||||
|
||||
_logger.LogDebug(
|
||||
"Debug view snapshot sent for {Instance}, correlationId={Id}",
|
||||
_instanceUniqueName, request.CorrelationId);
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// WP-25: Debug view unsubscribe (SiteRuntime-013).
|
||||
/// This handler is a deliberate no-op acknowledgement: the Instance Actor holds
|
||||
/// no per-subscriber state. The real debug-stream subscription lifecycle lives in
|
||||
/// <see cref="ZB.MOM.WW.ScadaBridge.SiteRuntime.Streaming.SiteStreamManager"/>
|
||||
/// (Subscribe / Unsubscribe / RemoveSubscriber); the gRPC stream is torn down
|
||||
/// there when the central side cancels the call. Nothing is removed here.
|
||||
/// </summary>
|
||||
private void HandleUnsubscribeDebugView(UnsubscribeDebugViewRequest request)
|
||||
{
|
||||
// No subscription state in the Instance Actor — see the XML doc above.
|
||||
_logger.LogDebug(
|
||||
"Debug view unsubscribe for {Instance}, correlationId={Id} " +
|
||||
"(no-op; subscription teardown handled by SiteStreamManager)",
|
||||
_instanceUniqueName, request.CorrelationId);
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// One-shot debug snapshot — returns current state without registering a subscriber.
|
||||
/// </summary>
|
||||
private void HandleDebugSnapshot(DebugSnapshotRequest request)
|
||||
{
|
||||
var now = DateTimeOffset.UtcNow;
|
||||
var attributeValues = _attributes.Select(kvp => new AttributeValueChanged(
|
||||
_instanceUniqueName,
|
||||
kvp.Key,
|
||||
kvp.Key,
|
||||
kvp.Value,
|
||||
_attributeQualities.GetValueOrDefault(kvp.Key, "Good"),
|
||||
_attributeTimestamps.GetValueOrDefault(kvp.Key, now))).ToList();
|
||||
|
||||
var alarmStates = _alarmActors.Keys.Select(name => new AlarmStateChanged(
|
||||
_instanceUniqueName,
|
||||
name,
|
||||
_alarmStates.GetValueOrDefault(name, AlarmState.Normal),
|
||||
_alarmPriorities.GetValueOrDefault(name, 0),
|
||||
_alarmTimestamps[name])).ToList();
|
||||
|
||||
var snapshot = new DebugViewSnapshot(
|
||||
_instanceUniqueName,
|
||||
attributeValues,
|
||||
alarmStates,
|
||||
DateTimeOffset.UtcNow);
|
||||
|
||||
Sender.Tell(snapshot);
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// Publishes attribute change to stream and notifies child Script/Alarm actors.
|
||||
/// WP-22: Tell for attribute notifications (fire-and-forget, never blocks).
|
||||
/// </summary>
|
||||
private void PublishAndNotifyChildren(AttributeValueChanged changed)
|
||||
{
|
||||
// WP-23: Publish to site-wide stream
|
||||
_streamManager?.PublishAttributeValueChanged(changed);
|
||||
|
||||
// Notify Script Actors (for value-change and conditional triggers)
|
||||
foreach (var scriptActor in _scriptActors.Values)
|
||||
{
|
||||
scriptActor.Tell(changed);
|
||||
}
|
||||
|
||||
// Notify Alarm Actors (for alarm evaluation)
|
||||
foreach (var alarmActor in _alarmActors.Values)
|
||||
{
|
||||
alarmActor.Tell(changed);
|
||||
}
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// Applies static overrides loaded from SQLite on top of default values.
|
||||
/// </summary>
|
||||
private void HandleOverridesLoaded(LoadOverridesResult result)
|
||||
{
|
||||
if (result.Error != null)
|
||||
{
|
||||
_logger.LogWarning(
|
||||
"Failed to load static overrides for {Instance}: {Error}",
|
||||
_instanceUniqueName, result.Error);
|
||||
return;
|
||||
}
|
||||
|
||||
foreach (var kvp in result.Overrides)
|
||||
{
|
||||
_attributes[kvp.Key] = kvp.Value;
|
||||
}
|
||||
|
||||
_logger.LogDebug(
|
||||
"Loaded {Count} static overrides for {Instance}",
|
||||
result.Overrides.Count, _instanceUniqueName);
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// Creates child Script Actors and Alarm Actors from the flattened configuration.
|
||||
/// WP-15: Script Actors spawned per script definition.
|
||||
/// WP-16: Alarm Actors spawned per alarm definition, as peers to Script Actors.
|
||||
/// WP-32: Compilation errors reject entire instance deployment (logged but actor still starts).
|
||||
///
|
||||
/// SiteRuntime-017: each child is seeded from a private point-in-time snapshot
|
||||
/// of <c>_attributes</c>, NOT the live dictionary. The snapshot is taken here on
|
||||
/// the Instance Actor thread, so it is race-free; handing the live mutable
|
||||
/// <see cref="System.Collections.Generic.Dictionary{TKey,TValue}"/> by reference
|
||||
/// would let a child constructor enumerate it on the child's mailbox thread while
|
||||
/// this actor mutates it in <c>HandleAttributeValueChanged</c>.
|
||||
/// </summary>
|
||||
private void CreateChildActors()
|
||||
{
|
||||
if (_configuration == null) return;
|
||||
|
||||
// SiteRuntime-017: snapshot the live attribute dictionary once, on the
|
||||
// Instance Actor thread, before any child is constructed. Each child
|
||||
// Props closure captures this immutable copy instead of the mutable
|
||||
// _attributes field, so no child constructor ever enumerates a
|
||||
// dictionary this actor is concurrently mutating.
|
||||
var attributeSnapshot = new Dictionary<string, object?>(_attributes);
|
||||
|
||||
// Create Script Actors
|
||||
foreach (var script in _configuration.Scripts)
|
||||
{
|
||||
var compilationResult = _compilationService.Compile(script.CanonicalName, script.Code);
|
||||
if (!compilationResult.IsSuccess)
|
||||
{
|
||||
_logger.LogError(
|
||||
"Script '{Script}' on instance '{Instance}' failed to compile: {Errors}",
|
||||
script.CanonicalName, _instanceUniqueName,
|
||||
string.Join("; ", compilationResult.Errors));
|
||||
continue;
|
||||
}
|
||||
|
||||
// Compile the trigger expression for Expression-triggered scripts.
|
||||
var triggerExpression = CompileTriggerExpression(
|
||||
script.TriggerType, script.TriggerConfiguration, $"script-trigger-{script.CanonicalName}");
|
||||
|
||||
var props = Props.Create(() => new ScriptActor(
|
||||
script.CanonicalName,
|
||||
_instanceUniqueName,
|
||||
Self,
|
||||
compilationResult.CompiledScript,
|
||||
script,
|
||||
_sharedScriptLibrary,
|
||||
_options,
|
||||
_logger,
|
||||
triggerExpression,
|
||||
attributeSnapshot,
|
||||
_healthCollector,
|
||||
_serviceProvider));
|
||||
|
||||
var actorRef = Context.ActorOf(props, $"script-{script.CanonicalName}");
|
||||
_scriptActors[script.CanonicalName] = actorRef;
|
||||
}
|
||||
|
||||
// Create Alarm Actors
|
||||
foreach (var alarm in _configuration.Alarms)
|
||||
{
|
||||
Script<object?>? onTriggerScript = null;
|
||||
|
||||
// Compile on-trigger script if defined
|
||||
if (!string.IsNullOrEmpty(alarm.OnTriggerScriptCanonicalName))
|
||||
{
|
||||
var triggerScriptDef = _configuration.Scripts
|
||||
.FirstOrDefault(s => s.CanonicalName == alarm.OnTriggerScriptCanonicalName);
|
||||
|
||||
if (triggerScriptDef != null)
|
||||
{
|
||||
var result = _compilationService.Compile(
|
||||
$"alarm-trigger-{alarm.CanonicalName}", triggerScriptDef.Code);
|
||||
if (result.IsSuccess)
|
||||
{
|
||||
onTriggerScript = result.CompiledScript;
|
||||
}
|
||||
else
|
||||
{
|
||||
_logger.LogWarning(
|
||||
"Alarm trigger script for {Alarm} on {Instance} failed to compile",
|
||||
alarm.CanonicalName, _instanceUniqueName);
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// Compile the trigger expression for Expression-triggered alarms.
|
||||
var triggerExpression = CompileTriggerExpression(
|
||||
alarm.TriggerType, alarm.TriggerConfiguration, $"alarm-trigger-expr-{alarm.CanonicalName}");
|
||||
|
||||
var props = Props.Create(() => new AlarmActor(
|
||||
alarm.CanonicalName,
|
||||
_instanceUniqueName,
|
||||
Self,
|
||||
alarm,
|
||||
onTriggerScript,
|
||||
_sharedScriptLibrary,
|
||||
_options,
|
||||
_logger,
|
||||
triggerExpression,
|
||||
attributeSnapshot,
|
||||
_healthCollector));
|
||||
|
||||
var actorRef = Context.ActorOf(props, $"alarm-{alarm.CanonicalName}");
|
||||
_alarmActors[alarm.CanonicalName] = actorRef;
|
||||
_alarmPriorities[alarm.CanonicalName] = alarm.PriorityLevel;
|
||||
_alarmTimestamps[alarm.CanonicalName] = DateTimeOffset.UtcNow;
|
||||
}
|
||||
|
||||
_logger.LogInformation(
|
||||
"Instance {Instance}: created {Scripts} script actors and {Alarms} alarm actors",
|
||||
_instanceUniqueName, _scriptActors.Count, _alarmActors.Count);
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// Compiles the boolean trigger expression for an Expression-triggered
|
||||
/// script or alarm. Returns null for non-Expression triggers, a blank
|
||||
/// expression, or a compilation failure (logged) — in which case the
|
||||
/// trigger is inert and the actor still starts.
|
||||
/// </summary>
|
||||
private Script<object?>? CompileTriggerExpression(
|
||||
string? triggerType, string? triggerConfigJson, string compileName)
|
||||
{
|
||||
if (!string.Equals(triggerType, "Expression", StringComparison.OrdinalIgnoreCase))
|
||||
return null;
|
||||
|
||||
var expression = TriggerExpressionGlobals.ExtractExpression(triggerConfigJson);
|
||||
if (expression == null)
|
||||
return null;
|
||||
|
||||
var result = _compilationService.CompileTriggerExpression(compileName, expression);
|
||||
if (result.IsSuccess)
|
||||
return result.CompiledScript;
|
||||
|
||||
_logger.LogError(
|
||||
"Trigger expression for {Name} on {Instance} failed to compile: {Errors}",
|
||||
compileName, _instanceUniqueName, string.Join("; ", result.Errors));
|
||||
return null;
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// Read-only access to current attribute count (for testing/diagnostics).
|
||||
/// </summary>
|
||||
public int AttributeCount => _attributes.Count;
|
||||
|
||||
/// <summary>
|
||||
/// Read-only access to script actor count (for testing/diagnostics).
|
||||
/// </summary>
|
||||
public int ScriptActorCount => _scriptActors.Count;
|
||||
|
||||
/// <summary>
|
||||
/// Read-only access to alarm actor count (for testing/diagnostics).
|
||||
/// </summary>
|
||||
public int AlarmActorCount => _alarmActors.Count;
|
||||
|
||||
/// <summary>
|
||||
/// Internal message for async override loading result.
|
||||
/// </summary>
|
||||
internal record LoadOverridesResult(Dictionary<string, string> Overrides, string? Error);
|
||||
}
|
||||
@@ -0,0 +1,584 @@
|
||||
using Akka.Actor;
|
||||
using Microsoft.CodeAnalysis.Scripting;
|
||||
using Microsoft.Extensions.DependencyInjection;
|
||||
using Microsoft.Extensions.Logging;
|
||||
using ZB.MOM.WW.ScadaBridge.Commons.Messages.ScriptExecution;
|
||||
using ZB.MOM.WW.ScadaBridge.Commons.Messages.Streaming;
|
||||
using ZB.MOM.WW.ScadaBridge.Commons.Types.Flattening;
|
||||
using ZB.MOM.WW.ScadaBridge.HealthMonitoring;
|
||||
using ZB.MOM.WW.ScadaBridge.SiteEventLogging;
|
||||
using ZB.MOM.WW.ScadaBridge.SiteRuntime.Scripts;
|
||||
using System.Globalization;
|
||||
using System.Text.Json;
|
||||
|
||||
namespace ZB.MOM.WW.ScadaBridge.SiteRuntime.Actors;
|
||||
|
||||
/// <summary>
|
||||
/// WP-15: Script Actor — coordinator actor, child of Instance Actor.
|
||||
/// Holds compiled script delegate, manages trigger configuration, and spawns
|
||||
/// ScriptExecutionActor children per invocation. Does not block on child completion.
|
||||
///
|
||||
/// Trigger types:
|
||||
/// - Interval: uses Akka timers to fire periodically
|
||||
/// - ValueChange: receives attribute change notifications from Instance Actor
|
||||
/// - Conditional: evaluates a threshold comparison on attribute change
|
||||
/// - Expression: evaluates a compiled boolean expression on attribute change
|
||||
/// Conditional and Expression triggers carry a <see cref="TriggerMode"/>:
|
||||
/// OnTrue fires as the condition becomes true; WhileTrue additionally re-fires
|
||||
/// on a timer (cadence = MinTimeBetweenRuns) while the condition stays true.
|
||||
///
|
||||
/// Supervision strategy: Resume on exception (coordinator preserves state).
|
||||
/// </summary>
|
||||
public class ScriptActor : ReceiveActor, IWithTimers
|
||||
{
|
||||
private readonly string _scriptName;
|
||||
private readonly string _instanceName;
|
||||
private readonly IActorRef _instanceActor;
|
||||
private readonly SharedScriptLibrary _sharedScriptLibrary;
|
||||
private readonly SiteRuntimeOptions _options;
|
||||
private readonly ILogger _logger;
|
||||
private readonly ISiteHealthCollector? _healthCollector;
|
||||
private readonly IServiceProvider? _serviceProvider;
|
||||
|
||||
private Script<object?>? _compiledScript;
|
||||
private ScriptTriggerConfig? _triggerConfig;
|
||||
private TimeSpan? _minTimeBetweenRuns;
|
||||
private DateTimeOffset _lastExecutionTime = DateTimeOffset.MinValue;
|
||||
private int _executionCounter;
|
||||
private readonly Commons.Types.Scripts.ScriptScope _scope;
|
||||
|
||||
// Expression trigger state: compiled expression, edge-tracking, and the
|
||||
// attribute snapshot the expression evaluates against.
|
||||
private readonly Script<object?>? _compiledTriggerExpression;
|
||||
private bool _lastExpressionResult;
|
||||
private readonly Dictionary<string, object?> _attributeSnapshot = new();
|
||||
|
||||
// WhileTrue trigger state: the most recent truth value of a Conditional
|
||||
// trigger's comparison, used to detect false->true / true->false edges.
|
||||
// (Expression triggers reuse _lastExpressionResult for the same purpose.)
|
||||
private bool _conditionState;
|
||||
|
||||
/// <summary>Timer key for the WhileTrue re-fire timer (cadence = MinTimeBetweenRuns).</summary>
|
||||
private const string WhileTrueTimerKey = "whiletrue-trigger";
|
||||
|
||||
/// <summary>
|
||||
/// SiteRuntime-017: the exact dictionary instance this actor was seeded from
|
||||
/// at construction. The Instance Actor must pass a private snapshot here, not
|
||||
/// its live <c>_attributes</c> field — sharing the live dictionary lets this
|
||||
/// constructor enumerate it while the Instance Actor mutates it on another
|
||||
/// thread. Exposed for regression coverage of that isolation contract.
|
||||
/// </summary>
|
||||
internal IReadOnlyDictionary<string, object?>? SeedAttributesReference { get; }
|
||||
|
||||
/// <summary>Gets or sets the Akka timer scheduler used to schedule interval and WhileTrue triggers.</summary>
|
||||
public ITimerScheduler Timers { get; set; } = null!;
|
||||
|
||||
/// <summary>
|
||||
/// Initializes the ScriptActor with its compiled script, trigger configuration, and supporting services.
|
||||
/// </summary>
|
||||
/// <param name="scriptName">Name of the script this actor manages.</param>
|
||||
/// <param name="instanceName">Unique name of the owning instance.</param>
|
||||
/// <param name="instanceActor">Reference to the parent Instance Actor.</param>
|
||||
/// <param name="compiledScript">Pre-compiled Roslyn script delegate, or null when compilation failed.</param>
|
||||
/// <param name="scriptConfig">Resolved script metadata including trigger type and configuration.</param>
|
||||
/// <param name="sharedScriptLibrary">Library of compiled shared scripts available for inline execution.</param>
|
||||
/// <param name="options">Site runtime configuration options.</param>
|
||||
/// <param name="logger">Logger for diagnostics.</param>
|
||||
/// <param name="compiledTriggerExpression">Pre-compiled boolean trigger expression, or null when not an expression trigger.</param>
|
||||
/// <param name="initialAttributes">Initial attribute snapshot used to seed expression trigger evaluation state.</param>
|
||||
/// <param name="healthCollector">Optional health metrics collector.</param>
|
||||
/// <param name="serviceProvider">Optional DI service provider for script execution context services.</param>
|
||||
public ScriptActor(
|
||||
string scriptName,
|
||||
string instanceName,
|
||||
IActorRef instanceActor,
|
||||
Script<object?>? compiledScript,
|
||||
ResolvedScript scriptConfig,
|
||||
SharedScriptLibrary sharedScriptLibrary,
|
||||
SiteRuntimeOptions options,
|
||||
ILogger logger,
|
||||
Script<object?>? compiledTriggerExpression = null,
|
||||
IReadOnlyDictionary<string, object?>? initialAttributes = null,
|
||||
ISiteHealthCollector? healthCollector = null,
|
||||
IServiceProvider? serviceProvider = null)
|
||||
{
|
||||
_scriptName = scriptName;
|
||||
_instanceName = instanceName;
|
||||
_instanceActor = instanceActor;
|
||||
_compiledScript = compiledScript;
|
||||
_sharedScriptLibrary = sharedScriptLibrary;
|
||||
_options = options;
|
||||
_logger = logger;
|
||||
_healthCollector = healthCollector;
|
||||
_serviceProvider = serviceProvider;
|
||||
_minTimeBetweenRuns = scriptConfig.MinTimeBetweenRuns;
|
||||
_scope = scriptConfig.Scope;
|
||||
_compiledTriggerExpression = compiledTriggerExpression;
|
||||
|
||||
// Seed the trigger-expression attribute snapshot from the instance's
|
||||
// initial attribute set so static attributes (which never re-emit an
|
||||
// AttributeValueChanged after deploy) evaluate correctly at startup.
|
||||
SeedAttributesReference = initialAttributes;
|
||||
if (initialAttributes != null)
|
||||
{
|
||||
foreach (var kvp in initialAttributes)
|
||||
_attributeSnapshot[kvp.Key] = kvp.Value;
|
||||
}
|
||||
|
||||
// Parse trigger configuration
|
||||
_triggerConfig = ParseTriggerConfig(scriptConfig.TriggerType, scriptConfig.TriggerConfiguration);
|
||||
|
||||
// Handle script call requests (Ask pattern from Instance Actor or ScriptRuntimeContext)
|
||||
Receive<ScriptCallRequest>(HandleScriptCallRequest);
|
||||
|
||||
// Handle attribute value changes for value-change and conditional triggers
|
||||
Receive<AttributeValueChanged>(HandleAttributeValueChanged);
|
||||
|
||||
// Handle interval tick
|
||||
Receive<IntervalTick>(_ => TrySpawnExecution(null));
|
||||
|
||||
// Handle WhileTrue re-fire tick
|
||||
Receive<WhileTrueTick>(_ => FireWhileTrueTick());
|
||||
|
||||
// Handle execution completion (for logging/metrics)
|
||||
Receive<ScriptExecutionCompleted>(HandleExecutionCompleted);
|
||||
}
|
||||
|
||||
/// <inheritdoc />
|
||||
protected override void PreStart()
|
||||
{
|
||||
base.PreStart();
|
||||
|
||||
// Set up interval trigger if configured
|
||||
if (_triggerConfig is IntervalTriggerConfig interval)
|
||||
{
|
||||
Timers.StartPeriodicTimer(
|
||||
"interval-trigger",
|
||||
IntervalTick.Instance,
|
||||
interval.Interval,
|
||||
interval.Interval);
|
||||
|
||||
_logger.LogDebug(
|
||||
"ScriptActor {Script} on {Instance}: interval trigger set to {Interval}",
|
||||
_scriptName, _instanceName, interval.Interval);
|
||||
}
|
||||
|
||||
_logger.LogInformation(
|
||||
"ScriptActor {Script} started on instance {Instance}",
|
||||
_scriptName, _instanceName);
|
||||
}
|
||||
|
||||
/// <inheritdoc />
|
||||
protected override SupervisorStrategy SupervisorStrategy()
|
||||
{
|
||||
return new OneForOneStrategy(
|
||||
maxNrOfRetries: -1,
|
||||
withinTimeRange: TimeSpan.FromMinutes(1),
|
||||
decider: Decider.From(ex =>
|
||||
{
|
||||
_logger.LogWarning(ex,
|
||||
"ScriptExecutionActor for {Script} on {Instance} failed, stopping",
|
||||
_scriptName, _instanceName);
|
||||
return Directive.Stop;
|
||||
}));
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// Handles CallScript ask from ScriptRuntimeContext or Instance Actor.
|
||||
/// Spawns a ScriptExecutionActor and forwards the sender for reply.
|
||||
/// </summary>
|
||||
private void HandleScriptCallRequest(ScriptCallRequest request)
|
||||
{
|
||||
if (_compiledScript == null)
|
||||
{
|
||||
Sender.Tell(new ScriptCallResult(
|
||||
request.CorrelationId,
|
||||
false,
|
||||
null,
|
||||
$"Script '{_scriptName}' is not compiled."));
|
||||
return;
|
||||
}
|
||||
|
||||
// Audit Log #23 (ParentExecutionId): carry any inbound-routed
|
||||
// ParentExecutionId through to the ScriptExecutionActor so the routed
|
||||
// script's ScriptRuntimeContext can record its spawner. Null for normal
|
||||
// (tag-change / timer) runs and nested Script.Call invocations.
|
||||
SpawnExecution(
|
||||
request.Parameters, request.CurrentCallDepth, Sender, request.CorrelationId,
|
||||
request.ParentExecutionId);
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// Handles attribute value changes — triggers script if configured for
|
||||
/// value-change, conditional, or expression. The attribute snapshot is
|
||||
/// updated for every change before any trigger logic runs.
|
||||
/// </summary>
|
||||
private void HandleAttributeValueChanged(AttributeValueChanged changed)
|
||||
{
|
||||
// Keep the snapshot current for every change, regardless of trigger type.
|
||||
_attributeSnapshot[changed.AttributeName] = changed.Value;
|
||||
|
||||
if (_triggerConfig is ValueChangeTriggerConfig valueTrigger)
|
||||
{
|
||||
if (valueTrigger.AttributeName == changed.AttributeName)
|
||||
{
|
||||
TrySpawnExecution(null);
|
||||
}
|
||||
}
|
||||
else if (_triggerConfig is ConditionalTriggerConfig conditional)
|
||||
{
|
||||
if (conditional.AttributeName == changed.AttributeName)
|
||||
{
|
||||
var conditionMet = EvaluateCondition(conditional, changed.Value);
|
||||
if (conditional.Mode == TriggerMode.WhileTrue)
|
||||
{
|
||||
// Edge-detect against the prior truth value; the timer does
|
||||
// the repeated firing while the condition stays true.
|
||||
HandleWhileTrueTransition(conditionMet, _conditionState);
|
||||
_conditionState = conditionMet;
|
||||
}
|
||||
else if (conditionMet)
|
||||
{
|
||||
// OnTrue: fire on each matching change (existing behavior).
|
||||
TrySpawnExecution(null);
|
||||
}
|
||||
}
|
||||
}
|
||||
else if (_triggerConfig is ExpressionTriggerConfig)
|
||||
{
|
||||
EvaluateExpressionTrigger();
|
||||
}
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// Evaluates the compiled trigger expression against the current attribute
|
||||
/// snapshot. In <see cref="TriggerMode.OnTrue"/> mode the script runs once
|
||||
/// per false→true transition; in <see cref="TriggerMode.WhileTrue"/> mode it
|
||||
/// fires on the edge and the re-fire timer is started/stopped with the
|
||||
/// expression's truth value. A throwing or non-bool expression is treated as
|
||||
/// false and logged as a script error; the actor never crashes.
|
||||
/// </summary>
|
||||
private void EvaluateExpressionTrigger()
|
||||
{
|
||||
if (_compiledTriggerExpression == null) return;
|
||||
if (_triggerConfig is not ExpressionTriggerConfig exprConfig) return;
|
||||
|
||||
bool result;
|
||||
try
|
||||
{
|
||||
var globals = new TriggerExpressionGlobals(_attributeSnapshot);
|
||||
// Bound evaluation with a short timeout. The CancellationToken
|
||||
// covers cooperative/async cases; a pathological CPU-bound
|
||||
// expression is not fully interruptible. Acceptable because
|
||||
// trigger expressions are authored by trusted Design-role users
|
||||
// and are compile-checked pre-deployment.
|
||||
using var cts = new CancellationTokenSource(TimeSpan.FromSeconds(2));
|
||||
var state = _compiledTriggerExpression
|
||||
.RunAsync(globals, cancellationToken: cts.Token)
|
||||
.GetAwaiter().GetResult();
|
||||
result = state.ReturnValue is bool b && b;
|
||||
}
|
||||
catch (Exception ex)
|
||||
{
|
||||
// OperationCanceledException (timeout) falls through here too,
|
||||
// and is correctly treated as false.
|
||||
LogExpressionError(ex);
|
||||
result = false;
|
||||
}
|
||||
|
||||
if (exprConfig.Mode == TriggerMode.WhileTrue)
|
||||
{
|
||||
HandleWhileTrueTransition(result, _lastExpressionResult);
|
||||
}
|
||||
else if (result && !_lastExpressionResult)
|
||||
{
|
||||
TrySpawnExecution(null);
|
||||
}
|
||||
|
||||
_lastExpressionResult = result;
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// Applies a WhileTrue trigger's condition-state transition: on the
|
||||
/// false→true edge, fire once and start the re-fire timer; on the
|
||||
/// true→false edge, stop the timer. While the state is unchanged, the
|
||||
/// already-running timer continues to drive re-firing.
|
||||
/// </summary>
|
||||
private void HandleWhileTrueTransition(bool nowTrue, bool wasTrue)
|
||||
{
|
||||
if (nowTrue && !wasTrue)
|
||||
{
|
||||
TrySpawnExecution(null);
|
||||
StartWhileTrueTimer();
|
||||
}
|
||||
else if (!nowTrue && wasTrue)
|
||||
{
|
||||
StopWhileTrueTimer();
|
||||
}
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// Starts the periodic WhileTrue re-fire timer. The cadence is the script's
|
||||
/// <c>MinTimeBetweenRuns</c>; with none configured the trigger cannot
|
||||
/// re-fire, so it degrades to the single edge fire and logs a warning.
|
||||
/// </summary>
|
||||
private void StartWhileTrueTimer()
|
||||
{
|
||||
if (_compiledScript == null) return;
|
||||
|
||||
if (_minTimeBetweenRuns is not { } interval)
|
||||
{
|
||||
_logger.LogWarning(
|
||||
"ScriptActor {Script} on {Instance}: WhileTrue trigger has no MinTimeBetweenRuns — " +
|
||||
"firing once on the edge only, no re-fire timer.",
|
||||
_scriptName, _instanceName);
|
||||
return;
|
||||
}
|
||||
|
||||
Timers.StartPeriodicTimer(WhileTrueTimerKey, WhileTrueTick.Instance, interval, interval);
|
||||
}
|
||||
|
||||
/// <summary>Cancels the WhileTrue re-fire timer (a no-op if it is not running).</summary>
|
||||
private void StopWhileTrueTimer() => Timers.Cancel(WhileTrueTimerKey);
|
||||
|
||||
/// <summary>
|
||||
/// Fires the script for a WhileTrue re-fire tick. The timer interval is
|
||||
/// itself the cadence, so this spawns directly — bypassing the
|
||||
/// MinTimeBetweenRuns skip-check that gates change-driven spawns (which
|
||||
/// could otherwise drop a tick to sub-millisecond timing jitter).
|
||||
/// </summary>
|
||||
private void FireWhileTrueTick()
|
||||
{
|
||||
if (_compiledScript == null) return;
|
||||
|
||||
_lastExecutionTime = DateTimeOffset.UtcNow;
|
||||
SpawnExecution(null, 0, ActorRefs.NoSender!, Guid.NewGuid().ToString());
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// Records a trigger-expression evaluation failure to the site event log,
|
||||
/// mirroring how ScriptExecutionActor reports script errors.
|
||||
/// </summary>
|
||||
private void LogExpressionError(Exception ex)
|
||||
{
|
||||
_healthCollector?.IncrementScriptError();
|
||||
var errorMsg = $"Trigger expression for script '{_scriptName}' on instance '{_instanceName}' failed: {ex.Message}";
|
||||
_logger.LogError(ex, "Trigger expression evaluation failed: {Script} on {Instance}", _scriptName, _instanceName);
|
||||
|
||||
_ = _serviceProvider?.GetService<ISiteEventLogger>()?.LogEventAsync(
|
||||
"script", "Error", _instanceName, $"ScriptActor:{_scriptName}", errorMsg, ex.ToString());
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// Attempts to spawn a script execution, respecting MinTimeBetweenRuns.
|
||||
/// </summary>
|
||||
private void TrySpawnExecution(IReadOnlyDictionary<string, object?>? parameters)
|
||||
{
|
||||
if (_compiledScript == null) return;
|
||||
|
||||
if (_minTimeBetweenRuns.HasValue)
|
||||
{
|
||||
var elapsed = DateTimeOffset.UtcNow - _lastExecutionTime;
|
||||
if (elapsed < _minTimeBetweenRuns.Value)
|
||||
{
|
||||
_logger.LogDebug(
|
||||
"Script {Script} on {Instance}: skipping execution, min time between runs not elapsed ({Elapsed} < {Min})",
|
||||
_scriptName, _instanceName, elapsed, _minTimeBetweenRuns.Value);
|
||||
return;
|
||||
}
|
||||
}
|
||||
|
||||
_lastExecutionTime = DateTimeOffset.UtcNow;
|
||||
SpawnExecution(parameters, 0, ActorRefs.NoSender!, Guid.NewGuid().ToString());
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// Spawns a new ScriptExecutionActor child for this invocation.
|
||||
/// Multiple concurrent executions are allowed.
|
||||
/// </summary>
|
||||
private void SpawnExecution(
|
||||
IReadOnlyDictionary<string, object?>? parameters,
|
||||
int callDepth,
|
||||
IActorRef replyTo,
|
||||
string correlationId,
|
||||
Guid? parentExecutionId = null)
|
||||
{
|
||||
var executionId = $"{_scriptName}-exec-{_executionCounter++}";
|
||||
|
||||
// SiteRuntime-009: the actor's mailbox stays on the default dispatcher, but the
|
||||
// script body itself runs on the dedicated ScriptExecutionScheduler (a bounded
|
||||
// set of dedicated threads), so blocking script I/O is contained there and
|
||||
// cannot starve the shared .NET thread pool.
|
||||
var props = Props.Create(() => new ScriptExecutionActor(
|
||||
_scriptName,
|
||||
_instanceName,
|
||||
_compiledScript!,
|
||||
parameters,
|
||||
callDepth,
|
||||
_instanceActor,
|
||||
_sharedScriptLibrary,
|
||||
_options,
|
||||
replyTo,
|
||||
correlationId,
|
||||
_logger,
|
||||
_scope,
|
||||
_healthCollector,
|
||||
_serviceProvider,
|
||||
// Audit Log #23 (ParentExecutionId): null for trigger-driven runs;
|
||||
// an inbound-API-routed call supplies the inbound request's id.
|
||||
parentExecutionId));
|
||||
|
||||
Context.ActorOf(props, executionId);
|
||||
}
|
||||
|
||||
private void HandleExecutionCompleted(ScriptExecutionCompleted msg)
|
||||
{
|
||||
_logger.LogDebug(
|
||||
"Script {Script} execution completed on {Instance}: success={Success}",
|
||||
_scriptName, _instanceName, msg.Success);
|
||||
}
|
||||
|
||||
private static bool EvaluateCondition(ConditionalTriggerConfig config, object? value)
|
||||
{
|
||||
if (value == null) return false;
|
||||
|
||||
try
|
||||
{
|
||||
// Use InvariantCulture so a string attribute value like "1.5" parses
|
||||
// consistently regardless of the host locale (SiteRuntime-023). For
|
||||
// purely-numeric inputs the culture argument is a no-op, but it is
|
||||
// safe and future-proof for string-typed attribute values arriving
|
||||
// from scripts or the data connection layer.
|
||||
var numericValue = Convert.ToDouble(value, CultureInfo.InvariantCulture);
|
||||
return config.Operator switch
|
||||
{
|
||||
">" => numericValue > config.Threshold,
|
||||
">=" => numericValue >= config.Threshold,
|
||||
"<" => numericValue < config.Threshold,
|
||||
"<=" => numericValue <= config.Threshold,
|
||||
"==" => Math.Abs(numericValue - config.Threshold) < 0.0001,
|
||||
"!=" => Math.Abs(numericValue - config.Threshold) >= 0.0001,
|
||||
_ => false
|
||||
};
|
||||
}
|
||||
catch
|
||||
{
|
||||
return string.Equals(value.ToString(), config.Threshold.ToString(), StringComparison.Ordinal);
|
||||
}
|
||||
}
|
||||
|
||||
private static ScriptTriggerConfig? ParseTriggerConfig(string? triggerType, string? triggerConfigJson)
|
||||
{
|
||||
if (string.IsNullOrEmpty(triggerType)) return null;
|
||||
|
||||
return triggerType.ToLowerInvariant() switch
|
||||
{
|
||||
"interval" => ParseIntervalTrigger(triggerConfigJson),
|
||||
"valuechange" => ParseValueChangeTrigger(triggerConfigJson),
|
||||
"conditional" => ParseConditionalTrigger(triggerConfigJson),
|
||||
"expression" => ParseExpressionTrigger(triggerConfigJson),
|
||||
"call" => null, // No automatic trigger — invoked only via Instance.CallScript()
|
||||
_ => null
|
||||
};
|
||||
}
|
||||
|
||||
private static ExpressionTriggerConfig? ParseExpressionTrigger(string? json)
|
||||
{
|
||||
var expr = TriggerExpressionGlobals.ExtractExpression(json);
|
||||
if (expr == null) return null;
|
||||
|
||||
// ExtractExpression already proved the JSON parses; read the mode too.
|
||||
var mode = TriggerMode.OnTrue;
|
||||
try
|
||||
{
|
||||
using var doc = JsonDocument.Parse(json!);
|
||||
mode = ParseTriggerMode(doc.RootElement);
|
||||
}
|
||||
catch (JsonException) { /* keep OnTrue */ }
|
||||
|
||||
return new ExpressionTriggerConfig(expr, mode);
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// Reads the optional <c>mode</c> field (Conditional + Expression triggers).
|
||||
/// An absent or unrecognized value (case-insensitive) yields
|
||||
/// <see cref="TriggerMode.OnTrue"/>, so pre-WhileTrue configs are unchanged.
|
||||
/// </summary>
|
||||
private static TriggerMode ParseTriggerMode(JsonElement root)
|
||||
{
|
||||
var raw = root.TryGetProperty("mode", out var m) ? m.GetString() : null;
|
||||
return string.Equals(raw?.Trim(), "WhileTrue", StringComparison.OrdinalIgnoreCase)
|
||||
? TriggerMode.WhileTrue
|
||||
: TriggerMode.OnTrue;
|
||||
}
|
||||
|
||||
private static IntervalTriggerConfig? ParseIntervalTrigger(string? json)
|
||||
{
|
||||
if (string.IsNullOrEmpty(json)) return null;
|
||||
try
|
||||
{
|
||||
var doc = JsonDocument.Parse(json);
|
||||
var ms = doc.RootElement.GetProperty("intervalMs").GetInt64();
|
||||
return new IntervalTriggerConfig(TimeSpan.FromMilliseconds(ms));
|
||||
}
|
||||
catch { return null; }
|
||||
}
|
||||
|
||||
private static ValueChangeTriggerConfig? ParseValueChangeTrigger(string? json)
|
||||
{
|
||||
if (string.IsNullOrEmpty(json)) return null;
|
||||
try
|
||||
{
|
||||
var doc = JsonDocument.Parse(json);
|
||||
var attr = doc.RootElement.GetProperty("attributeName").GetString()!;
|
||||
return new ValueChangeTriggerConfig(attr);
|
||||
}
|
||||
catch { return null; }
|
||||
}
|
||||
|
||||
private static ConditionalTriggerConfig? ParseConditionalTrigger(string? json)
|
||||
{
|
||||
if (string.IsNullOrEmpty(json)) return null;
|
||||
try
|
||||
{
|
||||
var doc = JsonDocument.Parse(json);
|
||||
var attr = doc.RootElement.GetProperty("attributeName").GetString()!;
|
||||
var op = doc.RootElement.GetProperty("operator").GetString()!;
|
||||
var threshold = doc.RootElement.GetProperty("threshold").GetDouble();
|
||||
return new ConditionalTriggerConfig(
|
||||
attr, op, threshold, ParseTriggerMode(doc.RootElement));
|
||||
}
|
||||
catch { return null; }
|
||||
}
|
||||
|
||||
// ── Internal messages ──
|
||||
|
||||
internal sealed class IntervalTick
|
||||
{
|
||||
public static readonly IntervalTick Instance = new();
|
||||
private IntervalTick() { }
|
||||
}
|
||||
|
||||
internal sealed class WhileTrueTick
|
||||
{
|
||||
public static readonly WhileTrueTick Instance = new();
|
||||
private WhileTrueTick() { }
|
||||
}
|
||||
|
||||
internal record ScriptExecutionCompleted(string ScriptName, bool Success, string? Error);
|
||||
}
|
||||
|
||||
// ── Trigger config types ──
|
||||
|
||||
/// <summary>
|
||||
/// When a Conditional/Expression trigger fires. <see cref="OnTrue"/> fires once
|
||||
/// as the condition becomes true; <see cref="WhileTrue"/> additionally re-fires
|
||||
/// on a timer (cadence = the script's MinTimeBetweenRuns) until it goes false.
|
||||
/// </summary>
|
||||
internal enum TriggerMode { OnTrue, WhileTrue }
|
||||
|
||||
internal record IntervalTriggerConfig(TimeSpan Interval) : ScriptTriggerConfig;
|
||||
internal record ValueChangeTriggerConfig(string AttributeName) : ScriptTriggerConfig;
|
||||
internal record ConditionalTriggerConfig(string AttributeName, string Operator, double Threshold, TriggerMode Mode) : ScriptTriggerConfig;
|
||||
internal record ExpressionTriggerConfig(string Expression, TriggerMode Mode) : ScriptTriggerConfig;
|
||||
internal abstract record ScriptTriggerConfig;
|
||||
@@ -0,0 +1,274 @@
|
||||
using Akka.Actor;
|
||||
using Microsoft.CodeAnalysis.Scripting;
|
||||
using Microsoft.Extensions.DependencyInjection;
|
||||
using Microsoft.Extensions.Logging;
|
||||
using ZB.MOM.WW.ScadaBridge.Commons.Interfaces;
|
||||
using ZB.MOM.WW.ScadaBridge.Commons.Interfaces.Services;
|
||||
using ZB.MOM.WW.ScadaBridge.Commons.Messages.ScriptExecution;
|
||||
using ZB.MOM.WW.ScadaBridge.Commons.Types;
|
||||
using ZB.MOM.WW.ScadaBridge.HealthMonitoring;
|
||||
using ZB.MOM.WW.ScadaBridge.SiteEventLogging;
|
||||
using ZB.MOM.WW.ScadaBridge.SiteRuntime.Scripts;
|
||||
using ZB.MOM.WW.ScadaBridge.StoreAndForward;
|
||||
|
||||
namespace ZB.MOM.WW.ScadaBridge.SiteRuntime.Actors;
|
||||
|
||||
/// <summary>
|
||||
/// WP-15: Script Execution Actor -- short-lived child of Script Actor.
|
||||
/// Receives compiled code, params, Instance Actor ref, and call depth.
|
||||
/// Executes the script via Script Runtime API, returns result, then stops.
|
||||
///
|
||||
/// The actor itself and its mailbox run on the default Akka dispatcher; only the
|
||||
/// script body is dispatched off the actor thread, onto the dedicated
|
||||
/// <see cref="ZB.MOM.WW.ScadaBridge.SiteRuntime.Scripts.ScriptExecutionScheduler"/>
|
||||
/// (SiteRuntime-009), so blocking script I/O cannot starve the shared thread pool
|
||||
/// or stall other Akka dispatchers.
|
||||
///
|
||||
/// WP-32: Script failures are logged but do not disable the script.
|
||||
/// Supervision: Stop on unhandled exception (parent ScriptActor decides).
|
||||
/// </summary>
|
||||
public class ScriptExecutionActor : ReceiveActor
|
||||
{
|
||||
/// <summary>
|
||||
/// Initializes the actor and immediately begins script execution on construction.
|
||||
/// </summary>
|
||||
/// <param name="scriptName">Name of the script being executed.</param>
|
||||
/// <param name="instanceName">Name of the instance that owns the script.</param>
|
||||
/// <param name="compiledScript">Compiled Roslyn script to execute.</param>
|
||||
/// <param name="parameters">Optional named parameter values for the script.</param>
|
||||
/// <param name="callDepth">Current call-nesting depth (used to enforce the max-depth limit).</param>
|
||||
/// <param name="instanceActor">Parent instance actor reference for attribute access.</param>
|
||||
/// <param name="sharedScriptLibrary">Library of shared scripts available during execution.</param>
|
||||
/// <param name="options">Site runtime options applied during execution.</param>
|
||||
/// <param name="replyTo">Actor reference that receives the script result.</param>
|
||||
/// <param name="correlationId">Application-level correlation id threaded through the execution.</param>
|
||||
/// <param name="logger">Logger for script execution events.</param>
|
||||
/// <param name="scope">Script scope controlling which APIs are available.</param>
|
||||
/// <param name="healthCollector">Optional health collector for recording execution metrics.</param>
|
||||
/// <param name="serviceProvider">Optional DI service provider for script execution services.</param>
|
||||
/// <param name="parentExecutionId">ExecutionId of the spawning inbound-API execution for audit correlation; null for normal runs.</param>
|
||||
public ScriptExecutionActor(
|
||||
string scriptName,
|
||||
string instanceName,
|
||||
Script<object?> compiledScript,
|
||||
IReadOnlyDictionary<string, object?>? parameters,
|
||||
int callDepth,
|
||||
IActorRef instanceActor,
|
||||
SharedScriptLibrary sharedScriptLibrary,
|
||||
SiteRuntimeOptions options,
|
||||
IActorRef replyTo,
|
||||
string correlationId,
|
||||
ILogger logger,
|
||||
Commons.Types.Scripts.ScriptScope scope,
|
||||
ISiteHealthCollector? healthCollector = null,
|
||||
IServiceProvider? serviceProvider = null,
|
||||
// Audit Log #23 (ParentExecutionId): the spawning execution's
|
||||
// ExecutionId for an inbound-API-routed call. Null for normal
|
||||
// (tag-change / timer) runs and nested Script.Call invocations.
|
||||
Guid? parentExecutionId = null)
|
||||
{
|
||||
// Immediately begin execution
|
||||
var self = Self;
|
||||
var parent = Context.Parent;
|
||||
|
||||
ExecuteScript(
|
||||
scriptName, instanceName, compiledScript, parameters, callDepth,
|
||||
instanceActor, sharedScriptLibrary, options, replyTo, correlationId,
|
||||
self, parent, logger, scope, healthCollector, serviceProvider,
|
||||
parentExecutionId);
|
||||
}
|
||||
|
||||
private static void ExecuteScript(
|
||||
string scriptName,
|
||||
string instanceName,
|
||||
Script<object?> compiledScript,
|
||||
IReadOnlyDictionary<string, object?>? parameters,
|
||||
int callDepth,
|
||||
IActorRef instanceActor,
|
||||
SharedScriptLibrary sharedScriptLibrary,
|
||||
SiteRuntimeOptions options,
|
||||
IActorRef replyTo,
|
||||
string correlationId,
|
||||
IActorRef self,
|
||||
IActorRef parent,
|
||||
ILogger logger,
|
||||
Commons.Types.Scripts.ScriptScope scope,
|
||||
ISiteHealthCollector? healthCollector,
|
||||
IServiceProvider? serviceProvider,
|
||||
Guid? parentExecutionId)
|
||||
{
|
||||
var timeout = TimeSpan.FromSeconds(options.ScriptExecutionTimeoutSeconds);
|
||||
|
||||
// SiteRuntime-009: run the script body on the dedicated script-execution
|
||||
// scheduler, not the shared .NET thread pool, so blocking script I/O cannot
|
||||
// starve the global pool and stall Akka dispatchers / HTTP handling.
|
||||
var scheduler = ScriptExecutionScheduler.Shared(options);
|
||||
|
||||
// Notification Outbox: the site communication actor that Notify.Status queries
|
||||
// central through. Resolved by actor path so the Notify helper does not need an
|
||||
// IActorRef threaded all the way down from the host wiring.
|
||||
var siteCommunicationActor = Context.System.ActorSelection("/user/site-communication");
|
||||
|
||||
// CTS must be created inside the async lambda so it outlives this method
|
||||
_ = Task.Factory.StartNew(async () =>
|
||||
{
|
||||
IServiceScope? serviceScope = null;
|
||||
// ISiteEventLogger is a singleton; resolve from the root provider so
|
||||
// it is available to the catch blocks regardless of scope state.
|
||||
var siteEventLogger = serviceProvider?.GetService<ISiteEventLogger>();
|
||||
using var cts = new CancellationTokenSource(timeout);
|
||||
try
|
||||
{
|
||||
// Resolve integration services from DI (scoped lifetime)
|
||||
IExternalSystemClient? externalSystemClient = null;
|
||||
IDatabaseGateway? databaseGateway = null;
|
||||
// Notification Outbox: the S&F engine is a singleton; the site identity
|
||||
// provider supplies the site id stamped on enqueued notifications.
|
||||
StoreAndForwardService? storeAndForward = null;
|
||||
var siteId = string.Empty;
|
||||
// Audit Log #23 (M2 Bundle F): the writer is a singleton (FallbackAuditWriter
|
||||
// composes the SQLite hot-path + drop-oldest ring); null in tests / hosts
|
||||
// that haven't called AddAuditLog, which the helper handles as a no-op.
|
||||
IAuditWriter? auditWriter = null;
|
||||
// Audit Log #23 (M3 Bundle A — Task A3): site-local tracking store
|
||||
// backing Tracking.Status(id). Singleton; null in tests / hosts
|
||||
// that haven't wired the store, which the helper handles by
|
||||
// throwing on access.
|
||||
IOperationTrackingStore? operationTrackingStore = null;
|
||||
// Audit Log #23 (M3 Bundle F — Task F1): site-side cached-call
|
||||
// telemetry forwarder. Singleton bound to the AuditLog
|
||||
// composition root; null in tests / hosts that haven't called
|
||||
// AddAuditLog, in which case the cached-call helpers degrade
|
||||
// to the no-emission path (the underlying S&F handoff still
|
||||
// happens and a TrackedOperationId is still returned).
|
||||
ICachedCallTelemetryForwarder? cachedForwarder = null;
|
||||
// SourceNode-stamping (Tasks 13/14): the local node name
|
||||
// resolved from INodeIdentityProvider — node-a/node-b on site
|
||||
// hosts. Null in tests / hosts that haven't registered the
|
||||
// provider, in which case NotificationSubmit.SourceNode and
|
||||
// SiteCallOperational.SourceNode stay null and central
|
||||
// persists the rows with SourceNode NULL.
|
||||
string? sourceNode = null;
|
||||
|
||||
if (serviceProvider != null)
|
||||
{
|
||||
serviceScope = serviceProvider.CreateScope();
|
||||
externalSystemClient = serviceScope.ServiceProvider.GetService<IExternalSystemClient>();
|
||||
databaseGateway = serviceScope.ServiceProvider.GetService<IDatabaseGateway>();
|
||||
storeAndForward = serviceScope.ServiceProvider.GetService<StoreAndForwardService>();
|
||||
siteId = serviceScope.ServiceProvider.GetService<ISiteIdentityProvider>()?.SiteId
|
||||
?? string.Empty;
|
||||
auditWriter = serviceScope.ServiceProvider.GetService<IAuditWriter>();
|
||||
operationTrackingStore = serviceScope.ServiceProvider.GetService<IOperationTrackingStore>();
|
||||
cachedForwarder = serviceScope.ServiceProvider.GetService<ICachedCallTelemetryForwarder>();
|
||||
sourceNode = serviceScope.ServiceProvider.GetService<INodeIdentityProvider>()?.NodeName;
|
||||
}
|
||||
|
||||
var context = new ScriptRuntimeContext(
|
||||
instanceActor,
|
||||
self,
|
||||
sharedScriptLibrary,
|
||||
callDepth,
|
||||
options.MaxScriptCallDepth,
|
||||
timeout,
|
||||
instanceName,
|
||||
logger,
|
||||
externalSystemClient,
|
||||
databaseGateway,
|
||||
storeAndForward,
|
||||
siteCommunicationActor,
|
||||
siteId,
|
||||
// Notification Outbox (FU3): stamp the executing script onto outbound
|
||||
// notifications using the Site Event Logging "Source" convention.
|
||||
sourceScript: $"ScriptActor:{scriptName}",
|
||||
// Audit Log #23 (M2 Bundle F): emit one ApiOutbound/ApiCall row per
|
||||
// ExternalSystem.Call. Writer is best-effort; failures are logged
|
||||
// and swallowed inside the helper so the script's call path is
|
||||
// never aborted by an audit failure.
|
||||
auditWriter: auditWriter,
|
||||
// Audit Log #23 (M3 Bundle A — Task A3): site-local tracking store
|
||||
// backing Tracking.Status(id). Authoritative source of truth for
|
||||
// cached-call status — read directly by the script API.
|
||||
operationTrackingStore: operationTrackingStore,
|
||||
// Audit Log #23 (M3 Bundle F — Task F1): cached-call telemetry
|
||||
// forwarder for ExternalSystem.CachedCall / Database.CachedWrite
|
||||
// CachedSubmit emission + the immediate-success terminal-row
|
||||
// emission. Best-effort: null degrades the helpers to a
|
||||
// no-emission path; the S&F handoff and TrackedOperationId
|
||||
// return are unaffected.
|
||||
cachedForwarder: cachedForwarder,
|
||||
// Audit Log #23 (ParentExecutionId): the spawning execution's
|
||||
// id for an inbound-API-routed call. The routed script still
|
||||
// mints its own fresh ExecutionId — this records the spawner.
|
||||
// Null for normal (tag-change / timer) runs.
|
||||
parentExecutionId: parentExecutionId,
|
||||
// SourceNode-stamping (Tasks 13/14): the local node name
|
||||
// (node-a/node-b on a site) — threaded down so Notify.Send
|
||||
// and the four cached-call telemetry constructors can stamp
|
||||
// it onto NotificationSubmit.SourceNode and
|
||||
// SiteCallOperational.SourceNode respectively.
|
||||
sourceNode: sourceNode);
|
||||
|
||||
var globals = new ScriptGlobals
|
||||
{
|
||||
Instance = context,
|
||||
Parameters = new ScriptParameters(parameters ?? new Dictionary<string, object?>()),
|
||||
CancellationToken = cts.Token,
|
||||
Scope = scope
|
||||
};
|
||||
|
||||
var state = await compiledScript.RunAsync(globals, cts.Token);
|
||||
|
||||
// Send result to requester if this was an Ask-based call
|
||||
if (!replyTo.IsNobody())
|
||||
{
|
||||
replyTo.Tell(new ScriptCallResult(correlationId, true, state.ReturnValue, null));
|
||||
}
|
||||
|
||||
// Notify parent of completion
|
||||
parent.Tell(new ScriptActor.ScriptExecutionCompleted(scriptName, true, null));
|
||||
}
|
||||
catch (OperationCanceledException)
|
||||
{
|
||||
healthCollector?.IncrementScriptError();
|
||||
var errorMsg = $"Script '{scriptName}' on instance '{instanceName}' timed out after {timeout.TotalSeconds}s";
|
||||
logger.LogWarning(errorMsg);
|
||||
|
||||
// WP-32: Failures recorded to site event log; script NOT disabled after failure.
|
||||
_ = siteEventLogger?.LogEventAsync(
|
||||
"script", "Error", instanceName, $"ScriptActor:{scriptName}", errorMsg);
|
||||
|
||||
if (!replyTo.IsNobody())
|
||||
{
|
||||
replyTo.Tell(new ScriptCallResult(correlationId, false, null, errorMsg));
|
||||
}
|
||||
|
||||
parent.Tell(new ScriptActor.ScriptExecutionCompleted(scriptName, false, errorMsg));
|
||||
}
|
||||
catch (Exception ex)
|
||||
{
|
||||
healthCollector?.IncrementScriptError();
|
||||
// WP-32: Failures recorded to site event log; script NOT disabled after failure.
|
||||
var errorMsg = $"Script '{scriptName}' on instance '{instanceName}' failed: {ex.Message}";
|
||||
logger.LogError(ex, "Script execution failed: {Script} on {Instance}", scriptName, instanceName);
|
||||
|
||||
_ = siteEventLogger?.LogEventAsync(
|
||||
"script", "Error", instanceName, $"ScriptActor:{scriptName}", errorMsg, ex.ToString());
|
||||
|
||||
if (!replyTo.IsNobody())
|
||||
{
|
||||
replyTo.Tell(new ScriptCallResult(correlationId, false, null, errorMsg));
|
||||
}
|
||||
|
||||
parent.Tell(new ScriptActor.ScriptExecutionCompleted(scriptName, false, errorMsg));
|
||||
}
|
||||
finally
|
||||
{
|
||||
// Dispose the DI scope (and scoped services) after script execution completes
|
||||
serviceScope?.Dispose();
|
||||
// Stop self after execution completes
|
||||
self.Tell(PoisonPill.Instance);
|
||||
}
|
||||
}, CancellationToken.None, TaskCreationOptions.DenyChildAttach, scheduler).Unwrap();
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,224 @@
|
||||
using Akka.Actor;
|
||||
using Akka.Cluster;
|
||||
using Akka.Event;
|
||||
using Microsoft.Extensions.Logging;
|
||||
using ZB.MOM.WW.ScadaBridge.SiteRuntime.Messages;
|
||||
using ZB.MOM.WW.ScadaBridge.SiteRuntime.Persistence;
|
||||
using ZB.MOM.WW.ScadaBridge.StoreAndForward;
|
||||
|
||||
namespace ZB.MOM.WW.ScadaBridge.SiteRuntime.Actors;
|
||||
|
||||
/// <summary>
|
||||
/// Runs on every site node (not a singleton). Handles both config and S&F replication
|
||||
/// between site cluster peers.
|
||||
///
|
||||
/// Outbound: receives local replication requests and forwards to peer via ActorSelection.
|
||||
/// Inbound: receives replicated operations from peer and applies to local SQLite.
|
||||
/// Uses fire-and-forget (Tell) — no ack wait per design.
|
||||
/// </summary>
|
||||
public class SiteReplicationActor : ReceiveActor
|
||||
{
|
||||
private readonly SiteStorageService _storage;
|
||||
private readonly StoreAndForwardStorage _sfStorage;
|
||||
private readonly ReplicationService _replicationService;
|
||||
private readonly string _siteRole;
|
||||
private readonly ILogger<SiteReplicationActor> _logger;
|
||||
private readonly Cluster _cluster;
|
||||
private Address? _peerAddress;
|
||||
|
||||
/// <summary>
|
||||
/// Initializes a new <see cref="SiteReplicationActor"/> and registers Akka message handlers.
|
||||
/// </summary>
|
||||
/// <param name="storage">Service for accessing local site storage.</param>
|
||||
/// <param name="sfStorage">Store-and-forward SQLite storage for replication of buffered messages.</param>
|
||||
/// <param name="replicationService">Service providing replication transport logic.</param>
|
||||
/// <param name="siteRole">Akka cluster role used to identify peer nodes to replicate to.</param>
|
||||
/// <param name="logger">Logger instance.</param>
|
||||
public SiteReplicationActor(
|
||||
SiteStorageService storage,
|
||||
StoreAndForwardStorage sfStorage,
|
||||
ReplicationService replicationService,
|
||||
string siteRole,
|
||||
ILogger<SiteReplicationActor> logger)
|
||||
{
|
||||
_storage = storage;
|
||||
_sfStorage = sfStorage;
|
||||
_replicationService = replicationService;
|
||||
_siteRole = siteRole;
|
||||
_logger = logger;
|
||||
_cluster = Cluster.Get(Context.System);
|
||||
|
||||
// Cluster member events
|
||||
Receive<ClusterEvent.MemberUp>(HandleMemberUp);
|
||||
Receive<ClusterEvent.MemberRemoved>(HandleMemberRemoved);
|
||||
Receive<ClusterEvent.CurrentClusterState>(HandleCurrentClusterState);
|
||||
|
||||
// Outbound — forward to peer
|
||||
Receive<ReplicateConfigDeploy>(msg => SendToPeer(new ApplyConfigDeploy(
|
||||
msg.InstanceName, msg.ConfigJson, msg.DeploymentId, msg.RevisionHash, msg.IsEnabled)));
|
||||
Receive<ReplicateConfigRemove>(msg => SendToPeer(new ApplyConfigRemove(msg.InstanceName)));
|
||||
Receive<ReplicateConfigSetEnabled>(msg => SendToPeer(new ApplyConfigSetEnabled(
|
||||
msg.InstanceName, msg.IsEnabled)));
|
||||
Receive<ReplicateArtifacts>(msg => SendToPeer(new ApplyArtifacts(msg.Command)));
|
||||
Receive<ReplicateStoreAndForward>(msg => SendToPeer(new ApplyStoreAndForward(msg.Operation)));
|
||||
|
||||
// Inbound — apply from peer
|
||||
Receive<ApplyConfigDeploy>(HandleApplyConfigDeploy);
|
||||
Receive<ApplyConfigRemove>(HandleApplyConfigRemove);
|
||||
Receive<ApplyConfigSetEnabled>(HandleApplyConfigSetEnabled);
|
||||
Receive<ApplyArtifacts>(HandleApplyArtifacts);
|
||||
Receive<ApplyStoreAndForward>(HandleApplyStoreAndForward);
|
||||
}
|
||||
|
||||
/// <inheritdoc />
|
||||
protected override void PreStart()
|
||||
{
|
||||
base.PreStart();
|
||||
_cluster.Subscribe(Self, ClusterEvent.SubscriptionInitialStateMode.InitialStateAsSnapshot,
|
||||
typeof(ClusterEvent.MemberUp),
|
||||
typeof(ClusterEvent.MemberRemoved));
|
||||
_logger.LogInformation("SiteReplicationActor started, subscribing to cluster events for role {Role}", _siteRole);
|
||||
}
|
||||
|
||||
/// <inheritdoc />
|
||||
protected override void PostStop()
|
||||
{
|
||||
_cluster.Unsubscribe(Self);
|
||||
base.PostStop();
|
||||
}
|
||||
|
||||
private void HandleCurrentClusterState(ClusterEvent.CurrentClusterState state)
|
||||
{
|
||||
foreach (var member in state.Members)
|
||||
{
|
||||
if (member.Status == MemberStatus.Up)
|
||||
TryTrackPeer(member);
|
||||
}
|
||||
}
|
||||
|
||||
private void HandleMemberUp(ClusterEvent.MemberUp evt)
|
||||
{
|
||||
TryTrackPeer(evt.Member);
|
||||
}
|
||||
|
||||
private void HandleMemberRemoved(ClusterEvent.MemberRemoved evt)
|
||||
{
|
||||
if (evt.Member.Address.Equals(_peerAddress))
|
||||
{
|
||||
_logger.LogInformation("Peer node removed: {Address}", _peerAddress);
|
||||
_peerAddress = null;
|
||||
}
|
||||
}
|
||||
|
||||
private void TryTrackPeer(Member member)
|
||||
{
|
||||
// Must have our site role, and must not be self
|
||||
if (member.HasRole(_siteRole) && !member.Address.Equals(_cluster.SelfAddress))
|
||||
{
|
||||
_peerAddress = member.Address;
|
||||
_logger.LogInformation("Peer node tracked: {Address}", _peerAddress);
|
||||
}
|
||||
}
|
||||
|
||||
private void SendToPeer(object message)
|
||||
{
|
||||
if (_peerAddress == null)
|
||||
{
|
||||
_logger.LogDebug("No peer available, dropping replication message {Type}", message.GetType().Name);
|
||||
return;
|
||||
}
|
||||
|
||||
var path = new RootActorPath(_peerAddress) / "user" / "site-replication";
|
||||
Context.ActorSelection(path).Tell(message);
|
||||
}
|
||||
|
||||
// ── Inbound handlers ──
|
||||
|
||||
private void HandleApplyConfigDeploy(ApplyConfigDeploy msg)
|
||||
{
|
||||
_logger.LogInformation("Applying replicated config deploy for {Instance}", msg.InstanceName);
|
||||
_storage.StoreDeployedConfigAsync(
|
||||
msg.InstanceName, msg.ConfigJson, msg.DeploymentId, msg.RevisionHash, msg.IsEnabled)
|
||||
.ContinueWith(t =>
|
||||
{
|
||||
if (t.IsFaulted)
|
||||
_logger.LogError(t.Exception, "Failed to apply replicated deploy for {Instance}", msg.InstanceName);
|
||||
});
|
||||
}
|
||||
|
||||
private void HandleApplyConfigRemove(ApplyConfigRemove msg)
|
||||
{
|
||||
_logger.LogInformation("Applying replicated config remove for {Instance}", msg.InstanceName);
|
||||
_storage.RemoveDeployedConfigAsync(msg.InstanceName)
|
||||
.ContinueWith(t =>
|
||||
{
|
||||
if (t.IsFaulted)
|
||||
_logger.LogError(t.Exception, "Failed to apply replicated remove for {Instance}", msg.InstanceName);
|
||||
});
|
||||
}
|
||||
|
||||
private void HandleApplyConfigSetEnabled(ApplyConfigSetEnabled msg)
|
||||
{
|
||||
_logger.LogInformation("Applying replicated set-enabled={Enabled} for {Instance}", msg.IsEnabled, msg.InstanceName);
|
||||
_storage.SetInstanceEnabledAsync(msg.InstanceName, msg.IsEnabled)
|
||||
.ContinueWith(t =>
|
||||
{
|
||||
if (t.IsFaulted)
|
||||
_logger.LogError(t.Exception, "Failed to apply replicated set-enabled for {Instance}", msg.InstanceName);
|
||||
});
|
||||
}
|
||||
|
||||
private void HandleApplyArtifacts(ApplyArtifacts msg)
|
||||
{
|
||||
var command = msg.Command;
|
||||
_logger.LogInformation("Applying replicated artifacts, deploymentId={DeploymentId}", command.DeploymentId);
|
||||
|
||||
Task.Run(async () =>
|
||||
{
|
||||
try
|
||||
{
|
||||
if (command.SharedScripts != null)
|
||||
foreach (var s in command.SharedScripts)
|
||||
await _storage.StoreSharedScriptAsync(s.Name, s.Code, s.ParameterDefinitions, s.ReturnDefinition);
|
||||
|
||||
if (command.ExternalSystems != null)
|
||||
foreach (var es in command.ExternalSystems)
|
||||
await _storage.StoreExternalSystemAsync(es.Name, es.EndpointUrl, es.AuthType, es.AuthConfiguration, es.MethodDefinitionsJson);
|
||||
|
||||
if (command.DatabaseConnections != null)
|
||||
foreach (var db in command.DatabaseConnections)
|
||||
await _storage.StoreDatabaseConnectionAsync(db.Name, db.ConnectionString, db.MaxRetries, db.RetryDelay);
|
||||
|
||||
if (command.NotificationLists != null)
|
||||
foreach (var nl in command.NotificationLists)
|
||||
await _storage.StoreNotificationListAsync(nl.Name, nl.RecipientEmails);
|
||||
|
||||
if (command.DataConnections != null)
|
||||
foreach (var dc in command.DataConnections)
|
||||
await _storage.StoreDataConnectionDefinitionAsync(dc.Name, dc.Protocol, dc.PrimaryConfigurationJson, dc.BackupConfigurationJson, dc.FailoverRetryCount);
|
||||
|
||||
if (command.SmtpConfigurations != null)
|
||||
foreach (var smtp in command.SmtpConfigurations)
|
||||
await _storage.StoreSmtpConfigurationAsync(smtp.Name, smtp.Server, smtp.Port, smtp.AuthMode,
|
||||
smtp.FromAddress, smtp.Username, smtp.Password, smtp.OAuthConfig);
|
||||
}
|
||||
catch (Exception ex)
|
||||
{
|
||||
_logger.LogError(ex, "Failed to apply replicated artifacts");
|
||||
}
|
||||
});
|
||||
}
|
||||
|
||||
private void HandleApplyStoreAndForward(ApplyStoreAndForward msg)
|
||||
{
|
||||
_logger.LogDebug("Applying replicated S&F operation {OpType} for message {Id}",
|
||||
msg.Operation.OperationType, msg.Operation.MessageId);
|
||||
|
||||
_replicationService.ApplyReplicatedOperationAsync(msg.Operation, _sfStorage)
|
||||
.ContinueWith(t =>
|
||||
{
|
||||
if (t.IsFaulted)
|
||||
_logger.LogError(t.Exception, "Failed to apply replicated S&F operation {Id}", msg.Operation.MessageId);
|
||||
});
|
||||
}
|
||||
}
|
||||
Reference in New Issue
Block a user