perf(site-runtime): deserialize a deploy's flattened config once and hand the parsed object down (review 2026-10-21 01/SR-P5)

The deploy compile warm parses once off-thread; the parsed FlattenedConfiguration rides DeployCompileWarmed to the gate (new Validate(FlattenedConfiguration) overload), the redeploy buffer, CreateInstanceActor's DCL ensure and a new InstanceActor constructor overload. Stored-row paths parse once inside CreateInstanceActor; an unparseable stored row still fails init via S6.
2026-10-22 06:47:18 -04:00
parent 960c35060c
commit dde22a3dd1
5 changed files with 296 additions and 40 deletions
@@ -11,6 +11,7 @@ using ZB.MOM.WW.ScadaBridge.Commons.Messages.Lifecycle;
 using ZB.MOM.WW.ScadaBridge.Commons.Messages.Management;
 using ZB.MOM.WW.ScadaBridge.Commons.Messages.ScriptExecution;
 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.SiteEventLogging;
 using ZB.MOM.WW.ScadaBridge.SiteRuntime.Deployment;
@@ -985,9 +986,16 @@ public class DeploymentManagerActor : ReceiveActor, IWithTimers
        var configJson = command.FlattenedConfigurationJson;
        Task.Run(() =>
        {
+            // Review 2026-10-21 01/SR-P5: this is the ONE deserialization of the deploy's config.
+            // The parsed object rides back on DeployCompileWarmed to the gate, the DCL-connection
+            // ensure and the Instance Actor, which each used to parse the same JSON again — two
+            // of those parses on this singleton's own thread.
+            if (!DeployCompileValidator.TryParse(configJson, out var parsed, out var parseError))
+                return new DeployCompileWarmed(command, null, parseError);
+
            try
            {
-                validator.Validate(configJson);
+                validator.Validate(parsed!);
            }
            catch (Exception ex)
            {
@@ -996,7 +1004,7 @@ public class DeploymentManagerActor : ReceiveActor, IWithTimers
                    command.InstanceUniqueName);
            }

-            return new DeployCompileWarmed(command);
+            return new DeployCompileWarmed(command, parsed, null);
        }).PipeTo(Self);
    }

@@ -1019,7 +1027,7 @@ public class DeploymentManagerActor : ReceiveActor, IWithTimers

        _deployWarms.Remove(instanceName);

-        RunDeployGateAndProceed(warming.Command, warming.ReplyTo);
+        RunDeployGateAndProceed(warming.Command, warming.ReplyTo, msg.Parsed, msg.ParseError);

        foreach (var queued in warming.Buffered)
            DispatchBufferedCommand(queued);
@@ -1072,7 +1080,13 @@ public class DeploymentManagerActor : ReceiveActor, IWithTimers
    /// only from <see cref="HandleDeployCompileWarmed"/>, i.e. after the same compile has
    /// been warmed off-thread.
    /// </summary>
-    private void RunDeployGateAndProceed(DeployInstanceCommand command, IActorRef replyTo)
+    /// <param name="command">The deploy command being gated.</param>
+    /// <param name="replyTo">The deployer to answer.</param>
+    /// <param name="parsed">The config parsed once by the warm (01/SR-P5), or null on a parse failure.</param>
+    /// <param name="parseError">The warm's deserialization error, when <paramref name="parsed"/> is null.</param>
+    private void RunDeployGateAndProceed(
+        DeployInstanceCommand command, IActorRef replyTo,
+        FlattenedConfiguration? parsed, string? parseError)
    {
        // Site-side compile gate (S3): a compile failure must reject the
        // deployment with NO partial state applied (no Instance Actor, no
@@ -1092,8 +1106,9 @@ public class DeploymentManagerActor : ReceiveActor, IWithTimers
        // The gate also rejects any instance/script/alarm/native-source name that is not a legal
        // Akka actor name (review 2026-10-21 01/SR-C1) — such a name used to throw out of
        // Context.ActorOf on THIS actor's thread and restart the singleton with every instance.
-        var errors = _deployCompileValidator.Validate(
-            command.FlattenedConfigurationJson, command.InstanceUniqueName);
+        var errors = parsed == null
+            ? [parseError ?? "Configuration JSON failed to deserialize."]
+            : _deployCompileValidator.Validate(parsed, command.InstanceUniqueName);
        if (errors.Count > 0)
        {
            var instanceName = command.InstanceUniqueName;
@@ -1109,14 +1124,15 @@ public class DeploymentManagerActor : ReceiveActor, IWithTimers
            return;
        }

-        ProceedWithDeploy(command, replyTo);
+        ProceedWithDeploy(command, replyTo, parsed!);
    }

    /// <summary>
    /// Core deploy application (redeploy-supersede + fresh apply), reached only
    /// after the synchronous compile gate in <see cref="HandleDeploy"/> passes.
    /// </summary>
-    private void ProceedWithDeploy(DeployInstanceCommand command, IActorRef replyTo)
+    private void ProceedWithDeploy(
+        DeployInstanceCommand command, IActorRef replyTo, FlattenedConfiguration parsed)
    {
        var instanceName = command.InstanceUniqueName;
        _logger.LogInformation(
@@ -1129,7 +1145,7 @@ public class DeploymentManagerActor : ReceiveActor, IWithTimers
        if (_instanceActors.TryGetValue(instanceName, out var existing))
        {
            _instanceActors.Remove(instanceName);
-            _pendingRedeploys[existing] = new PendingRedeploy(command, replyTo);
+            _pendingRedeploys[existing] = new PendingRedeploy(command, replyTo, parsed);
            _terminatingActorsByName[instanceName] = existing;
            Context.Watch(existing);
            _redeployStopQueue.Enqueue(existing);
@@ -1157,12 +1173,12 @@ public class DeploymentManagerActor : ReceiveActor, IWithTimers
                    $"superseded by newer deployment {command.DeploymentId} before predecessor finished terminating",
                    DateTimeOffset.UtcNow));
            }
-            _pendingRedeploys[terminatingRef] = new PendingRedeploy(command, replyTo);
+            _pendingRedeploys[terminatingRef] = new PendingRedeploy(command, replyTo, parsed);
            return;
        }

        // Fresh deployment — no existing actor to replace.
-        ApplyDeployment(command, replyTo, isRedeploy: false);
+        ApplyDeployment(command, replyTo, isRedeploy: false, parsed);
    }

    private bool DeferLifecycleStop(string name, string commandId, bool delete, IActorRef replyTo)
@@ -1224,7 +1240,7 @@ public class DeploymentManagerActor : ReceiveActor, IWithTimers
        {
            _terminatingActorsByName.Remove(pending.Command.InstanceUniqueName);
            if (flush.Error is null)
-                ApplyDeployment(pending.Command, pending.OriginalSender, isRedeploy: true);
+                ApplyDeployment(pending.Command, pending.OriginalSender, isRedeploy: true, pending.Parsed);
            else
                pending.OriginalSender.Tell(new DeploymentStatusResponse(pending.Command.DeploymentId,
                    pending.Command.InstanceUniqueName, DeploymentStatus.Failed,
@@ -1381,7 +1397,8 @@ public class DeploymentManagerActor : ReceiveActor, IWithTimers
    /// incorrect. The reply is sent from <see cref="HandleDeployPersistenceResult"/> once
    /// the persistence outcome is known.
    /// </summary>
-    private void ApplyDeployment(DeployInstanceCommand command, IActorRef sender, bool isRedeploy)
+    private void ApplyDeployment(
+        DeployInstanceCommand command, IActorRef sender, bool isRedeploy, FlattenedConfiguration? parsed)
    {
        var instanceName = command.InstanceUniqueName;

@@ -1413,7 +1430,7 @@ public class DeploymentManagerActor : ReceiveActor, IWithTimers
        // connections this config's data-sourced attributes subscribe through (register row 85).
        if (!CreateInstanceActor(
                instanceName, command.FlattenedConfigurationJson, command.RevisionHash,
-                out var createError, staticStateResetGate))
+                out var createError, staticStateResetGate, parsed))
        {
            // Unreachable while the name pre-check above holds; kept so an ActorOf refusal can
            // never escape to the supervisor and restart the singleton.
@@ -2888,10 +2905,32 @@ public class DeploymentManagerActor : ReceiveActor, IWithTimers
    {
        if (_dclManager == null) return;

+        FlattenedConfiguration? config;
        try
        {
-            var config = System.Text.Json.JsonSerializer.Deserialize<Commons.Types.Flattening.FlattenedConfiguration>(configJson);
-            if (config?.Connections == null) return;
+            config = System.Text.Json.JsonSerializer.Deserialize<FlattenedConfiguration>(configJson);
+        }
+        catch (Exception ex)
+        {
+            _logger.LogWarning(ex, "Failed to parse flattened config for DCL connections");
+            return;
+        }
+
+        if (config != null)
+            EnsureDclConnections(config);
+    }
+
+    /// <summary>
+    /// <see cref="EnsureDclConnections(string)"/> over an already-parsed configuration
+    /// (review 2026-10-21 01/SR-P5) — the form every creation path now uses.
+    /// </summary>
+    private void EnsureDclConnections(FlattenedConfiguration config)
+    {
+        if (_dclManager == null) return;
+
+        try
+        {
+            if (config.Connections == null) return;

            foreach (var (name, connConfig) in config.Connections)
            {
@@ -2928,7 +2967,7 @@ public class DeploymentManagerActor : ReceiveActor, IWithTimers
        }
        catch (Exception ex)
        {
-            _logger.LogWarning(ex, "Failed to parse flattened config for DCL connections");
+            _logger.LogWarning(ex, "Failed to apply flattened config DCL connections");
        }
    }

@@ -3745,15 +3784,27 @@ public class DeploymentManagerActor : ReceiveActor, IWithTimers
    /// <see langword="null"/> on every non-deploy start (boot batch, enable, self-heal), where
    /// nothing is being reset and the load runs immediately as before.
    /// </param>
+    /// <param name="parsed">
+    /// <paramref name="configJson"/> already deserialized by the caller (the deploy warm or the
+    /// boot-batch warm — review 2026-10-21 01/SR-P5), or <see langword="null"/> to parse it here,
+    /// ONCE, for both the DCL-connection ensure and the Instance Actor.
+    /// </param>
    /// <returns>
    /// <see langword="false"/> when Akka refused the actor name (review 2026-10-21 01/SR-C1);
    /// nothing was created or recorded. Callers turn that into a Failed reply / site event.
    /// </returns>
    internal bool CreateInstanceActor(
        string instanceName, string configJson, string revisionHash, out string? error,
-        Task? staticStateResetGate = null)
+        Task? staticStateResetGate = null, FlattenedConfiguration? parsed = null)
    {
        error = null;
+
+        // 01/SR-P5: one parse for the DCL ensure AND the actor. A config that does not parse is
+        // NOT handled here: the JSON is handed to the Instance Actor unchanged, whose constructor
+        // throws exactly as before (ActorInitializationException → Stop → the S6 path), so the
+        // failure semantics of a poison stored config are unchanged.
+        if (parsed == null && !DeployCompileValidator.TryParse(configJson, out parsed, out _))
+            parsed = null;
        // Deferred register row 85 — an Instance Actor is useless without the DCL connection
        // actors its data-sourced attributes subscribe through, so ensuring them belongs in the
        // single creation funnel rather than at each caller.
@@ -3777,7 +3828,10 @@ public class DeploymentManagerActor : ReceiveActor, IWithTimers
        // their exact previous semantics (they called it unconditionally). It is cheap and
        // silent when there is nothing to do: EnsureDclConnection is hash-guarded on
        // _createdConnections, so an unchanged connection is a no-op with no message and no log.
-        EnsureDclConnections(configJson);
+        if (parsed != null)
+            EnsureDclConnections(parsed);
+        else
+            EnsureDclConnections(configJson);

        if (_instanceActors.ContainsKey(instanceName))
        {
@@ -3787,20 +3841,36 @@ public class DeploymentManagerActor : ReceiveActor, IWithTimers

        // Reuse the shared, host-configured logger factory
        // instead of allocating (and leaking) a fresh LoggerFactory per instance.
-        var props = Props.Create(() => new InstanceActor(
-            instanceName,
-            configJson,
-            _storage,
-            _compilationService,
-            _sharedScriptLibrary,
-            _streamManager,
-            _options,
-            _loggerFactory.CreateLogger<InstanceActor>(),
-            _dclManager,
-            _healthCollector,
-            _serviceProvider,
-            null,
-            staticStateResetGate));
+        var instanceLogger = _loggerFactory.CreateLogger<InstanceActor>();
+        var props = parsed != null
+            ? Props.Create(() => new InstanceActor(
+                instanceName,
+                parsed,
+                _storage,
+                _compilationService,
+                _sharedScriptLibrary,
+                _streamManager,
+                _options,
+                instanceLogger,
+                _dclManager,
+                _healthCollector,
+                _serviceProvider,
+                null,
+                staticStateResetGate))
+            : Props.Create(() => new InstanceActor(
+                instanceName,
+                configJson,
+                _storage,
+                _compilationService,
+                _sharedScriptLibrary,
+                _streamManager,
+                _options,
+                instanceLogger,
+                _dclManager,
+                _healthCollector,
+                _serviceProvider,
+                null,
+                staticStateResetGate));

        // Review 2026-10-21 01/SR-C1: ActorOf throws InvalidActorNameException SYNCHRONOUSLY on
        // this actor's thread for a name Akka refuses (space, '#', '/', non-ASCII, ...). Escaping
@@ -3978,7 +4048,11 @@ public class DeploymentManagerActor : ReceiveActor, IWithTimers
    /// command instance it warmed so a result superseded mid-warm is recognised by reference
    /// and dropped.
    /// </summary>
-    internal sealed record DeployCompileWarmed(DeployInstanceCommand Command);
+    /// <param name="Command">The exact command instance the warm ran for.</param>
+    /// <param name="Parsed">The config parsed once off-thread (01/SR-P5); null when it did not parse.</param>
+    /// <param name="ParseError">Why it did not parse, when <paramref name="Parsed"/> is null.</param>
+    internal sealed record DeployCompileWarmed(
+        DeployInstanceCommand Command, FlattenedConfiguration? Parsed = null, string? ParseError = null);

    /// <summary>
    /// Piped back to self once a staggered startup batch's compiles have been
@@ -4100,7 +4174,11 @@ public class DeploymentManagerActor : ReceiveActor, IWithTimers
    /// <summary>
    /// A redeployment command buffered until the previous Instance Actor terminates.
    /// </summary>
-    internal record PendingRedeploy(DeployInstanceCommand Command, IActorRef OriginalSender);
+    /// <param name="Command">The buffered deploy.</param>
+    /// <param name="OriginalSender">The deployer to answer.</param>
+    /// <param name="Parsed">The config the gate already parsed (01/SR-P5), reused on apply.</param>
+    internal record PendingRedeploy(
+        DeployInstanceCommand Command, IActorRef OriginalSender, FlattenedConfiguration? Parsed = null);

    /// <summary>
    /// Join state for a deployment awaiting confirmation. Deployment <c>Success</c> is