Merge pull request (#136) - Admin RedundancyTab

src/ZB.MOM.WW.OtOpcUa.Admin/Components/Pages/Clusters/ClusterDetail.razor

@@ -52,6 +52,7 @@ else
         <li class="nav-item"><button class="nav-link @Tab("namespaces")" @onclick='() => _tab = "namespaces"'>Namespaces</button></li>
         <li class="nav-item"><button class="nav-link @Tab("drivers")" @onclick='() => _tab = "drivers"'>Drivers</button></li>
         <li class="nav-item"><button class="nav-link @Tab("acls")" @onclick='() => _tab = "acls"'>ACLs</button></li>
+        <li class="nav-item"><button class="nav-link @Tab("redundancy")" @onclick='() => _tab = "redundancy"'>Redundancy</button></li>
         <li class="nav-item"><button class="nav-link @Tab("audit")" @onclick='() => _tab = "audit"'>Audit</button></li>
     </ul>
 
@@ -92,6 +93,10 @@ else
     {
         <AclsTab GenerationId="@_currentDraft.GenerationId" ClusterId="@ClusterId"/>
     }
+    else if (_tab == "redundancy")
+    {
+        <RedundancyTab ClusterId="@ClusterId"/>
+    }
     else if (_tab == "audit")
     {
         <AuditTab ClusterId="@ClusterId"/>

src/ZB.MOM.WW.OtOpcUa.Admin/Components/Pages/Clusters/RedundancyTab.razor

@@ -0,0 +1,136 @@
+@using ZB.MOM.WW.OtOpcUa.Admin.Services
+@using ZB.MOM.WW.OtOpcUa.Configuration.Entities
+@using ZB.MOM.WW.OtOpcUa.Configuration.Enums
+@inject ClusterNodeService NodeSvc
+
+<h4>Redundancy topology</h4>
+<p class="text-muted small">
+    One row per <code>ClusterNode</code> in this cluster. Role, <code>ApplicationUri</code>,
+    and <code>ServiceLevelBase</code> are authored separately; the Admin UI shows them read-only
+    here so operators can confirm the published topology without touching it. LastSeen older than
+    @((int)ClusterNodeService.StaleThreshold.TotalSeconds)s is flagged Stale — the node has
+    stopped heart-beating and is likely down. Role swap goes through the server-side
+    <code>RedundancyCoordinator</code> apply-lease flow, not direct DB edits.
+</p>
+
+@if (_nodes is null)
+{
+    <p>Loading…</p>
+}
+else if (_nodes.Count == 0)
+{
+    <div class="alert alert-warning">
+        No ClusterNode rows for this cluster. The server process needs at least one entry
+        (with a non-blank <code>ApplicationUri</code>) before it can start up per OPC UA spec.
+    </div>
+}
+else
+{
+    var primaries = _nodes.Count(n => n.RedundancyRole == RedundancyRole.Primary);
+    var secondaries = _nodes.Count(n => n.RedundancyRole == RedundancyRole.Secondary);
+    var standalone = _nodes.Count(n => n.RedundancyRole == RedundancyRole.Standalone);
+    var staleCount = _nodes.Count(ClusterNodeService.IsStale);
+
+    <div class="row g-3 mb-4">
+        <div class="col-md-3"><div class="card"><div class="card-body">
+            <h6 class="text-muted mb-1">Nodes</h6>
+            <div class="fs-3">@_nodes.Count</div>
+        </div></div></div>
+        <div class="col-md-3"><div class="card border-success"><div class="card-body">
+            <h6 class="text-muted mb-1">Primary</h6>
+            <div class="fs-3 text-success">@primaries</div>
+        </div></div></div>
+        <div class="col-md-3"><div class="card border-info"><div class="card-body">
+            <h6 class="text-muted mb-1">Secondary</h6>
+            <div class="fs-3 text-info">@secondaries</div>
+        </div></div></div>
+        <div class="col-md-3"><div class="card @(staleCount > 0 ? "border-warning" : "")"><div class="card-body">
+            <h6 class="text-muted mb-1">Stale</h6>
+            <div class="fs-3 @(staleCount > 0 ? "text-warning" : "")">@staleCount</div>
+        </div></div></div>
+    </div>
+
+    @if (primaries == 0 && standalone == 0)
+    {
+        <div class="alert alert-danger small mb-3">
+            No Primary or Standalone node — the cluster has no authoritative write target. Secondaries
+            stay read-only until one of them gets promoted via <code>RedundancyCoordinator</code>.
+        </div>
+    }
+    else if (primaries > 1)
+    {
+        <div class="alert alert-danger small mb-3">
+            <strong>Split-brain:</strong> @primaries nodes claim the Primary role. Apply-lease
+            enforcement should have made this impossible at the coordinator level. Investigate
+            immediately — one of the rows was likely hand-edited.
+        </div>
+    }
+
+    <table class="table table-sm table-hover align-middle">
+        <thead>
+            <tr>
+                <th>Node</th>
+                <th>Role</th>
+                <th>Host</th>
+                <th class="text-end">OPC UA port</th>
+                <th class="text-end">ServiceLevel base</th>
+                <th>ApplicationUri</th>
+                <th>Enabled</th>
+                <th>Last seen</th>
+            </tr>
+        </thead>
+        <tbody>
+        @foreach (var n in _nodes)
+        {
+            <tr class="@RowClass(n)">
+                <td><code>@n.NodeId</code></td>
+                <td><span class="badge @RoleBadge(n.RedundancyRole)">@n.RedundancyRole</span></td>
+                <td>@n.Host</td>
+                <td class="text-end"><code>@n.OpcUaPort</code></td>
+                <td class="text-end">@n.ServiceLevelBase</td>
+                <td class="small text-break"><code>@n.ApplicationUri</code></td>
+                <td>
+                    @if (n.Enabled) { <span class="badge bg-success">Enabled</span> }
+                    else { <span class="badge bg-secondary">Disabled</span> }
+                </td>
+                <td class="small @(ClusterNodeService.IsStale(n) ? "text-warning fw-bold" : "")">
+                    @(n.LastSeenAt is null ? "never" : FormatAge(n.LastSeenAt.Value))
+                    @if (ClusterNodeService.IsStale(n)) { <span class="badge bg-warning text-dark ms-1">Stale</span> }
+                </td>
+            </tr>
+        }
+        </tbody>
+    </table>
+}
+
+@code {
+    [Parameter] public string ClusterId { get; set; } = string.Empty;
+
+    private List<ClusterNode>? _nodes;
+
+    protected override async Task OnParametersSetAsync()
+    {
+        _nodes = await NodeSvc.ListByClusterAsync(ClusterId, CancellationToken.None);
+    }
+
+    private static string RowClass(ClusterNode n) =>
+        ClusterNodeService.IsStale(n) ? "table-warning" :
+        !n.Enabled ? "table-secondary" : "";
+
+    private static string RoleBadge(RedundancyRole r) => r switch
+    {
+        RedundancyRole.Primary    => "bg-success",
+        RedundancyRole.Secondary  => "bg-info",
+        RedundancyRole.Standalone => "bg-primary",
+        _                         => "bg-secondary",
+    };
+
+    private static string FormatAge(DateTime t)
+    {
+        var age = DateTime.UtcNow - t;
+        if (age.TotalSeconds < 60) return $"{(int)age.TotalSeconds}s ago";
+        if (age.TotalMinutes < 60) return $"{(int)age.TotalMinutes}m ago";
+        if (age.TotalHours < 24) return $"{(int)age.TotalHours}h ago";
+        return t.ToString("yyyy-MM-dd HH:mm 'UTC'");
+    }
+}

src/ZB.MOM.WW.OtOpcUa.Admin/Program.cs

@@ -48,6 +48,7 @@ builder.Services.AddScoped<ReservationService>();
 builder.Services.AddScoped<DraftValidationService>();
 builder.Services.AddScoped<AuditLogService>();
 builder.Services.AddScoped<HostStatusService>();
+builder.Services.AddScoped<ClusterNodeService>();
 builder.Services.AddScoped<EquipmentImportBatchService>();
 builder.Services.AddScoped<ZB.MOM.WW.OtOpcUa.Configuration.Services.ILdapGroupRoleMappingService,
     ZB.MOM.WW.OtOpcUa.Configuration.Services.LdapGroupRoleMappingService>();

src/ZB.MOM.WW.OtOpcUa.Admin/Services/ClusterNodeService.cs

@@ -0,0 +1,28 @@
+using Microsoft.EntityFrameworkCore;
+using ZB.MOM.WW.OtOpcUa.Configuration;
+using ZB.MOM.WW.OtOpcUa.Configuration.Entities;
+
+namespace ZB.MOM.WW.OtOpcUa.Admin.Services;
+
+/// <summary>
+///     Read-side service for ClusterNode rows + their cluster-scoped redundancy view. Consumed
+///     by the RedundancyTab on the cluster detail page. Writes (role swap, node enable/disable)
+///     are not supported here — role swap happens through the RedundancyCoordinator apply-lease
+///     flow on the server side and would conflict with any direct DB mutation from Admin.
+/// </summary>
+public sealed class ClusterNodeService(OtOpcUaConfigDbContext db)
+{
+    /// <summary>Stale-threshold matching <c>HostStatusService.StaleThreshold</c> — 30s of clock
+    /// tolerance covers a missed heartbeat plus publisher GC pauses.</summary>
+    public static readonly TimeSpan StaleThreshold = TimeSpan.FromSeconds(30);
+
+    public Task<List<ClusterNode>> ListByClusterAsync(string clusterId, CancellationToken ct) =>
+        db.ClusterNodes.AsNoTracking()
+            .Where(n => n.ClusterId == clusterId)
+            .OrderByDescending(n => n.ServiceLevelBase)
+            .ThenBy(n => n.NodeId)
+            .ToListAsync(ct);
+
+    public static bool IsStale(ClusterNode node) =>
+        node.LastSeenAt is null || DateTime.UtcNow - node.LastSeenAt.Value > StaleThreshold;
+}

tests/ZB.MOM.WW.OtOpcUa.Admin.Tests/ClusterNodeServiceTests.cs

@@ -0,0 +1,78 @@
+using Microsoft.EntityFrameworkCore;
+using Shouldly;
+using Xunit;
+using ZB.MOM.WW.OtOpcUa.Admin.Services;
+using ZB.MOM.WW.OtOpcUa.Configuration;
+using ZB.MOM.WW.OtOpcUa.Configuration.Entities;
+using ZB.MOM.WW.OtOpcUa.Configuration.Enums;
+
+namespace ZB.MOM.WW.OtOpcUa.Admin.Tests;
+
+[Trait("Category", "Unit")]
+public sealed class ClusterNodeServiceTests
+{
+    [Fact]
+    public void IsStale_NullLastSeen_Returns_True()
+    {
+        var node = NewNode("A", RedundancyRole.Primary, lastSeenAt: null);
+        ClusterNodeService.IsStale(node).ShouldBeTrue();
+    }
+
+    [Fact]
+    public void IsStale_RecentLastSeen_Returns_False()
+    {
+        var node = NewNode("A", RedundancyRole.Primary, lastSeenAt: DateTime.UtcNow.AddSeconds(-5));
+        ClusterNodeService.IsStale(node).ShouldBeFalse();
+    }
+
+    [Fact]
+    public void IsStale_Old_LastSeen_Returns_True()
+    {
+        var node = NewNode("A", RedundancyRole.Primary,
+            lastSeenAt: DateTime.UtcNow - ClusterNodeService.StaleThreshold - TimeSpan.FromSeconds(1));
+        ClusterNodeService.IsStale(node).ShouldBeTrue();
+    }
+
+    [Fact]
+    public async Task ListByClusterAsync_OrdersByServiceLevelBase_Descending_Then_NodeId()
+    {
+        using var ctx = NewContext();
+        ctx.ClusterNodes.AddRange(
+            NewNode("B-low", RedundancyRole.Secondary, serviceLevelBase: 150, clusterId: "c1"),
+            NewNode("A-high", RedundancyRole.Primary, serviceLevelBase: 200, clusterId: "c1"),
+            NewNode("other-cluster", RedundancyRole.Primary, serviceLevelBase: 200, clusterId: "c2"));
+        await ctx.SaveChangesAsync();
+
+        var svc = new ClusterNodeService(ctx);
+        var rows = await svc.ListByClusterAsync("c1", CancellationToken.None);
+
+        rows.Count.ShouldBe(2);
+        rows[0].NodeId.ShouldBe("A-high");   // higher ServiceLevelBase first
+        rows[1].NodeId.ShouldBe("B-low");
+    }
+
+    private static ClusterNode NewNode(
+        string nodeId,
+        RedundancyRole role,
+        DateTime? lastSeenAt = null,
+        int serviceLevelBase = 200,
+        string clusterId = "c1") => new()
+    {
+        NodeId = nodeId,
+        ClusterId = clusterId,
+        RedundancyRole = role,
+        Host = $"{nodeId}.example",
+        ApplicationUri = $"urn:{nodeId}",
+        ServiceLevelBase = (byte)serviceLevelBase,
+        LastSeenAt = lastSeenAt,
+        CreatedBy = "test",
+    };
+
+    private static OtOpcUaConfigDbContext NewContext()
+    {
+        var opts = new DbContextOptionsBuilder<OtOpcUaConfigDbContext>()
+            .UseInMemoryDatabase(Guid.NewGuid().ToString())
+            .Options;
+        return new OtOpcUaConfigDbContext(opts);
+    }
+}