dohertj2/lmxopcua
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Close all four stability-review 2026-04-13 findings so a failed runtime probe subscription can no longer leave a phantom entry that Tick() flips to Stopped and fans out false BadOutOfService quality across a host's subtree, a silently-failed dashboard bind no longer lets the service advertise a successful start while an operator-visible endpoint is dead, the seven sync-over-async sites in LmxNodeManager (rebuild probe sync, Read, Write, four HistoryRead overrides) can no longer park the OPC UA stack thread indefinitely on a hung backend, and alarm auto-subscribe + transferred-subscription restore no longer race shutdown as untracked fire-and-forget tasks.

Browse Source 
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This commit is contained in:
Joseph Doherty
2026-04-14 00:48:07 -04:00
parent 731092595f
commit c76ab8fdee
21 changed files with 869 additions and 53 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
docs/MxAccessBridge.md
View File

@@ -108,6 +108,7 @@ Enabled by default via `MxAccess.RuntimeStatusProbesEnabled`; see [Configuration
2. **Transition predicate** — A probe callback is interpreted as `isRunning = vtq.Quality.IsGood() && vtq.Value is bool b && b`. Everything else (explicit `ScanState = false`, bad quality, communication errors from the broker) means **Stopped**.
3. **On-change-only delivery** — `ScanState` is delivered **only when the value actually changes**. A stably Running host may go hours without a callback. The probe manager's `Tick()` explicitly does NOT run a starvation check on Running entries — the only time-based transition is **Unknown → Stopped** when the initial callback hasn't arrived within `RuntimeStatusUnknownTimeoutSeconds` (default 15s). This protects against a probe that fails to resolve at all without incorrectly flipping healthy long-running hosts.
4. **Transport gating** — When `IMxAccessClient.State != Connected`, `GetSnapshot()` forces every entry to `Unknown` regardless of underlying state. The dashboard shows the Connection panel as the primary signal in that case rather than misleading operators with "every host stopped."
5. **Subscribe failure rollback** — If `SubscribeAsync` throws for a new probe (SDK failure, broker rejection, transport error), the manager rolls back both `_byProbe` and `_probeByGobjectId` so the probe never appears in `GetSnapshot()`. Without this rollback, a failed subscribe would leave the entry in `Unknown` forever, and `Tick()` would later transition it to `Stopped` after the unknown-resolution timeout, fanning out a **false-negative** host-down signal that invalidates the subtree of a host that was never actually advised. Stability review 2026-04-13 Finding 1.
### Subtree quality invalidation on transition
@@ -137,6 +138,14 @@ See the `runtimestatus.md` plan file and the `service_info.md` entry for the in-
See [Status Dashboard](StatusDashboard.md#galaxy-runtime) for the field table and [Configuration](Configuration.md#mxaccess) for the two new config fields.
## Request Timeout Safety Backstop
Every sync-over-async site on the OPC UA stack thread that calls into MxAccess (`Read`, `Write`, address-space rebuild probe sync) is wrapped in a bounded `SyncOverAsync.WaitSync(...)` helper with timeout `MxAccess.RequestTimeoutSeconds` (default 30s). This is a backstop: `MxAccessClient.Read/Write` already enforce inner `ReadTimeoutSeconds` / `WriteTimeoutSeconds` bounds on the async path. The outer wrapper exists so a scheduler stall, slow reconnect, or any other non-returning async path cannot park the stack thread indefinitely.
On timeout, the underlying task is **not** cancelled — it runs to completion on the thread pool and is abandoned. This is acceptable because MxAccess clients are shared singletons and the abandoned continuation does not capture request-scoped state. The OPC UA stack receives `StatusCodes.BadTimeout` on the affected operation.
`ConfigurationValidator` enforces `RequestTimeoutSeconds >= 1` and warns when it is set below the inner Read/Write timeouts (operator misconfiguration). Stability review 2026-04-13 Finding 3.
## Why Marshal.ReleaseComObject Is Needed
The .NET runtime's garbage collector releases COM references non-deterministically. For MXAccess, delayed release can leave stale COM connections open, preventing clean re-registration. `MxProxyAdapter.Unregister` calls `Marshal.ReleaseComObject(_lmxProxy)` in a `finally` block to immediately release the COM reference count to zero. This ensures the underlying COM server is freed before a reconnect attempt creates a new instance.