feat(opcua): surgical pure-remove deploys — scoped teardown, no full rebuild (R2-07 T11)

This commit is contained in:
Joseph Doherty
2026-07-13 12:18:08 -04:00
parent 4504ed930c
commit 502d7650d0
3 changed files with 257 additions and 18 deletions
@@ -128,8 +128,12 @@ public sealed class AddressSpaceApplier
// every client subscription server-wide survives; coincident surgical tag deltas + folder
// renames are applied IN PLACE via ISurgicalAddressSpaceSink below (any false/throw there falls
// back to a full rebuild — the F10b contract).
// • PureRemove / AddRemoveMix ⇒ full rebuild UNTIL their phases ship (R2-07 T11 / T13); mapping
// them here keeps each phase independently shippable with today's behaviour as the floor.
// • PureRemove ⇒ NO full rebuild. The removed nodes are torn down IN PLACE, scoped to the affected
// subtree, via the ISurgicalAddressSpaceSink remove members below (each preceded by a terminal
// Bad / RemovedConditionState write so in-flight MonitoredItems observe the removal); subscribers
// of OTHER nodes are untouched. Any false/throw from a remove ⇒ full-rebuild fallback (ratchet).
// • AddRemoveMix ⇒ full rebuild UNTIL Phase 3 ships (R2-07 T13); mapping it here keeps each phase
// independently shippable with today's behaviour as the floor.
// • Rebuild ⇒ the default-closed safety valve for any node-affecting change (ChangedEquipment /
// ChangedAlarms / non-surgical ChangedEquipmentTags / node-relevant ChangedEquipmentVirtualTags)
// — the classifier's rule 2, which also catches any future plan field that makes a changed
@@ -138,7 +142,6 @@ public sealed class AddressSpaceApplier
// a driver-only plan AttributeOnly, so it never rebuilds here.
var kind = AddressSpaceChangeClassifier.Classify(plan);
var mustRebuild = kind is AddressSpaceChangeKind.Rebuild
or AddressSpaceChangeKind.PureRemove
or AddressSpaceChangeKind.AddRemoveMix;
var surgicalTagDeltas = plan.ChangedEquipmentTags.Where(AddressSpaceChangeClassifier.TagDeltaIsSurgicalEligible).ToList();
@@ -155,7 +158,7 @@ public sealed class AddressSpaceApplier
rebuildFailed = !SafeRebuild();
rebuilt = true;
}
else if (surgicalTagDeltas.Count > 0 || renamedFolders.Count > 0)
else if (surgicalTagDeltas.Count > 0 || renamedFolders.Count > 0 || kind == AddressSpaceChangeKind.PureRemove)
{
if (_sink is ISurgicalAddressSpaceSink surgical)
{
@@ -191,6 +194,13 @@ public sealed class AddressSpaceApplier
}
if (!ok) { allApplied = false; break; }
}
// R2-07 Phase 2 — PureRemove scoped teardown (runs after any coincident surgical
// attribute updates / renames succeeded). Terminal Bad / RemovedConditionState writes then
// in-place removes; any false/throw flips allApplied → the rebuild ratchet below.
if (allApplied && kind == AddressSpaceChangeKind.PureRemove)
{
allApplied = ApplyPureRemove(surgical, plan, ts, ref failedNodes);
}
if (!allApplied) { rebuildFailed = !SafeRebuild(); rebuilt = true; }
}
else
@@ -218,6 +228,108 @@ public sealed class AddressSpaceApplier
return new AddressSpaceApplyOutcome(removedCount, addedCount, changedCount, rebuilt, rebuildFailed, failedNodes);
}
/// <summary>
/// R2-07 Phase 2 — apply a PureRemove plan by tearing down ONLY the affected nodes IN PLACE (no full
/// rebuild). Removed equipment "own" their child tags/vtags/alarms, so any removed child whose
/// <c>EquipmentId</c> is itself in <see cref="AddressSpacePlan.RemovedEquipment"/> is SUBSUMED by the
/// subtree removal and skipped individually. For surviving-equipment removals: a removed value
/// variable gets a terminal Bad <c>WriteValue</c> then <see cref="ISurgicalAddressSpaceSink.RemoveVariableNode"/>;
/// a removed alarm-bearing tag (the pre-R2-07 today-gap: it got no terminal condition write) and a
/// removed scripted alarm get a terminal <see cref="RemovedConditionState"/> write then
/// <see cref="ISurgicalAddressSpaceSink.RemoveAlarmConditionNode"/>; each removed equipment gets one
/// <see cref="ISurgicalAddressSpaceSink.RemoveEquipmentSubtree"/>. The terminal write is what an
/// in-flight MonitoredItem observes (a final Bad); after the node is gone, re-subscription gets
/// <c>BadNodeIdUnknown</c>. Returns false on the FIRST remove that reports the node unknown / throws
/// (the caller's one-way rebuild ratchet takes over), leaving no further surgical work attempted.
/// </summary>
/// <param name="surgical">The surgical sink to route removes through.</param>
/// <param name="plan">The PureRemove plan.</param>
/// <param name="ts">The timestamp for the terminal writes.</param>
/// <param name="failedNodes">Accumulator for swallowed terminal-write failures (archreview 01/S-1).</param>
/// <returns>True when every removal succeeded; false on the first failure (caller rebuilds).</returns>
private bool ApplyPureRemove(ISurgicalAddressSpaceSink surgical, AddressSpacePlan plan, DateTime ts, ref int failedNodes)
{
var removedEquipmentIds = new HashSet<string>(
plan.RemovedEquipment.Select(e => e.EquipmentId), StringComparer.Ordinal);
bool NotSubsumed(string equipmentId) => !removedEquipmentIds.Contains(equipmentId);
// Removed equipment tags — value variables OR alarm-bearing conditions — for SURVIVING equipment.
foreach (var tag in plan.RemovedEquipmentTags.Where(t => NotSubsumed(t.EquipmentId)))
{
var nodeId = EquipmentNodeIds.Variable(tag.EquipmentId, tag.FolderPath, tag.Name);
if (tag.Alarm is not null)
{
// Alarm-bearing tag → a Part 9 condition node. Terminal RemovedConditionState (today-gap
// closed), then remove the condition in place.
if (!SafeWriteAlarmCondition(nodeId, RemovedConditionState, ts)) failedNodes++;
if (!SafeRemoveAlarmCondition(surgical, nodeId)) return false;
}
else
{
// Plain value variable → terminal Bad so an in-flight MonitoredItem sees the removal.
SafeWriteValue(nodeId, null, OpcUaQuality.Bad, ts);
if (!SafeRemoveVariable(surgical, nodeId)) return false;
}
}
// Removed VirtualTags (always plain value variables) for surviving equipment.
foreach (var v in plan.RemovedEquipmentVirtualTags.Where(t => NotSubsumed(t.EquipmentId)))
{
var nodeId = EquipmentNodeIds.Variable(v.EquipmentId, v.FolderPath, v.Name);
SafeWriteValue(nodeId, null, OpcUaQuality.Bad, ts);
if (!SafeRemoveVariable(surgical, nodeId)) return false;
}
// Removed scripted alarms for surviving equipment — the terminal RemovedConditionState was already
// written by the top-of-Apply removal block; here we tear the condition node down in place.
foreach (var alarm in plan.RemovedAlarms.Where(a => NotSubsumed(a.EquipmentId)))
{
if (!SafeRemoveAlarmCondition(surgical, alarm.ScriptedAlarmId)) return false;
}
// Removed equipment — one subtree teardown each (subsumes their child tags/vtags/alarms). The
// top-of-Apply block already wrote the equipment id's terminal condition state.
foreach (var eq in plan.RemovedEquipment)
{
if (!SafeRemoveEquipmentSubtree(surgical, eq.EquipmentId)) return false;
}
return true;
}
/// <summary>Remove a value-variable node in place, treating a throw as a false (⇒ rebuild fallback).</summary>
/// <returns><c>true</c> when the node was removed; <c>false</c> when unknown or the sink threw.</returns>
private bool SafeRemoveVariable(ISurgicalAddressSpaceSink surgical, string nodeId)
{
try { return surgical.RemoveVariableNode(nodeId); }
catch (Exception ex) { _logger.LogError(ex, "AddressSpaceApplier: surgical RemoveVariableNode threw for {Node}", nodeId); return false; }
}
/// <summary>Remove an alarm-condition node in place, treating a throw as a false (⇒ rebuild fallback).</summary>
/// <returns><c>true</c> when the node was removed; <c>false</c> when unknown or the sink threw.</returns>
private bool SafeRemoveAlarmCondition(ISurgicalAddressSpaceSink surgical, string nodeId)
{
try { return surgical.RemoveAlarmConditionNode(nodeId); }
catch (Exception ex) { _logger.LogError(ex, "AddressSpaceApplier: surgical RemoveAlarmConditionNode threw for {Node}", nodeId); return false; }
}
/// <summary>Remove an equipment subtree in place, treating a throw as a false (⇒ rebuild fallback).</summary>
/// <returns><c>true</c> when the subtree was removed; <c>false</c> when unknown or the sink threw.</returns>
private bool SafeRemoveEquipmentSubtree(ISurgicalAddressSpaceSink surgical, string nodeId)
{
try { return surgical.RemoveEquipmentSubtree(nodeId); }
catch (Exception ex) { _logger.LogError(ex, "AddressSpaceApplier: surgical RemoveEquipmentSubtree threw for {Node}", nodeId); return false; }
}
/// <summary>Write a value, swallowing (and Warning-logging) any sink fault — used for the terminal Bad
/// on a removed variable so an in-flight MonitoredItem observes the removal.</summary>
/// <returns><c>true</c> when the write landed; <c>false</c> when the sink threw.</returns>
private bool SafeWriteValue(string nodeId, object? value, OpcUaQuality quality, DateTime ts)
{
try { _sink.WriteValue(nodeId, value, quality, ts); return true; }
catch (Exception ex) { _logger.LogWarning(ex, "AddressSpaceApplier: WriteValue threw for {Node}", nodeId); return false; }
}
/// <summary>
/// Compute the deduplicated, deterministically-ordered set of affected PARENT node ids to announce
/// (Part 3 <c>NodeAdded</c>) after a pure-add apply's Materialise passes have created the new nodes.