Files
Joseph Doherty
793c787315
Phase 3 PR 53 -- Transport reconnect-on-drop + SO_KEEPALIVE for DL205 no-keepalive quirk. AutomationDirect H2-ECOM100 does NOT send TCP keepalives per docs/v2/dl205.md behavioral-oddities section -- any NAT/firewall device between the gateway and the PLC can silently close an idle socket after 2-5 minutes of inactivity. The PLC itself never notices and the first SendAsync after the drop would previously surface as IOException / EndOfStreamException / SocketException to the caller even though the PLC is perfectly healthy. PR 53 makes ModbusTcpTransport survive mid-session socket drops: SendAsync wraps the previous body as SendOnceAsync; on the first attempt, if the failure is a socket-layer error (IOException, SocketException, EndOfStreamException, ObjectDisposedException) AND autoReconnect is enabled (default true), the transport tears down the dead socket, calls ConnectAsync to re-establish, and resends the PDU exactly once. Deliberately single-retry -- further failures propagate so the driver health surface reflects the real state, no masking a dead PLC. Protocol-layer failures (e.g. ModbusException with exception code 02) are specifically NOT caught by the reconnect path -- they would just come back with the same exception code after the reconnect, so retrying is wasted wire time. Socket-level vs protocol-level is a discriminator inside IsSocketLevelFailure. Also enables SO_KEEPALIVE on the TcpClient with aggressive timing: TcpKeepAliveTime=30s, TcpKeepAliveInterval=10s, TcpKeepAliveRetryCount=3. Total time-to-detect-dead-socket = 30 + 10*3 = 60s, vs the Windows default 2-hour idle + 9 retries = 2h40min. Best-effort: older OSes that don't expose the fine-grained keepalive knobs silently skip them (catch {}). New ModbusDriverOptions.AutoReconnect bool (default true) threads through to the default transport factory in ModbusDriver -- callers wanting the old 'fail loud on drop' behavior can set AutoReconnect=false, or use a custom transportFactory that ignores the option. Unit tests: ModbusTcpReconnectTests boots a FlakeyModbusServer in-process (real TcpListener on loopback) that serves one valid FC03 response then forcibly shuts down the socket. Transport_recovers_from_mid_session_drop_and_retries_successfully issues two consecutive SendAsync calls and asserts both return valid PDUs -- the second must trigger the reconnect path transparently. Transport_without_AutoReconnect_propagates_drop_to_caller asserts the legacy behavior when the opt-out is taken. Validates real socket semantics rather than mocked exceptions. 142/142 Modbus.Tests pass (113 prior + 2 mapper + 2 reconnect + 25 accumulated across PRs 45-52); 11/11 DL205 integration tests still pass with MODBUS_SIM_PROFILE=dl205 -- no regression from the transport change.
Phase 3 PR 34 — Host-status publisher (Server) + /hosts drill-down page (Admin). Closes LMX follow-up #7 by wiring together the data layer from PR 33. Server.HostStatusPublisher is a BackgroundService that walks every driver registered in DriverHost every 10 seconds, skips drivers that don't implement IHostConnectivityProbe, calls GetHostStatuses() on each probe-capable driver, and upserts one DriverHostStatus row per (NodeId, DriverInstanceId, HostName) into the central config DB. Upsert path: SingleOrDefaultAsync on the composite PK; if no row exists, Add a new one; if a row exists, LastSeenUtc advances unconditionally (heartbeat) and State + StateChangedUtc update only on transitions so Admin UI can distinguish 'still reporting, still Running' from 'freshly transitioned to Running'. MapState translates Core.Abstractions.HostState to Configuration.Enums.DriverHostState (intentional duplicate enum — Configuration project stays free of driver-runtime deps per PR 33's choice). If a driver's GetHostStatuses throws, log warning and skip that driver this tick — never take down the Server on a publisher failure. If the DB is unreachable, log warning + retry next heartbeat (no buffering — next tick's current-state snapshot is more useful than replaying stale transitions after a long outage). 2-second startup delay so NodeBootstrap's RegisterAsync calls land before the first publish tick, then tick runs immediately so a freshly-started Server surfaces its host topology in the Admin UI without waiting a full interval.
Phase 3 PR 33 — DriverHostStatus entity + EF migration (data-layer for LMX #7). New DriverHostStatus entity with composite key (NodeId, DriverInstanceId, HostName) persists each server node's per-host connectivity view — one row per (server node, driver instance, probe-reported host), which means a redundant 2-node cluster with one Galaxy driver reporting 3 platforms produces 6 rows because each server node owns its own runtime view of the shared host topology, not 3. Fields: NodeId (64), DriverInstanceId (64), HostName (256 — fits Galaxy FQDNs and Modbus host:port strings), State (DriverHostState enum — Unknown/Running/Stopped/Faulted, persisted as nvarchar(16) via HasConversion<string> so DBAs inspecting the table see readable state names not ordinals), StateChangedUtc + LastSeenUtc (datetime2(3) — StateChangedUtc tracks actual transitions while LastSeenUtc advances on every publisher heartbeat so the Admin UI can flag stale rows from a crashed Server independent of State), Detail (nullable 1024 — exception message from the driver's probe when Faulted, null otherwise).
Phase 3 PR 38 — DriverNodeManager HistoryRead override (LMX #1 finish). Wires the OPC UA HistoryRead service through CustomNodeManager2's four protected per-kind hooks — HistoryReadRawModified / HistoryReadProcessed / HistoryReadAtTime / HistoryReadEvents — each dispatching to the driver's IHistoryProvider capability (PR 35 for ReadAtTime + ReadEvents on top of PR 19-era ReadRaw + ReadProcessed). Was the last missing piece of the end-to-end HistoryRead path: PR 10 + PR 11 shipped the Galaxy.Host IPC contracts, PR 35 surfaced them on IHistoryProvider + GalaxyProxyDriver, but no server-side handler bridged OPC UA HistoryRead service requests onto the capability interface. Now it does.