master
7 Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
 Joseph Doherty
|
1a2856526a |
mbproxy: strip historical phase/wave/plan references from source comments
Comments described the *history* of how the code arrived (phase numbers, wave IDs, review IDs, dated TODOs) instead of what it does today. That scaffolding rotted as the codebase evolved. Cleaned 60 source files + .gitignore; behaviour unchanged (387/387 tests still pass). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> |
||
|
Joseph Doherty
|
7a435957ee |
mbproxy: Wave 4 — fix issues introduced by the Wave-1/2 fixes
Closes the new findings from the post-remediation re-review (codereviews/2026-05-14/ReReviewAfterRemediation.md): NC1 — ProxyWorker.StopAsync drain loop is structurally always-zero Wave 1's W1.5 inherited the original ShutdownCoordinator bug it was meant to replace. Supervisor.StopAsync nulls the per-mux counter provider before the drain loop runs, so CountInFlight always returns 0 and the drain budget is never spent on actual draining. Fix: snapshot the in-flight count BEFORE supervisor stop, drop the theatrical post-stop loop, and report InFlightAtCancel as the snapshot count (= the number of in-flight requests dropped by the stop). The supervisor stop IS the drain — there is nothing to drain that wouldn't be killed by the stop itself. NM1 — TearDownBackendAsync._connectGate.WaitAsync uncancellable Without a token, a long Polly-wrapped EnsureBackendConnectedAsync against an unreachable host could hold the gate for the full BackendConnectTimeoutMs * MaxAttempts window, blocking DisposeAsync (and therefore ProxyWorker.StopAsync) for that duration. Fix: bound the wait with a 2 s teardown deadline; on timeout proceed best-effort without the gate. Worst-case consequence is one orphaned in-flight cycle on the dying backend, surfaced to upstream as exception 0x0B by the watchdog. NM2 — ReplaceContext non-atomic ctx + provider swap Snapshot path reads `_cacheStatsProvider` independently of `_ctx`. If `_ctx` was swapped first, a snapshot taken in the gap would still hold the OLD adapter wrapping the OLD cache — which the supervisor disposes immediately after we return. Fix: set the provider FIRST, then swap `_ctx`. Snapshots in the swap window now read either (old, old) or (new, new), never (old-after-disposed). NM5 — Self-cascade ObjectDisposedException after dispose Writer/reader fault catches fired `_ = TearDownBackendAsync(...)` unconditionally. After DisposeAsync runs `_connectGate.Dispose()`, the fire-and-forget TearDown threw ObjectDisposedException on WaitAsync as an unobserved Task exception. Fix: skip self-cascade when `_disposeCts.IsCancellationRequested` — DisposeAsync runs an explicit TearDown anyway. Nm1 — Saturation cleanup uses await SendResponseAsync W1.2's per-attacher delivery loop awaited the blocking SendResponseAsync, which would serialise on a wedged late-attacher's full bounded channel and stall delivery to its peers — contradicting the W1.3 doctrine that the fan-out path must never await per-pipe writes. Fix: use TrySendResponse and increment ResponseDropForFullUpstream on drop. T2 — WatchdogVsResponse_Race seeded Random fragility Used `new Random(12345)` over [350, 450) ms with watchdog at 400 ms; Random's algorithm is implementation-defined across .NET major versions (legacy → Xoshiro128 in .NET 6) so a runtime upgrade could land all samples on one side of the deadline and break the "both branches must fire" assertion. Fix: deterministic counter-based alternation (15 fast + 15 slow across 30 iterations) — guaranteed by construction. Latent items NM3 (_supervisorCts leak on re-Start) and NM4 (TCS single-shot semantics) are unfixed: no caller actually re-Starts a supervisor today; both become real only if the reconciler ever changes to re-Start instead of dispose-and-rebuild. Documented in the re-review. Tests: 387 pass / 0 fail. Three back-to-back race-test runs in isolation all green (T2 alternation is deterministic). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> |
||
|
Joseph Doherty
|
53f842a655 |
mbproxy: close all 5 race-hard W3 test gaps from 2026-05-14 review
Closes the 5 deterministically-race-hard test gaps that were previously
documented as known omissions (#5–9 in codereviews/2026-05-14/RemediationPlan.md).
Tests: 387 pass / 0 fail (baseline 382 + 5 new race tests). Three back-to-back
runs in isolation all green — no observable flakes.
Each test reaches the relevant code path deterministically by either:
- reaching into the multiplexer's private state via reflection (only used
for pre-saturating the TxIdAllocator — the test path that's externally
impossible to hit otherwise without spawning 65,536 real connections),
- constructing a backend stub that exercises the timing window directly, or
- asserting only the externally-observable contract that holds across all
valid interleavings (no-double-delivery, no-hang) rather than asserting
a specific ordering that flakes.
W3 #5 — TxIdAllocator_Saturated_NextRequest_GetsException04_WithOriginalTxId
Pre-saturates the multiplexer's _allocator via reflection (TryAllocate
×65536), then sends one FC06 write. The next request hits the
!_allocator.TryAllocate branch immediately and the test verifies exception
04 with the original TxId echoed.
W3 #6 — TxIdAllocator_Saturated_TwoConcurrentIdenticalReads_BothPipesGetException04
Pre-saturates the allocator, then fires two concurrent identical FC03 reads
from two pipes. Both pipes must receive exception 04 — regardless of whether
pipe B coalesces onto pipe A's stub (W1.2's deliver-to-late-attachers path)
OR opens its own factory failure path. The contract verified is "no late
attacher hangs" — the externally-observable invariant from the W1.2 fix.
W3 #7 — SlowUpstream_DoesNotStallPeerResponses_DropCounterIncrements
Wedges upstream A by leaving its socket-receive side undrained, pumps 500
FC03 requests through A so the bounded response channel + kernel buffer
fill, then sends one request from a healthy upstream B. B's response must
arrive within seconds (would block forever pre-W1.3) and A's
ResponseDropForFullUpstream counter must increment — proving the W1.3
TrySendResponse non-blocking fan-out works as designed.
W3 #8 — WatchdogVsResponse_Race_AlwaysExactlyOneOutcome_PerRequest
Custom SlowResponseBackend stub responds at a randomized 350–450 ms delay
while BackendRequestTimeoutMs=400. Across 30 iterations, the request races
the watchdog's per-tick scan. The contract asserts: every request gets
exactly ONE response (normal or exception 0x0B), the original TxId is
always echoed, and BOTH branches are exercised (proving the race window is
real). The W1 claim-then-dispatch design (CorrelationMap.TryRemove as the
single source of truth) makes this contract hold across all interleavings.
W3 #9 — CascadeVsNewAccept_StressChurn_NoCrash_NoHang
Stress-test: 3 cascade cycles, 8 concurrent connect+request tasks per
cycle. Backend is killed mid-cascade-storm to force teardown to race the
in-flight connect attempts. After all churn the multiplexer must still
serve a normal request. The originally-flagged race (a pipe added between
_pipes.Values.ToArray() and _pipes.Clear() in TearDownBackendAsync) is
microseconds wide and not deterministically reproducible without test
seams; this stress test instead proves the no-crash-under-churn property
that operators care about.
Helpers added:
DrainAllocator(PlcMultiplexer) — reflection-based saturation primitive,
only used by tests #5 and #6.
SlowResponseBackend — backend stub with caller-supplied per-request delay
via a Func<int>, only used by test #8.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
|
||
|
Joseph Doherty
|
ce32c5cee8 |
mbproxy: Wave 1 fixes from 2026-05-14 code review
Resolves the four critical correctness defects + the ShutdownCoordinator double-stop ordering bug called out in codereviews/2026-05-14/Overview.md. Tests: 362 pass / 0 fail (baseline 358 + 4 new W1 regression tests). W1.1 — Context swap on running multiplexer. PlcMultiplexer._ctx becomes volatile with a new ReplaceContext() method that re-registers the cache stats provider on the (preserved) counters. PlcListener exposes its multiplexer; PlcListenerSupervisor.ReplaceContextAsync swaps the running mux first, then disposes the old cache. Hot-reload tag-list changes and the cache-flush-on-reload contract now actually take effect on the next PDU instead of waiting for the next listener fault. W1.2 — Coalescing factory leak. When the InFlightByKey factory soft-fails (allocator saturation or duplicate TxId), the cleanup path now TryRemoves the stub and walks every party on it (including late attachers) to deliver Modbus exception 0x04. Previously only the leader got the exception; late attachers waited forever for a response that no backend round-trip would ever fire. W1.3 — Backend-reader head-of-line block. UpstreamPipe gains TrySendResponse for non-blocking enqueue. The per-PLC backend reader's fan-out loop uses it instead of awaiting SendResponseAsync, so a wedged upstream's full bounded response channel can no longer stall the single backend reader and starve every other client on that PLC. New responseDropForFullUpstream counter on ProxyCounters / CounterSnapshot records the drops. W1.4 — Stranded outbound frames after cascade. TearDownBackendAsync acquires _connectGate and drains any frames left in _outboundChannel after the writer task faulted/cancelled, releasing their proxy TxIds back to the allocator. Without this, a fresh EnsureBackendConnectedAsync racing the cascade would send stranded frames with old TxIds onto the new backend socket; the responses would arrive with no correlation entry and the upstream peers would hang on the watchdog until BackendRequestTimeoutMs. W1.5 — Delete ShutdownCoordinator (Option B). Drain logic moved into ProxyWorker.StopAsync. AdminEndpointHost is no longer registered as IHostedService; ProxyWorker drives its lifecycle directly so admin starts after listeners are bound and stops AFTER the in-flight drain (the design's documented contract). Admin is resolved lazily in ExecuteAsync to break the circular DI graph (Admin -> StatusSnapshotBuilder -> ProxyWorker). GracefulShutdownTimeoutMs is now read fresh from IOptionsMonitor.CurrentValue at stop time, so a hot-reloaded value is honoured. Removes ShutdownCoordinator + tests. New tests: PlcMultiplexerTests.ReplaceContext_NewTagMap_VisibleOnNextPdu PlcMultiplexerTests.ReplaceContext_NewCache_NextReadGoesToBackend_NotOldCache UpstreamPipeTests.TrySendResponse_WhenChannelFull_ReturnsFalse_WithoutBlocking UpstreamPipeTests.TrySendResponse_AfterDispose_ReturnsFalse Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> |
||
|
Joseph Doherty
|
1db900edef |
mbproxy: add opt-in response cache (Phase 11)
Layers a per-PLC, per-tag response cache on top of Phase 10's coalescing.
Cache is OFF by default per tag (CacheTtlMs = 0); a fresh deployment with no
TTL config behaves identically to Phase 10. Operators opt tags in by setting
CacheTtlMs > 0 on a BcdTagOptions entry (or DefaultCacheTtlMs > 0 on a
PlcOptions entry), explicitly acknowledging the staleness window.
Cache lookup order: cache -> coalesce -> backend. A cache hit short-circuits
both Phase 10's coalescing path and Phase 9's backend send. Cache stores
POST-rewriter PDU bytes so hits never re-invoke the BCD rewriter. FC06/FC16
write responses invalidate every cached entry whose address range overlaps
the write (half-open interval math).
New types (Mbproxy.Proxy.Cache, all internal):
- CacheKey (record-struct, same shape as CoalescingKey but kept SEPARATE so
the two phases evolve independently).
- CacheEntry, ResponseCache (IDisposable; LRU + PeriodicTimer eviction
loop), CacheInvalidator (pure overlap matcher), CacheLogEvents (stable
mbproxy.cache.* names).
Multi-tag range TTL = min(TTLs); any tag with TTL = 0 in the range disables
caching for the whole read (conservative-by-design).
Options surface:
- BcdTagOptions.CacheTtlMs (nullable int; null = fall through to PLC default)
- PlcOptions.DefaultCacheTtlMs
- MbproxyOptions.Cache.{AllowLongTtl, MaxEntriesPerPlc, EvictionIntervalMs}
- TTL > 60_000 ms requires Cache.AllowLongTtl = true (reload validation).
Admin counters (Tier 1.8 + Tier 2 cache-memory KPIs from docs/kpi.md):
- CacheHitCount, CacheMissCount, CacheInvalidations on ProxyCounters.
- CacheEntryCount, CacheBytes via a new ICacheStatsProvider snapshot path.
- /status.json and the HTML page surface a new Cache cell per PLC row.
Hot-reload: any tag-list change to a PLC reseats the per-PLC context with a
fresh cache; the old cache is disposed inside ReplaceContextAsync. Per-tag
flush granularity is intentionally not implemented in v1.
PLCs with no cache-eligible tags (every resolved tag has CacheTtlMs = 0)
get Cache = null on the context and skip the eviction timer entirely, so
the no-cache path is byte-identical to Phase 10.
Tests (32 new unit + 5 new E2E = 37 new; suite now 314 unit + 48 E2E):
- CacheKeyTests, CacheEntryTests (records + boundary semantics).
- CacheInvalidatorTests: full overlap, both partials, adjacent-not-
overlapping, disjoint, different unit ID + auxiliary FC-filter / zero-qty.
- ResponseCacheTests: round-trip, lazy expiry, range invalidation,
unit-id filter, LRU bound, LRU access tracking, concurrent get/set,
dispose, clear, approximate-bytes accounting.
- ResponseCacheMultiplexerTests (stub-backend): hit short-circuits
coalescing, BCD-decoded bytes are cached not raw, FC06 invalidates
overlapping, non-overlapping write does not invalidate, multi-tag
TTL=min rule, regression-cache-disabled-by-default-is-Phase-10, hit
works even when backend unreachable.
- ResponseCacheE2ETests (pymodbus DL205 sim, sequential reads):
* Headline: 10 reads with TTL=1000 ms -> 9 hits, 1 miss, 1 backend trip.
* TTL expiry path with sleep > TTL.
* Write invalidation through the proxy on a scratch register.
* BCD-decoded bytes are cached, not raw BCD nibbles.
* Regression: Cache disabled by default -> behaviour byte-identical to
Phase 10.
Pre-existing flake hardened: BackendDisconnect_CascadesToAllUpstreams now
polls briefly for the cascade counter to absorb the inherent scheduling
gap between "upstream EOF observed" and "counter incremented inside
TearDownBackendAsync." Counter semantics unchanged.
Phase doc updated with implementation clarifications discovered during
this work (CacheKey kept separate from CoalescingKey, LastUsedTick is
long, FC06/FC16 startAddr/qty parsing extension, cache-pre-connect
short-circuit, write-invalidation only on successful responses).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
|
||
|
Joseph Doherty
|
a2dba4bd07 |
mbproxy: add in-flight read coalescing (Phase 10)
When two or more upstream clients send the same FC03/FC04 read while a matching request is already in flight on the same PLC's multiplexed backend socket, attach the late arrivals to the existing InFlightRequest .InterestedParties list instead of opening a second backend round-trip. The single backend response fans out to every attached party with each party's original MBAP TxId restored individually. Zero post-response staleness — coalescing operates entirely within the in-flight window (microseconds to ~10 ms typical); the proxy is NOT a cache layer. Headline mechanism: - New record struct CoalescingKey(UnitId, Fc, StartAddress, Qty) keys the per-PLC InFlightByKeyMap. FC03 and FC04 are separate Modbus tables and never share a key; different unit IDs never coalesce; writes (FC06/FC16) bypass the coalescing path entirely. - InFlightByKeyMap uses a simple lock around a Dictionary; atomic TryAttachOrCreate either appends a new party to the in-flight request's mutable List<InterestedParty> or invokes a factory to build a fresh entry. Per-entry MaxParties cap (default 32) bounds fan-out cost; past the cap, the next arrival opens a new entry. - PlcMultiplexer.OnUpstreamFrameAsync takes the coalescing path for FC03/FC04 when Mbproxy.Resilience.ReadCoalescing.Enabled. The factory closure does the Phase-9 work (allocate TxId, add to CorrelationMap); the channel send happens AFTER returning from TryAttachOrCreate so the map lock is not held across the async send. - Response fan-out in RunBackendReaderAsync removes the entry from InFlightByKeyMap before iterating InterestedParties, ensuring no concurrent attach can mutate the list during iteration. - Cascade + watchdog paths also drain the key map so a stale entry cannot outlive its backend round-trip. Counter accounting balance (per snapshot): CoalescedHitCount + CoalescedMissCount equals total FC03 + FC04 requests since startup. Even with coalescing disabled, every read still bumps Miss so dashboard math stays balanced. New surface (additive only): - src/Mbproxy/Proxy/Multiplexing/CoalescingKey.cs - src/Mbproxy/Proxy/Multiplexing/InFlightByKeyMap.cs - src/Mbproxy/Proxy/Multiplexing/CoalescingLogEvents.cs - ReadCoalescingOptions on ResilienceOptions - CoalescedHitCount / CoalescedMissCount / CoalescedResponseToDeadUpstream counters surfaced on /status.json per PLC and as a compact "Coal" cell on the HTML status page. Phase 9 test patch: TwoUpstreams_ProxyTxIds_AreDistinct_OnTheWire previously read the same register from both clients (which now coalesces). Patched to read two different addresses so the test still proves distinct backend TxIds without violating the coalescing contract. Tests added: 24 new (19 unit + 5 E2E): - CoalescingKeyTests (5) - InFlightByKeyMapTests (6, includes concurrent stress) - ReadCoalescingTests (8, stub-backend with deterministic delay) - ReadCoalescingE2ETests (5, pymodbus simulator; coalescing-active during overlap is proven against the stub, not the sim, due to pymodbus 3.13's known concurrent-frame bug) Total: 325 tests passing (282 unit + 43 E2E). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> |
||
|
Joseph Doherty
|
56eee3c563 |
mbproxy: initial commit through Phase 9 (TxId multiplexing)
Adds the mbproxy service end-to-end. Phases 00-08 implement the production-ready single-listener / 1:1-backend transparent Modbus TCP proxy with bidirectional BCD rewriting for the ~54-PLC DL205/DL260 fleet. Phase 9 replaces the connection layer with a single backend socket per PLC plus MBAP TxId rewriting, lifting the H2-ECOM100's 4-concurrent-client cap as an operational ceiling. Phase 9 additions of note: - PlcMultiplexer + UpstreamPipe + TxIdAllocator + CorrelationMap - InFlightRequest with IReadOnlyList<InterestedParty> (load-bearing for Phase 10 read coalescing — do not collapse to a single field) - Per-request watchdog: surfaces Modbus exception 0x0B to upstream on BackendRequestTimeoutMs, defending against lost responses, dead-PLC paths, and pymodbus 3.13.0's concurrent-multiplexed- request bug (its ServerRequestHandler.last_pdu state race) - Status DTO + HTML gain inFlight / maxInFlight / txIdWraps / disconnectCascades / queueDepth (Tier 1.6 in docs/kpi.md) Tests: 263 unit + 38 E2E. Multiplexer correctness under truly concurrent backend traffic is proved against a stub backend in PlcMultiplexerTests; MultiplexerE2ETests paces requests so pymodbus 3.13's single-PDU framer stays in known-good mode. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> |