diff --git a/docs/drivers/AbLegacy-Test-Fixture.md b/docs/drivers/AbLegacy-Test-Fixture.md
index 2c7f459a..ecf9894d 100644
--- a/docs/drivers/AbLegacy-Test-Fixture.md
+++ b/docs/drivers/AbLegacy-Test-Fixture.md
@@ -36,8 +36,9 @@ supplies a `FakeAbLegacyTag`.
The `AbLegacy.IntegrationTests` suites now run **once per transport** (`TransportMatrix`: `Libplctag` and
`Native`). Set `AB_LEGACY_COMPOSE_PROFILE` to the running profile. On the 2026-10-07 run native passed
-everything libplctag passed on `slc500`, `micrologix` and `plc5` (the PLC-5 test reads N7 only, so the PLC-5
-refusal paths and status-bit writes are not exercised by it). Native PCCC is **simulator-only**. Write-capable
+everything libplctag passed on `slc500`, `micrologix` and `plc5` (the smoke tests read N7 only; the PLC-5
+refusal paths and status-bit write refusals are exercised by `AbLegacyUnsupportedPathTests`, native only, #593 — the
+refusals are the driver's own, raised before the wire, so each test carries a positive-control read). Native PCCC is **simulator-only**. Write-capable
tests (the SLC500 N7:5 round trip) require `AB_SERVER_ALLOW_WRITES=1`. Golden frames:
`scripts/ab-capture/capture.sh`, fixtures under
`tests/Drivers/ZB.MOM.WW.OtOpcUa.Driver.Ab.Transport.Tests/Captures/`. Evidence:
diff --git a/docs/drivers/AbServer-Test-Fixture.md b/docs/drivers/AbServer-Test-Fixture.md
index 74a071e8..7c22622e 100644
--- a/docs/drivers/AbServer-Test-Fixture.md
+++ b/docs/drivers/AbServer-Test-Fixture.md
@@ -37,7 +37,7 @@ quirk. UDT / alarm / quirk behavior is verified only by unit tests with
Every integration test in `AbCip.IntegrationTests` now runs **once per transport** (`TransportMatrix`:
`Libplctag` and `Native`). Run one compose profile at a time and name it with `AB_SERVER_COMPOSE_PROFILE`
-(all profiles bind `:44818`, so set it to the running profile). On the 2026-10-07 run native passed everything
+(all profiles bind `:44818`, so set it to the running profile; **unset, only the default `controllogix` rows run** — ab_server does not self-identify its `--plc` mode, so crossing every family at one container failed the non-matching rows instead of skipping them, #594). `NativeLeaseReleaseSoakTests` is the ab_server soak of the background lease release (`AB_LEASE_SOAK_CYCLES`, default 100; read-only, skips without the fixture, #593). On the 2026-10-07 run native passed everything
libplctag passed on all seven profiles; on `micro800` libplctag failed on the empty-path defect described in
[AbCip.md](AbCip.md#known-gaps) and native passed. That defect is now fixed: a re-run of the `micro800`
smoke test passes on both transports. Evidence:
diff --git a/tests/Drivers/ZB.MOM.WW.OtOpcUa.Driver.AbCip.IntegrationTests/NativeLeaseReleaseSoakTests.cs b/tests/Drivers/ZB.MOM.WW.OtOpcUa.Driver.AbCip.IntegrationTests/NativeLeaseReleaseSoakTests.cs
new file mode 100644
index 00000000..0624ebb8
--- /dev/null
+++ b/tests/Drivers/ZB.MOM.WW.OtOpcUa.Driver.AbCip.IntegrationTests/NativeLeaseReleaseSoakTests.cs
@@ -0,0 +1,77 @@
+using Shouldly;
+using Xunit;
+using ZB.MOM.WW.OtOpcUa.Driver.Ab.Transport;
+
+namespace ZB.MOM.WW.OtOpcUa.Driver.AbCip.IntegrationTests;
+
+///
+/// Gitea #579 / #593: soak of ConnectionLease.ReleaseInBackground against a real ab_server.
+/// The offline NativeAbCipLeaseReleaseTests pin the ordering against a fake CIP target for 25 cycles;
+/// this drives the same release-then-immediately-recreate loop over a real socket and the simulator's own
+/// connection bookkeeping, where a Forward Open racing the predecessor's still-in-flight Forward Close would
+/// be refused or would leak a connection slot.
+///
+///
+/// Skips when ab_server is unreachable (). Cycle count:
+/// AB_LEASE_SOAK_CYCLES (default 100). Read-only against the target (one DINT read per cycle), so it
+/// needs no write opt-in.
+///
+[Trait("Category", "Integration")]
+[Trait("Requires", "AbServer")]
+public sealed class NativeLeaseReleaseSoakTests
+{
+ private static readonly TimeSpan Guard = TimeSpan.FromSeconds(60);
+
+ private static int Cycles =>
+ int.TryParse(Environment.GetEnvironmentVariable("AB_LEASE_SOAK_CYCLES"), out var n) && n > 0 ? n : 100;
+
+ ///
+ /// Each cycle: create a native runtime, initialise, read, then Dispose (the background release) and
+ /// re-create at once. Every read must be Good, exactly one Forward Open and one Forward Close per cycle
+ /// must hit the wire (a re-create that overlapped the close, or a close that never ran, shows as a count
+ /// mismatch), and the pool must end with no live entry.
+ ///
+ [AbServerFact]
+ public async Task Release_in_background_then_immediate_recreate_never_overlaps_or_leaks_against_ab_server()
+ {
+ var fixture = new AbServerFixture(KnownProfiles.ControlLogix);
+ var pool = new DeviceSessionPool();
+ var factory = new NativeAbCipTagFactory(pool);
+ var cycles = Cycles;
+ var teardowns = new List(cycles);
+
+ var createParams = new AbCipTagCreateParams(
+ Gateway: fixture.Host,
+ Port: fixture.Port,
+ CipPath: "1,0",
+ LibplctagPlcAttribute: "controllogix",
+ TagName: "TestDINT",
+ Timeout: TimeSpan.FromSeconds(5),
+ AllowPacking: false,
+ ElementCount: 1,
+ IsArray: false,
+ AuthoredType: AbCipDataType.DInt,
+ UseConnectedMessaging: true);
+
+ using var budget = CancellationTokenSource.CreateLinkedTokenSource(TestContext.Current.CancellationToken);
+ budget.CancelAfter(TimeSpan.FromSeconds(120));
+
+ for (var i = 0; i < cycles; i++)
+ {
+ var runtime = (NativeAbCipTagRuntime)factory.Create(createParams);
+ await runtime.InitializeAsync(budget.Token);
+ await runtime.ReadAsync(budget.Token);
+ runtime.GetStatus().ShouldBe(0, $"cycle {i}: read status");
+ runtime.DecodeValue(AbCipDataType.DInt, null).ShouldNotBeNull($"cycle {i}: decoded value");
+
+ runtime.Dispose(); // background release; the next Create must queue behind its close
+ teardowns.Add(runtime.TeardownTask ?? Task.CompletedTask);
+ }
+
+ await Task.WhenAll(teardowns).WaitAsync(Guard, TestContext.Current.CancellationToken);
+
+ pool.EntryCount.ShouldBe(0, "every lease was released, so no pooled connection may remain");
+ pool.Diagnostics.ForwardOpens.ShouldBe(cycles, "one Forward Open per cycle: a re-create must wait for its predecessor's close");
+ pool.Diagnostics.ForwardClosesSent.ShouldBe(cycles, "one Forward Close per cycle: none dropped, none doubled");
+ }
+}
diff --git a/tests/Drivers/ZB.MOM.WW.OtOpcUa.Driver.AbLegacy.IntegrationTests/Docker/README.md b/tests/Drivers/ZB.MOM.WW.OtOpcUa.Driver.AbLegacy.IntegrationTests/Docker/README.md
index 3aec0909..18b8aef2 100644
--- a/tests/Drivers/ZB.MOM.WW.OtOpcUa.Driver.AbLegacy.IntegrationTests/Docker/README.md
+++ b/tests/Drivers/ZB.MOM.WW.OtOpcUa.Driver.AbLegacy.IntegrationTests/Docker/README.md
@@ -54,7 +54,7 @@ families stop the current service + start another.
| `AB_LEGACY_ENDPOINT` | *unset* — per-family ports on `10.100.0.35` | `host:port` of a single PCCC device, pinning **every** family to it (real-hardware mode). |
| `AB_LEGACY_CIP_PATH` | `1,0` | CIP routing path portion of the `ab://host:port/` URI. ab_server rejects empty paths at the CIP unconnected-send layer; real SLC/MicroLogix/PLC-5 hardware accepts empty (no backplane). Set to empty (`AB_LEGACY_CIP_PATH=`) when pointing at real hardware. |
| `AB_SERVER_ALLOW_WRITES` | *unset* | **Write opt-in.** Every write-capable test (today: the SLC500 N7:5 write-then-read round trip) skips unless this is `1`. Set it for simulator runs; **never** set it when an endpoint variable points at production hardware. Write tests also refuse while `AB_DIFF_ENDPOINT` (the AbCip read-only differential) is set. |
-| `AB_LEGACY_COMPOSE_PROFILE` | *unset* | Only consulted when `AB_LEGACY_ENDPOINT` pins an endpoint: names which family that one device is, so the SLC500-only write test skips against other hardware. Against the Docker fixture it is unnecessary — each family has its own port, and each test skips itself if its own port is down. |
+| `AB_LEGACY_COMPOSE_PROFILE` | *unset* | Only consulted when `AB_LEGACY_ENDPOINT` pins an endpoint: names which family that one device is, so the SLC500-only write test skips against other hardware. When `AB_LEGACY_ENDPOINT` is set and this is not, only the `slc500` rows run (the family of a pinned device cannot be detected). Against the Docker fixture it is unnecessary — each family has its own port, and each test skips itself if its own port is down. |
## Run the integration tests