Merge branch 'fix/standby-secret-mirror-status' (M1: standby mirror failures on the health report)

# Conflicts:
#	CLAUDE.md
#	docs/plans/2026-09-28-site-secrets-artifact-deploy-design.md
This commit is contained in:
Joseph Doherty
2026-09-28 21:37:03 -04:00
29 changed files with 1738 additions and 106 deletions
@@ -1,6 +1,7 @@
using System.Text.Json;
using Microsoft.Extensions.Logging.Abstractions;
using ZB.MOM.WW.ScadaBridge.Commons.Messages.Artifacts;
using ZB.MOM.WW.ScadaBridge.Commons.Messages.Health;
using ZB.MOM.WW.ScadaBridge.SiteRuntime.Persistence;
using ZB.MOM.WW.ScadaBridge.TestSupport;
@@ -303,6 +304,54 @@ public class DeployedSecretsStorageTests : IAsyncLifetime, IDisposable
Assert.Equal("A", mirrored.Name);
}
// ── secret_mirror_status (review finding M1) ──
[Fact]
public async Task SecretMirrorStatus_RoundTrips_OneRowPerNode_OrderedByNodeName()
{
var at = new DateTimeOffset(2026, 9, 28, 6, 0, 0, TimeSpan.Zero);
await _storage.SetSecretMirrorStatusAsync(
"node-b", [new SecretMirrorFailure("A", "did not decrypt")], at);
await _storage.SetSecretMirrorStatusAsync("node-a", [], at.AddMinutes(1));
var rows = await _storage.GetSecretMirrorStatusesAsync();
Assert.Equal(["node-a", "node-b"], rows.Select(r => r.NodeName));
Assert.Empty(rows[0].Failures);
Assert.Equal(at.AddMinutes(1), rows[0].CheckedAtUtc);
var failure = Assert.Single(rows[1].Failures);
Assert.Equal(new SecretMirrorFailure("A", "did not decrypt"), failure);
Assert.Equal(at, rows[1].CheckedAtUtc);
}
[Fact]
public async Task SecretMirrorStatus_Set_UpsertsTheNodesRow()
{
var at = new DateTimeOffset(2026, 9, 28, 6, 0, 0, TimeSpan.Zero);
await _storage.SetSecretMirrorStatusAsync("node-a", [new SecretMirrorFailure("A", "x")], at);
await _storage.SetSecretMirrorStatusAsync("node-a", [], at.AddHours(1));
var row = Assert.Single(await _storage.GetSecretMirrorStatusesAsync());
Assert.Empty(row.Failures);
Assert.Equal(at.AddHours(1), row.CheckedAtUtc);
}
[Fact]
public async Task SecretMirrorStatus_UnparseableRow_ReadsAsAFailure_NotAsHealthy()
{
await using (var connection = _storage.CreateConnection())
await using (var insert = connection.CreateCommand())
{
insert.CommandText =
"INSERT INTO secret_mirror_status (node_name, failures_json, checked_at) VALUES ('node-b', '{not json', '2026-09-28T00:00:00.0000000+00:00')";
await insert.ExecuteNonQueryAsync();
}
var row = Assert.Single(await _storage.GetSecretMirrorStatusesAsync());
Assert.Equal("node-b", row.NodeName);
Assert.Single(row.Failures);
}
private static SecretArtifact Art(string name, long revision) => new(
Name: name,
ContentType: "application/octet-stream",
@@ -23,14 +23,14 @@ public class SiteStorageSchemaTests
SiteStorageSchema.Apply(connection);
SiteStorageSchema.Apply(connection); // second run must not throw
// Every table is named explicitly on purpose — a loop asserting "11 tables"
// Every table is named explicitly on purpose — a loop asserting "12 tables"
// would still pass if one were renamed.
foreach (var table in new[]
{
"deployed_configurations", "static_attribute_overrides", "shared_scripts",
"external_systems", "database_connections", "notification_lists",
"data_connection_definitions", "smtp_configurations", "native_alarm_state",
"deployed_secrets", "mirrored_secrets",
"deployed_secrets", "mirrored_secrets", "secret_mirror_status",
})
{
Assert.True(TableExists(connection, table), $"missing table: {table}");
@@ -2,6 +2,7 @@ using System.Collections.Concurrent;
using Microsoft.Extensions.Logging;
using Microsoft.Extensions.Logging.Abstractions;
using Microsoft.Extensions.Options;
using ZB.MOM.WW.ScadaBridge.Commons.Interfaces.Services;
using ZB.MOM.WW.ScadaBridge.Commons.Messages.Artifacts;
using ZB.MOM.WW.ScadaBridge.SiteRuntime.Persistence;
using ZB.MOM.WW.ScadaBridge.SiteRuntime.Secrets;
@@ -655,6 +656,132 @@ public class DeployedSecretsMirrorTests : IDisposable
}
}
// ── Per-node status row for central (review finding M1) ──
[Fact]
public async Task Status_FirstPassAfterStartup_WritesThisNodesRow_EvenWhenHealthy()
{
var time = new FakeTimeProvider(new DateTimeOffset(2026, 9, 28, 6, 0, 0, TimeSpan.Zero));
var (mirror, storage, _, _, _) = await CreateAsync(time: time, nodeIdentity: new FixedNodeIdentity("node-b"));
await mirror.ReconcileAsync(CancellationToken.None);
var status = Assert.Single(await storage.GetSecretMirrorStatusesAsync());
Assert.Equal("node-b", status.NodeName);
Assert.Empty(status.Failures);
Assert.Equal(time.Now, status.CheckedAtUtc);
}
[Fact]
public async Task Status_UnchangedFailureSet_IsNotRewrittenOnLaterPasses()
{
var time = new FakeTimeProvider(new DateTimeOffset(2026, 9, 28, 6, 0, 0, TimeSpan.Zero));
var (mirror, storage, _, resolver, _) = await CreateAsync(time: time, nodeIdentity: new FixedNodeIdentity("node-b"));
resolver.ThrowFor("a");
await storage.UpsertDeployedSecretsAsync([Art("a", 1, kek: "kek-central")], "d1");
await mirror.ReconcileAsync(CancellationToken.None);
var firstWrite = Assert.Single(await storage.GetSecretMirrorStatusesAsync());
// Same failure on every later pass: the row is replicated, so it must not be rewritten.
time.Now = time.Now.AddMinutes(5);
await mirror.ReconcileAsync(CancellationToken.None);
await mirror.ReconcileAsync(CancellationToken.None);
var after = Assert.Single(await storage.GetSecretMirrorStatusesAsync());
Assert.Equal(firstWrite.CheckedAtUtc, after.CheckedAtUtc);
var failure = Assert.Single(after.Failures);
Assert.Equal("a", failure.SecretName);
Assert.Contains("kek-central", failure.Message);
}
[Fact]
public async Task Status_FailureSetChange_RewritesTheRow_AndClearsWhenHealed()
{
var time = new FakeTimeProvider(new DateTimeOffset(2026, 9, 28, 6, 0, 0, TimeSpan.Zero));
var (mirror, storage, _, resolver, _) = await CreateAsync(time: time, nodeIdentity: new FixedNodeIdentity("node-a"));
await storage.UpsertDeployedSecretsAsync([Art("a", 1)], "d1");
await mirror.ReconcileAsync(CancellationToken.None);
Assert.Empty(Assert.Single(await storage.GetSecretMirrorStatusesAsync()).Failures);
// A decrypt failure appears -> row rewritten with the failure.
resolver.ThrowFor("a");
time.Now = time.Now.AddMinutes(1);
await mirror.ReconcileAsync(CancellationToken.None);
var failing = Assert.Single(await storage.GetSecretMirrorStatusesAsync());
Assert.Equal("a", Assert.Single(failing.Failures).SecretName);
Assert.Equal(time.Now, failing.CheckedAtUtc);
// Healed -> row rewritten empty.
resolver.ClearFailures();
time.Now = time.Now.AddMinutes(1);
await mirror.ReconcileAsync(CancellationToken.None);
var healed = Assert.Single(await storage.GetSecretMirrorStatusesAsync());
Assert.Empty(healed.Failures);
Assert.Equal(time.Now, healed.CheckedAtUtc);
}
[Fact]
public async Task Status_RowCarriesNamesAndMessagesOnly_NeverCiphertext()
{
var (mirror, storage, _, resolver, _) = await CreateAsync(nodeIdentity: new FixedNodeIdentity("node-a"));
resolver.ThrowFor("a");
await storage.UpsertDeployedSecretsAsync([Art("a", 1)], "d1");
await mirror.ReconcileAsync(CancellationToken.None);
await using var connection = storage.CreateConnection();
await using var command = connection.CreateCommand();
command.CommandText = "SELECT failures_json FROM secret_mirror_status WHERE node_name = 'node-a'";
var json = (string)(await command.ExecuteScalarAsync())!;
Assert.Contains("\"SecretName\":\"a\"", json);
Assert.DoesNotContain("Ciphertext", json);
Assert.DoesNotContain(Convert.ToBase64String(new byte[] { 1, 2, 3, 4, 5 }), json);
}
[Fact]
public async Task Status_WriteFailure_DoesNotBreakThePass_AndIsRetriedNextPass()
{
var logger = new CapturingLogger<DeployedSecretsMirror>();
var (mirror, storage, store, _, _) = await CreateAsync(logger: logger, nodeIdentity: new FixedNodeIdentity("node-a"));
await storage.UpsertDeployedSecretsAsync([Art("a", 1)], "d1");
await using (var connection = storage.CreateConnection())
await using (var drop = connection.CreateCommand())
{
drop.CommandText = "DROP TABLE secret_mirror_status";
await drop.ExecuteNonQueryAsync();
}
var r = await mirror.ReconcileAsync(CancellationToken.None);
Assert.Equal(["a"], r.Applied);
Assert.True(store.Contains("a"));
Assert.Contains(logger.Entries, e => e.Level == LogLevel.Warning && e.Message.Contains("mirror status"));
// Table back (e.g. transient fault cleared): the unchanged failure set is still
// written, because the earlier write never landed.
await storage.InitializeAsync();
await mirror.ReconcileAsync(CancellationToken.None);
Assert.Equal("node-a", Assert.Single(await storage.GetSecretMirrorStatusesAsync()).NodeName);
}
[Fact]
public async Task Status_WithoutNodeIdentity_FallsBackToMachineName()
{
var (mirror, storage, _, _, _) = await CreateAsync();
await mirror.ReconcileAsync(CancellationToken.None);
Assert.Equal(Environment.MachineName, Assert.Single(await storage.GetSecretMirrorStatusesAsync()).NodeName);
}
private sealed class FixedNodeIdentity(string? name) : INodeIdentityProvider
{
public string? NodeName { get; } = name;
}
// ── Harness ──
private async Task<(
@@ -665,7 +792,8 @@ public class DeployedSecretsMirrorTests : IDisposable
FakeSecretCacheInvalidator Invalidator)> CreateAsync(
CapturingLogger<DeployedSecretsMirror>? logger = null,
TimeSpan? interval = null,
TimeProvider? time = null)
TimeProvider? time = null,
INodeIdentityProvider? nodeIdentity = null)
{
var storage = new SiteStorageService(_localDb.Db, NullLogger<SiteStorageService>.Instance);
await storage.InitializeAsync();
@@ -685,7 +813,8 @@ public class DeployedSecretsMirrorTests : IDisposable
invalidator,
options,
(ILogger<DeployedSecretsMirror>?)logger ?? NullLogger<DeployedSecretsMirror>.Instance,
time ?? TimeProvider.System);
time ?? TimeProvider.System,
nodeIdentity);
return (mirror, storage, store, resolver, invalidator);
}
@@ -0,0 +1,95 @@
using Microsoft.Extensions.Logging.Abstractions;
using ZB.MOM.WW.ScadaBridge.Commons.Messages.Health;
using ZB.MOM.WW.ScadaBridge.HealthMonitoring;
using ZB.MOM.WW.ScadaBridge.SiteRuntime.Persistence;
using ZB.MOM.WW.ScadaBridge.SiteRuntime.Secrets;
using ZB.MOM.WW.ScadaBridge.TestSupport;
namespace ZB.MOM.WW.ScadaBridge.SiteRuntime.Tests.Secrets;
/// <summary>
/// Review finding M1: <see cref="SecretMirrorStatusReporter"/> forwards every node's
/// replicated <c>secret_mirror_status</c> row onto the site health report via the collector.
/// </summary>
public class SecretMirrorStatusReporterTests : IAsyncLifetime, IDisposable
{
private readonly TestLocalDb _localDb = TestLocalDb.CreateTemp("secret-mirror-status");
private SiteStorageService _storage = null!;
public async Task InitializeAsync()
{
_storage = new SiteStorageService(_localDb.Db, NullLogger<SiteStorageService>.Instance);
await _storage.InitializeAsync();
}
public Task DisposeAsync() => Task.CompletedTask;
public void Dispose()
{
var path = _localDb.Path;
_localDb.Dispose();
TestLocalDb.DeleteFiles(path);
}
[Fact]
public async Task Probe_PushesEveryNodesRow_OntoTheNextReport()
{
var at = new DateTimeOffset(2026, 9, 28, 6, 0, 0, TimeSpan.Zero);
await _storage.SetSecretMirrorStatusAsync("node-a", [], at);
await _storage.SetSecretMirrorStatusAsync(
"node-b", [new SecretMirrorFailure("OPC-PWD", "did not decrypt on this node")], at);
var collector = new SiteHealthCollector();
var reporter = new SecretMirrorStatusReporter(
collector, _storage, NullLogger<SecretMirrorStatusReporter>.Instance);
await reporter.ProbeAsync();
var report = collector.CollectReport("site-a");
Assert.NotNull(report.SecretMirrorStatuses);
Assert.Equal(["node-a", "node-b"], report.SecretMirrorStatuses!.Select(s => s.NodeName));
Assert.Equal("OPC-PWD", Assert.Single(report.SecretMirrorStatuses[1].Failures).SecretName);
}
[Fact]
public void BeforeAnyProbe_TheReportFieldIsNull()
{
var report = new SiteHealthCollector().CollectReport("site-a");
Assert.Null(report.SecretMirrorStatuses);
}
[Fact]
public async Task Probe_NoRowsYet_PushesAnEmptyList_NotNull()
{
var collector = new SiteHealthCollector();
var reporter = new SecretMirrorStatusReporter(
collector, _storage, NullLogger<SecretMirrorStatusReporter>.Instance);
await reporter.ProbeAsync();
Assert.Empty(collector.CollectReport("site-a").SecretMirrorStatuses!);
}
[Fact]
public async Task Probe_ReadFailure_KeepsThePreviousSnapshot_AndDoesNotThrow()
{
await _storage.SetSecretMirrorStatusAsync("node-a", [], DateTimeOffset.UtcNow);
var collector = new SiteHealthCollector();
var reporter = new SecretMirrorStatusReporter(
collector, _storage, NullLogger<SecretMirrorStatusReporter>.Instance);
await reporter.ProbeAsync();
await using (var connection = _storage.CreateConnection())
await using (var drop = connection.CreateCommand())
{
drop.CommandText = "DROP TABLE secret_mirror_status";
await drop.ExecuteNonQueryAsync();
}
await reporter.ProbeAsync();
// Point-in-time: not drained on collect, and not cleared by a failed read.
Assert.Single(collector.CollectReport("site-a").SecretMirrorStatuses!);
Assert.Single(collector.CollectReport("site-a").SecretMirrorStatuses!);
}
}