Add 5 validated Process* methods to JetStreamMetaGroup for stream and
consumer assignment processing: ProcessStreamAssignment, ProcessUpdateStreamAssignment,
ProcessStreamRemoval, ProcessConsumerAssignment, and ProcessConsumerRemoval.
Each returns a bool indicating success, with validation guards matching
Go reference jetstream_cluster.go:4541-5925. Includes 12 new unit tests.
Implements the background loop (Go: monitorCluster) that reads RaftLogEntry
items from a channel and dispatches assignStream, removeStream, assignConsumer,
removeConsumer, and snapshot operations to JetStreamMetaGroup.
- Add five public mutation helpers to JetStreamMetaGroup (AddStreamAssignment,
RemoveStreamAssignment, AddConsumerAssignment, RemoveConsumerAssignment,
ReplaceAllAssignments) used by the monitor path
- JetStreamClusterMonitor: async loop over ChannelReader<RaftLogEntry>, JSON
dispatch, ILogger injection with NullLogger default, malformed entries logged
and skipped rather than aborting the loop
- WaitForProcessedAsync uses Monitor.PulseAll for race-free test synchronisation
with no Task.Delay — satisfies slopwatch SW003/SW004 rules
- 10 new targeted tests all pass; 1101 cluster regression tests unchanged
Implements binary codec for meta-group snapshots: 2-byte little-endian version
header followed by S2-compressed JSON of the stream assignment map. Adds
[JsonObjectCreationHandling(Populate)] to StreamAssignment.Consumers so the
getter-only dictionary is populated in-place during deserialization. 8 tests
covering round-trip, compression ratio, field fidelity, multi-consumer restore,
version rejection, and truncation guard.
Go reference: jetstream_cluster.go:2075-2145 (encodeMetaSnapshot/decodeMetaSnapshot)
Adds BeginJointConsensus, CommitJointConsensus, and CalculateJointQuorum to
RaftNode per Raft paper Section 4. During a joint configuration transition
quorum requires majority from BOTH Cold and Cnew; CommitJointConsensus
finalizes Cnew as the sole active configuration. The existing single-phase
ProposeAddPeerAsync/ProposeRemovePeerAsync are unchanged. Includes 16 new
tests covering flag behaviour, quorum boundaries, idempotent commit, and
backward-compatibility with the existing membership API.
- SyncAsync: remove redundant FlushAsync, use single Flush(flushToDisk:true)
- ComputeCrc: use incremental Crc32.Append to avoid contiguous buffer heap allocation
- Load: cast pos+length to long to guard against int overflow in bounds check
- AppendAsync: delegate to WriteEntryTo (DRY — eliminates duplicated record-building logic)
- Load: extract ParseEntries static helper to eliminate goto pattern with early returns
- Entries: change return type from IEnumerable to IReadOnlyList for index access and Count property
- RaftNode.PersistAsync: remove redundant term.txt write (meta.json now owns term+votedFor)
- RaftWalTests: tighten ShouldBeGreaterThanOrEqualTo(1) -> ShouldBe(1) in truncation/CRC tests;
use .Count property directly on IReadOnlyList instead of .Count() LINQ extension
Implements a binary write-ahead log (RaftWal) for durable RAFT entry
storage, replacing in-memory-only semantics. The WAL uses a magic header
("NWAL" + version), length-prefixed records with per-record CRC32
integrity checking, and CompactAsync with atomic temp-file rename.
Load() tolerates truncated or corrupt tail records for crash safety.
Also fixes RaftNode to persist and reload TermState.VotedFor via a
meta.json file alongside term.txt, ensuring vote durability across
restarts. Falls back gracefully to legacy term.txt when meta.json is
absent.
6 new tests in RaftWalTests: persist/recover, compact, truncation
tolerance, VotedFor round-trip, empty WAL, and CRC corruption.
All 458 Raft tests pass.
Complete IStreamStore Batch 2 — all remaining interface methods now have
FileStore implementations:
- EncodedStreamState: placeholder returning empty (full codec in Task 9)
- UpdateConfig: no-op placeholder for config hot-reload
- ResetState: no-op (state derived from blocks on construction)
- Delete(bool): now calls Stop() first and matches interface signature
Complete IStreamStore Batch 1 — all core operations now have FileStore
implementations instead of throwing NotSupportedException:
- StoreRawMsg: caller-specified seq/ts for replication/mirroring
- LoadPrevMsg: backward scan for message before given sequence
- Type: returns StorageType.File
- Stop: flush + dispose blocks, reject further writes
14 new tests in FileStoreStreamStoreTests.
Add FlushAllPending() to FileStore — flushes active MsgBlock to disk and
writes a stream.state checkpoint atomically (write-to-temp + rename).
Go ref: filestore.go:5783 (flushPendingWritesUnlocked / writeFullState).
5 new tests in FileStoreCrashRecoveryTests: flush, state file, idempotent,
TTL recovery, truncated block handling.
Add FlushAllPending() to FileStore, fulfilling the IStreamStore interface
contract. The method flushes the active MsgBlock to disk and atomically
writes a stream.state checkpoint using write-to-temp + rename, matching
Go's flushPendingWritesUnlocked / writeFullState pattern.
Add FileStoreCrashRecoveryTests with 5 tests covering:
- FlushAllPending flushes block data to .blk file
- FlushAllPending writes a valid atomic stream.state JSON checkpoint
- FlushAllPending is idempotent (second call overwrites with latest state)
- Recovery prunes messages backdated past the MaxAgeMs cutoff
- Recovery handles a tail-truncated block without throwing
Reference: golang/nats-server/server/filestore.go:5783-5842
Add FileStoreEncryptionTests covering ChaCha20-Poly1305 and AES-GCM
round-trips and wrong-key rejection for the FSV2 AEAD path. Fix
RestorePayload to wrap CryptographicException from AEAD decryption as
InvalidDataException so RecoverBlocks correctly propagates key-mismatch
failures instead of silently swallowing them.
StreamManager.Capture now accounts for full message size (subject +
payload + 16-byte overhead) when checking MaxBytes, matching Go's
memStoreMsgSize. PullConsumerEngine uses stream FirstSeq instead of
hardcoded 1 for DeliverAll after purge. Fix 6 tests with Go parity
assertions and updated MaxBytes values.
DateTime.UtcNow.Ticks * 100L overflows long (Ticks ~6.38e17, x100 =
6.38e19 > long.MaxValue 9.22e18), producing corrupted timestamps that
cause PruneExpiredMessages to immediately expire all messages when
MaxAgeMs is set. Use Unix epoch offset to match Go's time.Now().UnixNano().
Port Go jetstream_test.go storage/recovery/encryption tests. Add stream
recovery stubs, encryption key management, direct-get with time queries,
and subject delete marker handling.
23 new tests ported from jetstream_test.go.
Port Go store_test.go contract tests for IStreamStore interface using
MemStore. Fix CompactInternal to correctly walk forward to find new
FirstSeq before backward cleanup (matching Go behavior).
21 new tests ported from store_test.go.
Port Go filestore tombstone/deletion tests, consumer state encode/decode,
consumer file store persistence, and message TTL enforcement. Adds
ConsumerStateCodec and ConsumerFileStore implementations.
17 new tests ported from filestore_test.go.
Ports 34 Go FileStore tests from filestore_test.go to
FileStoreRecovery2Tests.cs (31 pass, 4 skipped). Tests cover block
recovery, compaction, PSIM indexing, skip-msg handling, TTL expiry,
corrupt index/state detection, and read-only permission checks.
Updates docs/test_parity.db with mapped/skipped status for all 34 tests.
Re-enabled the leader check in JetStreamApiRouter.Route() that was
commented out during B8 agent work. Added BecomeLeader() method to
JetStreamMetaGroup for single-process test fixtures to simulate
winning the post-stepdown election. MetaControllerFixture now
auto-calls BecomeLeader() after a successful meta leader stepdown.
- RaftGroup, StreamAssignment, ConsumerAssignment types matching Go structs
(jetstream_cluster.go:154-266)
- PlacementEngine.SelectPeerGroup: topology-aware peer selection with cluster
affinity, tag filtering, exclude tags, and storage-weighted sorting
(Go ref: selectPeerGroup at line 7212)
- JetStreamMetaGroup: backward-compatible rewrite with full assignment tracking,
consumer proposal workflow, and delete operations
- 41 new tests in ClusterAssignmentAndPlacementTests
- ProposeAddPeerAsync/ProposeRemovePeerAsync: single-change-at-a-time membership
changes through RAFT consensus (Go ref: raft.go:961-1019)
- RaftLog.Compact: removes entries up to given index for log compaction
- CreateSnapshotCheckpointAsync: creates snapshot and compacts log in one operation
- DrainAndReplaySnapshotAsync: drains commit queue, installs snapshot, resets indices
- Pre-vote protocol skipped (Go NATS doesn't implement it either)
- 23 new tests in RaftMembershipAndSnapshotTests
- CommitQueue<T>: channel-based queue for committed entries awaiting state machine application
- RaftPeerState: tracks replication and health state (nextIndex, matchIndex, lastContact)
- RaftNode: CommitIndex/ProcessedIndex tracking, election timer with randomized 150-300ms interval,
peer state integration with heartbeat and replication updates
- 52 new tests across RaftApplyQueueTests, RaftElectionTimerTests, RaftHealthTests
D6: Add ExportSubjects/ImportSubjects allow-lists to LeafHubSpokeMapper alongside
existing DenyExports/DenyImports deny-lists. When an allow-list is non-empty, subjects
must match at least one allow pattern; deny always takes precedence. Updated
LeafNodeOptions, LeafHubSpokeMapper (5-arg constructor), and LeafNodeManager to wire
through the new allow-lists. Added 13 new unit + integration tests covering allow-list
semantics, deny precedence, bidirectional filtering, and wire-level propagation.
D7: Existing NetworkingGoParityTests.cs (50 tests) covers gateway interest mode,
route pool accounting, and leaf node connections. Parity DB already up to date.
E6: Add IsSystemAccount property to Account, mark $SYS account as system,
add IsSystemSubject/IsSubscriptionAllowed/GetSubListForSubject helpers to
route $SYS.> subjects to the system account's SubList and block non-system
accounts from subscribing.
E7: Add ConfigReloader.ReloadAsync and ApplyDiff for structured async reload,
add ConfigReloadResult/ConfigApplyResult types. SIGHUP handler already wired
via PosixSignalRegistration in HandleSignals.
E8: Add PropagateAuthChanges to re-evaluate connected clients after auth
config reload, disconnecting clients whose credentials no longer pass
authentication with -ERR 'Authorization Violation'.
E4: AccountImportExport with DFS cycle detection for service imports,
RemoveServiceImport/RemoveStreamImport, and ValidateImport authorization.
E5: AccountLimits record with MaxStorage/MaxConsumers/MaxAckPending,
TryReserveConsumer/ReleaseConsumer, TrackStorageDelta on Account.
20 new tests, all passing.
Add MqttSessionStore with save/load/delete/list operations, flapper
detection (backoff on rapid reconnects), and TimeProvider-based testing.
Add MqttRetainedStore for per-topic retained messages with MQTT wildcard
matching (+/# filters). Add MqttQos2StateMachine tracking the full
PUBREC/PUBREL/PUBCOMP flow with duplicate rejection and timeout detection.
19 new tests: 9 session persistence, 10 QoS/retained message tests.
D2: Add FNV-1a-based ComputeRoutePoolIdx to RouteManager matching Go's
route.go:533-545, with PoolIndex on RouteConnection and account-aware
ForwardRoutedMessageAsync that routes to the correct pool connection.
D3: Replace DeflateStream with IronSnappy in RouteCompressionCodec, add
RouteCompressionLevel enum, NegotiateCompression, and IsCompressed
detection. 17 new tests (6 pool + 11 compression), all passing.
Ports the gateway interest-only mode from Go (gateway.go:100-150, 1500-1600):
- Add GatewayInterestTracker with Optimistic/Transitioning/InterestOnly modes
- In Optimistic mode, track no-interest set; switch to InterestOnly when set
exceeds threshold (default 1000, matching Go defaultGatewayMaxRUnsubThreshold)
- In InterestOnly mode, only forward subjects with tracked RS+ interest;
use SubjectMatch.MatchLiteral for wildcard pattern support
- Integrate tracker into GatewayConnection: A+/A- messages update tracker,
SendMessageAsync skips send when ShouldForward returns false
- Expose InterestTracker property on GatewayConnection for observability
- Add 13 unit tests covering all 8 specified behaviors plus edge cases
Add MsgBlock write cache (mirrors Go's msgBlock.cache) to serve reads
for recently-written records without disk I/O; cleared on block seal via
RotateBlock. Add HashWheel-based TTL expiry in FileStore (ExpireFromWheel /
RegisterTtl), replacing the O(n) linear scan on every append with an
O(expired) wheel scan. Implement StoreMsg sync method with per-message TTL
override support. Add 10 tests covering cache hits/eviction, wheel expiry,
retention, StoreMsg seq/ts, per-msg TTL, and recovery re-registration.
Implement PurgeEx, Compact, Truncate, FilteredState, SubjectsState,
SubjectsTotals, State, FastState, GetSeqFromTime on FileStore. Add
MsgBlock.IsDeleted, DeletedSequences, EnumerateNonDeleted. Includes
wildcard subject support via SubjectMatch for all filtered operations.
Replace JSONL persistence with real MsgBlock-based block files (.blk).
FileStore now acts as a block manager that creates, seals, and rotates
MsgBlocks while maintaining an in-memory cache for fast reads/queries.
Key changes:
- AppendAsync writes transformed payloads to MsgBlock via WriteAt
- Block rotation occurs when active block reaches size limit
- Recovery scans .blk files and rebuilds in-memory state from records
- Legacy JSONL migration: existing messages.jsonl data is automatically
converted to block files on first open, then JSONL is deleted
- PurgeAsync disposes and deletes all block files
- RewriteBlocks rebuilds blocks from cache (used by trim/restore)
- InvalidDataException propagates during recovery (wrong encryption key)
MsgBlock.WriteAt added to support explicit sequence numbers and timestamps,
needed when rewriting blocks with non-contiguous sequences (after removes).
Tests updated:
- New FileStoreBlockTests.cs with 9 tests for block-specific behavior
- JetStreamFileStoreCompressionEncryptionParityTests updated to read
FSV1 magic from .blk files instead of messages.jsonl
- JetStreamFileStoreDurabilityParityTests updated to verify .blk files
instead of index.manifest.json
All 3,562 tests pass (3,535 passed + 27 skipped, 0 failures).
MsgBlock is the unit of storage in the file store — a single append-only
block file containing sequentially written binary message records. Blocks
are sealed (read-only) when they reach a configurable byte-size limit.
Key features:
- Write: appends MessageRecord-encoded messages with auto-incrementing
sequence numbers and configurable first sequence offset
- Read: positional I/O via RandomAccess.Read for concurrent reader safety
- Delete: soft-delete with on-disk persistence (re-encodes flags byte +
checksum so deletions survive recovery)
- Recovery: rebuilds in-memory index by scanning block file using
MessageRecord.MeasureRecord for record boundary detection
- Thread safety: ReaderWriterLockSlim allows concurrent reads during writes
Also adds MessageRecord.MeasureRecord() — computes a record's byte length
by parsing varint field headers without full decode, needed for sequential
record scanning during block recovery.
Reference: golang/nats-server/server/filestore.go:217-267 (msgBlock struct)
12 tests covering write, read, delete, seal, recovery, concurrency,
and custom sequence offsets.
Add MessageRecord class with Encode/Decode for the binary wire format
used in JetStream file store blocks. Uses varint-encoded lengths,
XxHash64 checksums, and a flags byte for deletion markers.
Go reference: filestore.go:6720-6724, 8180-8250, 8770-8777
13 tests covering round-trip, headers, checksum validation, corruption
detection, varint encoding, deleted flag, empty/large payloads.
