Files
lmxopcua/tests/Server/ZB.MOM.WW.OtOpcUa.Host.IntegrationTests/LocalDb/LocalDbPairHarness.cs
T
Joseph Doherty 232bff5df7 fix(localdb): close both Phase 1 live-gate follow-ups
1. Wiped-node back-fill (live gate check 4). Fixed upstream in
   ZB.MOM.WW.LocalDb 0.1.2 and pinned here. The gate's diagnosis was slightly
   off: the oplog cap was not the gate. Snapshot detection measured the peer's
   gap against the oldest surviving oplog row and read an empty oplog as "no
   gap possible" — and an empty oplog is the steady state of a converged pair,
   since ack-pruning deletes everything the peer confirmed. The healthy state
   was the one state that could not heal a wiped peer. Now measured against
   last_acked_seq when the oplog is empty.

   Pinned at this level too, not just the library's: the pair harness grows a
   WipePassiveAsync (a NEW database — a rebuilt node comes back with a new node
   id and a zero watermark), and the new scenario asserts the emptied oplog as
   its precondition before wiping, then requires the deployment artifact to
   come back byte-identical with no new deploy. Verified RED against the pinned
   0.1.1 (times out waiting for back-fill) and green on 0.1.2.

2. Oplog growth on default-OFF nodes (live gate check 8), fixed at the source:
   StoreAsync now skips entirely when the pointer already names this
   deployment/revision, the SHA matches the bytes, and the expected chunk count
   is present. Re-caching an artifact the node already holds — every restart's
   boot-from-cache, every RestoreApplied — writes nothing and mints no oplog
   rows, where before it cost a delete plus an insert per chunk plus a pointer
   update, all identical to what was already there.

   Identity is over the bytes and the chunks are counted, both deliberately: a
   re-composed artifact can carry the same ids with different bytes, and a
   pointer can name a deployment whose chunks are missing, where skipping would
   make an unreadable cache permanent. Both guards verified RED against a naive
   pointer-only skip.

   The gate's stated bound was also wrong and is corrected in the doc: growth
   was never headed for the 1M row cap. AddZbLocalDbReplication is registered
   unconditionally, so MaintenanceBackgroundService runs on default-OFF nodes
   and the 7-day MaxOplogAge cap prunes.

Runtime.Tests 412/0/31, Host.IntegrationTests LocalDb 45/45. Pre-existing and
unrelated: AbCip_Green_AgainstSim (needs the AB CIP docker fixture, red on a
clean master too) and a flaky Roslyn race test (green 2/2 in isolation).

Claude-Session: https://claude.ai/code/session_01GASWkNEi68FSCtvr6rLoEW
2026-07-21 01:27:02 -04:00

367 lines
15 KiB
C#

using System.Net;
using Microsoft.AspNetCore.Builder;
using Microsoft.AspNetCore.Hosting;
using Microsoft.AspNetCore.Hosting.Server;
using Microsoft.AspNetCore.Hosting.Server.Features;
using Microsoft.AspNetCore.Server.Kestrel.Core;
using Microsoft.Extensions.Configuration;
using Microsoft.Extensions.DependencyInjection;
using Microsoft.Extensions.Hosting;
using Microsoft.Extensions.Logging.Abstractions;
using Xunit;
using ZB.MOM.WW.LocalDb;
using ZB.MOM.WW.LocalDb.Replication;
using ZB.MOM.WW.OtOpcUa.Host.Configuration;
using ZB.MOM.WW.OtOpcUa.Runtime.DeploymentCache;
namespace ZB.MOM.WW.OtOpcUa.Host.IntegrationTests.LocalDb;
/// <summary>
/// Serializes the two-node convergence tests against one another: each stands up a real Kestrel
/// h2c listener plus two SQLite files, and running them concurrently under CI contention is a
/// flakiness risk. The rest of the assembly parallelizes normally.
/// </summary>
[CollectionDefinition("LocalDbPairConvergence")]
public sealed class LocalDbPairConvergenceCollection;
/// <summary>
/// Two driver-role OtOpcUa nodes replicating the deployment-artifact cache over a REAL loopback
/// gRPC transport (Kestrel h2c on 127.0.0.1), through the REAL fail-closed
/// <see cref="LocalDbSyncAuthInterceptor"/>. Node A is the initiator (it dials the peer); node B
/// is passive (it hosts <c>MapZbLocalDbSync</c>).
/// </summary>
/// <remarks>
/// <para>
/// Both databases are initialised through the production <see cref="LocalDbSetup.OnReady"/> —
/// a hand-written schema here would prove only that the test agrees with itself. The tables,
/// their primary keys, and the DDL→<c>RegisterReplicated</c> ordering under test are the ones
/// the host actually runs.
/// </para>
/// <para>
/// The two <see cref="ILocalDb"/> instances are owned by the harness and registered into the
/// hosts as pre-constructed singletons. MS.DI does not dispose instances it did not create,
/// so tearing a host down (the transport-loss scenario) leaves the databases intact and
/// writable — which is exactly what lets node A accumulate writes while B is down.
/// </para>
/// <para>
/// A real loopback socket (not an in-memory TestServer) is deliberate: killing the passive
/// host disposes the server and closes the socket, faulting the initiator's active stream
/// promptly. An in-memory TestServer does not model a connection drop and leaves the client
/// hanging.
/// </para>
/// <para>
/// The API keys are per-node so the wrong-key scenario can hand A and B different keys and
/// assert non-convergence — with a matching-key positive control proving the same harness
/// does converge when the keys agree (an absence assertion without a positive control passed
/// vacuously in ScadaBridge).
/// </para>
/// <para>Offline: no docker, no external services.</para>
/// </remarks>
public sealed class LocalDbPairHarness : IAsyncDisposable
{
/// <summary>The key both nodes share unless a test overrides one of them.</summary>
public const string DefaultApiKey = "otopcua-localdb-pair-convergence-key";
/// <summary>How long a scenario waits for the pair to agree (or to stay diverged) before deciding.</summary>
public static readonly TimeSpan ConvergeTimeout = TimeSpan.FromSeconds(30);
private readonly string _apiKeyA;
private readonly string _apiKeyB;
private readonly string _pathA = Path.Combine(Path.GetTempPath(), $"otopcua-pairA-{Guid.NewGuid():N}.db");
// Not readonly: WipePassiveAsync replaces node B's database wholesale, which is what a rebuilt
// node with a lost volume actually is — a new file, and with it a new node id.
private string _pathB = Path.Combine(Path.GetTempPath(), $"otopcua-pairB-{Guid.NewGuid():N}.db");
private readonly ServiceProvider _dbProviderA;
private ServiceProvider _dbProviderB;
private IHost? _serverHost; // node B — passive
private IHost? _initiatorHost; // node A — dials the peer
static LocalDbPairHarness() =>
// Grpc.Net.Client dials the loopback server over HTTP/2 cleartext (h2c).
AppContext.SetSwitch("System.Net.Http.SocketsHttpHandler.Http2UnencryptedSupport", true);
/// <param name="apiKeyA">Node A's replication key. Defaults to <see cref="DefaultApiKey"/>.</param>
/// <param name="apiKeyB">Node B's replication key. Defaults to <see cref="DefaultApiKey"/>.</param>
public LocalDbPairHarness(string? apiKeyA = null, string? apiKeyB = null)
{
_apiKeyA = apiKeyA ?? DefaultApiKey;
_apiKeyB = apiKeyB ?? DefaultApiKey;
_dbProviderA = BuildDatabaseProvider(_pathA);
_dbProviderB = BuildDatabaseProvider(_pathB);
}
/// <summary>Node A — the initiator, which dials the peer.</summary>
public ILocalDb A => _dbProviderA.GetRequiredService<ILocalDb>();
/// <summary>Node B — the passive node, which listens.</summary>
public ILocalDb B => _dbProviderB.GetRequiredService<ILocalDb>();
/// <summary>The production artifact cache over node A's database.</summary>
public IDeploymentArtifactCache CacheA =>
new LocalDbDeploymentArtifactCache(A, NullLogger<LocalDbDeploymentArtifactCache>.Instance);
/// <summary>The production artifact cache over node B's database.</summary>
public IDeploymentArtifactCache CacheB =>
new LocalDbDeploymentArtifactCache(B, NullLogger<LocalDbDeploymentArtifactCache>.Instance);
// ---- lifecycle --------------------------------------------------------------------------
/// <summary>Forces both databases to construct (running OnReady) then brings the pair online.</summary>
public async Task StartAsync()
{
_ = A;
_ = B;
await StartPassiveAsync();
await StartInitiatorAsync();
}
/// <summary>Takes node B's listener down, leaving its database intact and writable.</summary>
public async Task StopPassiveAsync()
{
await StopHostAsync(_serverHost);
_serverHost = null;
}
/// <summary>
/// Destroys node B's database and replaces it with an empty one, leaving the pairing config
/// untouched — the rebuilt-node case: a lost volume, a re-imaged host, a fresh container.
/// </summary>
/// <remarks>
/// A new file, not a truncated one, because that is what the real thing is: the node id lives
/// in the database, so a rebuilt node comes back with a new identity and a zero watermark. The
/// caller follows with <see cref="RestartPairAsync"/> to bring the wiped node back online.
/// </remarks>
public async Task WipePassiveAsync()
{
await StopPassiveAsync();
await _dbProviderB.DisposeAsync();
// Pooled connections outlive the provider; without this the delete silently no-ops on
// Windows and the "wiped" node would still hold its rows.
Microsoft.Data.Sqlite.SqliteConnection.ClearAllPools();
DeleteDatabaseFiles(_pathB);
_pathB = Path.Combine(Path.GetTempPath(), $"otopcua-pairB-{Guid.NewGuid():N}.db");
_dbProviderB = BuildDatabaseProvider(_pathB);
_ = B;
}
/// <summary>
/// Brings node B back on a NEW loopback port and re-dials from A. The initiator re-reads the
/// peer address on each reconnect, so this is a genuine rejoin over the same databases.
/// </summary>
public async Task RestartPairAsync()
{
await StartPassiveAsync();
await StopHostAsync(_initiatorHost);
_initiatorHost = null;
await StartInitiatorAsync();
}
// ---- convergence helpers ----------------------------------------------------------------
/// <summary>Polls <paramref name="condition"/> until true or the deadline passes; fails otherwise.</summary>
public static async Task WaitUntilAsync(Func<Task<bool>> condition, string because)
{
var deadline = DateTime.UtcNow + ConvergeTimeout;
while (DateTime.UtcNow < deadline)
{
if (await condition())
return;
await Task.Delay(50);
}
Assert.Fail($"Timed out after {ConvergeTimeout.TotalSeconds:0}s waiting for: {because}");
}
/// <summary>Waits out a bounded window and returns whether <paramref name="condition"/> ever held.</summary>
/// <remarks>
/// For the negative half of the wrong-key scenario: it must NOT converge. A short, fixed
/// window keeps the test quick while still giving a matching-key control ample time to
/// converge (the control uses <see cref="WaitUntilAsync"/> with the full timeout).
/// </remarks>
public static async Task<bool> HeldWithinAsync(Func<Task<bool>> condition, TimeSpan window)
{
var deadline = DateTime.UtcNow + window;
while (DateTime.UtcNow < deadline)
{
if (await condition())
return true;
await Task.Delay(50);
}
return false;
}
/// <summary>
/// Dumps the <c>__localdb_row_version</c> rows for one table, ordered by pk, as
/// <c>pk_json|hlc|node_id|is_tombstone</c>. Equal dumps on both nodes prove B holds A's
/// origin-stamped row (same HLC + node id), not a locally re-derived one.
/// </summary>
public static async Task<string> DumpRowVersionAsync(ILocalDb db, string table)
{
var rows = await db.QueryAsync(
"""
SELECT pk_json, hlc, node_id, is_tombstone
FROM __localdb_row_version
WHERE table_name = @Table
ORDER BY pk_json
""",
static r => $"{r.GetString(0)}|{r.GetInt64(1)}|{r.GetString(2)}|{r.GetInt64(3)}",
new { Table = table });
return string.Join("\n", rows);
}
/// <summary>Counts <c>deployment_artifacts</c> chunk rows for one deployment on a node.</summary>
public static async Task<long> ChunkCountAsync(ILocalDb db, string deploymentId)
{
var rows = await db.QueryAsync(
"SELECT COUNT(*) FROM deployment_artifacts WHERE deployment_id = @DeploymentId",
static r => r.GetInt64(0),
new { DeploymentId = deploymentId });
return rows[0];
}
/// <summary>
/// Counts oplog rows on a node. Zero means every local change has been acked by the peer and
/// pruned — the steady state of a converged pair, and the state from which a wiped peer can
/// only be healed by a snapshot.
/// </summary>
public static async Task<long> OplogDepthAsync(ILocalDb db)
{
var rows = await db.QueryAsync("SELECT COUNT(*) FROM __localdb_oplog", static r => r.GetInt64(0));
return rows[0];
}
/// <summary>Counts tombstone rows in <c>__localdb_row_version</c> for one table on a node.</summary>
public static async Task<long> TombstoneCountAsync(ILocalDb db, string table)
{
var rows = await db.QueryAsync(
"""
SELECT COUNT(*) FROM __localdb_row_version
WHERE table_name = @Table AND is_tombstone = 1
""",
static r => r.GetInt64(0),
new { Table = table });
return rows[0];
}
// ---- fixture internals ------------------------------------------------------------------
/// <summary>
/// A provider owning one deployment-cache database, initialised through the host's own
/// <see cref="LocalDbSetup.OnReady"/> — same schema, same registration order.
/// </summary>
private static ServiceProvider BuildDatabaseProvider(string path)
{
var config = new ConfigurationBuilder()
.AddInMemoryCollection(new Dictionary<string, string?> { ["LocalDb:Path"] = path })
.Build();
return new ServiceCollection()
.AddZbLocalDb(config, LocalDbSetup.OnReady)
.BuildServiceProvider();
}
private IConfiguration ReplicationConfig(string apiKey, string? peerAddress)
{
var values = new Dictionary<string, string?>
{
// Tight flush + bounded reconnect backoff so convergence is observable well inside the
// poll deadline. The 60 s production default would let the doubling backoff overrun it
// after a peer outage.
["LocalDb:Replication:FlushInterval"] = "00:00:00.050",
["LocalDb:Replication:ReconnectBackoffMax"] = "00:00:02",
["LocalDb:Replication:ApiKey"] = apiKey,
};
if (peerAddress is not null)
values["LocalDb:Replication:PeerAddress"] = peerAddress;
return new ConfigurationBuilder().AddInMemoryCollection(values).Build();
}
/// <summary>Starts node B, the passive listener, behind the real auth interceptor.</summary>
private async Task StartPassiveAsync()
{
var config = ReplicationConfig(_apiKeyB, peerAddress: null);
_serverHost = await new HostBuilder()
.ConfigureWebHost(web =>
{
web.UseKestrel(o =>
o.Listen(IPAddress.Loopback, 0, listen => listen.Protocols = HttpProtocols.Http2));
web.ConfigureServices(services =>
{
services.AddLogging();
services.AddRouting();
// The REAL interceptor, not a stand-in. If it rejected legitimate peer traffic,
// every matching-key scenario would fail — which is the point.
services.AddGrpc(o => o.Interceptors.Add<LocalDbSyncAuthInterceptor>());
services.AddSingleton(B);
services.AddZbLocalDbReplication(config);
})
.Configure(app =>
{
app.UseRouting();
app.UseEndpoints(e => e.MapZbLocalDbSync());
});
})
.StartAsync();
}
/// <summary>Starts node A, which dials the passive node.</summary>
private async Task StartInitiatorAsync()
{
var config = ReplicationConfig(_apiKeyA, PassiveAddress());
_initiatorHost = await new HostBuilder()
.ConfigureServices(services =>
{
services.AddLogging();
services.AddSingleton(A);
services.AddZbLocalDbReplication(config);
})
.StartAsync();
}
private string PassiveAddress()
=> _serverHost!.Services.GetRequiredService<IServer>()
.Features.Get<IServerAddressesFeature>()!.Addresses.Single();
private static async Task StopHostAsync(IHost? host)
{
if (host is null)
return;
try { await host.StopAsync(TimeSpan.FromSeconds(5)); } catch { /* teardown */ }
host.Dispose();
}
public async ValueTask DisposeAsync()
{
await StopHostAsync(_initiatorHost);
await StopHostAsync(_serverHost);
await _dbProviderA.DisposeAsync();
await _dbProviderB.DisposeAsync();
Microsoft.Data.Sqlite.SqliteConnection.ClearAllPools();
DeleteDatabaseFiles(_pathA);
DeleteDatabaseFiles(_pathB);
}
/// <summary>Deletes a SQLite database and its WAL/shm sidecars, best effort.</summary>
private static void DeleteDatabaseFiles(string path)
{
foreach (var suffix in new[] { "", "-wal", "-shm" })
{
try { File.Delete(path + suffix); } catch { /* best effort */ }
}
}
}