Bump WolverineFx and WolverineFx.Http

[dependabot]

Updated WolverineFx from 6.17.2 to 6.23.1.

Release notes

Sourced from WolverineFx's releases.

6.23.1

Agent distribution

TL;DR: if you pause a projection from CritterWatch on 6.23.0, restarting it appears to do nothing for a full minute. This fixes that.

• A paused projection or subscription agent now resumes immediately when you restart it (#​3663). On 6.23.0 the restart was accepted, the pause restriction was cleared, and then nothing happened until the pending-assignment ledger's TTL expired — 2 × CheckAssignmentPeriod, so 60 seconds with the defaults. Long enough that an operator reasonably concludes the agent is never coming back.

  The ledger introduced in 6.23.0 (#​3622) only counted an assignment as confirmed if a later evaluation saw the agent running and still assigned to the same node. A pause makes those two conditions mutually exclusive: the first evaluation that can observe the delivered assignment is the same one that detaches the agent. The entry was never confirmed, nothing on the stop path cleared it, and the restart's AssignAgent was suppressed as a duplicate still in flight. Delivery alone now confirms the entry, which is the only question the ledger was ever asking.

• Pausing an agent no longer briefly starts it first (#​3666). ApplyRestrictionsAsync kickstarted a health check before persisting the operator's restriction change, so that evaluation ran against the old restrictions and could act against the very intent being applied — for a pause, re-assigning and starting the agent one beat before the merged evaluation stopped it again. Besides the wasted daemon start/stop cycle, this is what armed the stale ledger entry behind #​3663.

PostgreSQL

• Advisory-lock sessions stay invisible to Marten's async-daemon gap detection (#​3664). Marten 9.16.1+ will not skip a stale event-sequence gap while any session whose open transaction predates that gap is still alive (marten#​4953). A session parked in an open transaction for the life of the process therefore reads as a permanent "possible reserver" and can hold the high-water mark — and every async projection — behind a gap that is genuinely dead.

  Wolverine's long-held locks were already shaped correctly: leader election and node coordination hold session-scoped advisory locks on a dedicated connection with no transaction, so they show up as state='idle' with a NULL xact_start. Those sessions are now also tagged application_name = 'wolverine-advisory-lock:<database>', which turns a pg_stat_activity investigation from guesswork into something you can read at a glance. The constraints are pinned in tests and in the Postgres durability docs, including the trap worth knowing in your own code: never add a keepalive query inside a long-lived open transaction — it bumps state_change, makes the session look active, and re-promotes it to candidate reserver.

  Note for combined Marten + Wolverine deployments: older guidance suggested Postgres's idle_in_transaction_session_timeout as a dead-gap backstop. Prefer upgrading Marten and using SkipStaleGapsDespiteLiveTransactionsAfter instead.

6.23.0

Projection & agent distribution

TL;DR: this prevents Wolverine from going into a panic doing agent assignments and churning crazily hard during Kubernetes rollouts or cluster starts and that's a very good thing

The bulk of this release. A wave of fixes (WO-1..8) to the agent assignment plane that together remove the re-assignment churn and livelock that could leave projection agents flapping or wedged.

• Node heartbeats no longer block behind command execution (#​3620)
• Node resurrection restores the node's real identity rather than a skeleton record (#​3621)
• A pending-assignment ledger suppresses duplicate AssignAgent floods (#​3622)
• Agent batch starts are chunked and bounded-parallel, so they stay safe at scale (#​3623)
• Ejection hysteresis plus leader protection (#​3624)
• The assignment row is re-upserted for agents that are already running (#​3619)
• Local agents drain with bounded parallelism on shutdown (#​3636)
• Paused shards now say why. IEventSubscriptionAgent.Failure surfaces a ShardFailure — category, the failing event's sequence and type, and the root exception type — through health checks and a new IWolverineObserver.AgentPaused hook plus a NodeRecordType.AgentPaused record. Failures bound to a specific event (ApplyEvent, EventSerialization, UnknownEventType) or to two processes racing one shard (ProgressionOutOfOrder) are no longer auto-restarted, since they would die on the identical event every time (#​3637, #​3638)
• Agent starts retry locally, bounded by Durability.AgentStartRetryAttempts / AgentStartRetryDelay, instead of idling a full CheckAssignmentPeriod after a startup race (#​3519)
• Store-scoped FindAgentUriAsync overload (#​3647) and a store-aware TryRebuildRegisteredProjectionAsync overload (#​3618)

Bumps JasperFx.Events to 2.36.1, Marten to 9.20.0 and Polecat to 5.7.0.

HTTP

• Endpoints that advertise a request body their HTTP method cannot carry now fail fast at startup instead of silently 404ing out of the route matcher (#​3648)
• A missing Content-Type on an [AcceptsContentType] route returns 415 rather than 404 (#​3649)
• Query-only [AsParameters] endpoints no longer advertise a form body, which had been dropping them from route matching entirely (#​3630)
• Explicitly-routed chains — PublishMessage<T> and SendMessage<T> — now describe the message they read from the body. They had been advertising no request body at all in OpenAPI (#​3646)
• Enum values in query strings and form posts parse case-insensitively (#​3629)
• Compiled queries returned as a Wolverine.Http resource no longer execute twice (#​3625)

Transports

• Conventional routing respects named brokers (#​3633)
• Google Pub/Sub: named-broker support for sharded and partitioned topics (#​3632), and ListenToPubsubSubscriptionOnNamedBroker (#​3631)
• Kafka: topic-already-exists error codes are handled rather than thrown (#​3640)
• Redis: scheduled retries survive an unreadable timestamp (#​3613), and scheduled entries that repeatedly fail to deserialize are dead-lettered instead of looping (#​3644)
• EnvelopeMapper reads both timestamp header formats (#​3645)
• NServiceBus interop: the EnclosedMessageTypes header is split before the message type is resolved, so interop works across Azure Service Bus, SNS, SQS and the database transports (#​3628)

Persistence

• Oracle runs the durability agent through the shared batching mechanics (#​3614). Note that ODP.NET exposes no DbBatch, so the work splits per statement.

Other

• Resource discovery no longer eager-connects broker transports (#​3617)

Contributors

Thank you to everyone who contributed to this release:

... (truncated)

6.22.0

Wolverine 6.22.0 rolls up the claim-check backend wave, distributed-agent and durability hardening, HTTP/OpenAPI binding fixes, and the Marten 9.18 / JasperFx 2.34 critter-stack alignment.

Dependency alignment

• JasperFx family → 2.34.0
• Marten → 9.18.0 (with Marten.AspNetCore / Marten.Newtonsoft), Polecat [5.5.0,6.0.0)
• Weasel family → 9.18.1

Claim-check offloading

• New NATS JetStream Object Store claim-check backend (#​3506)
• New PostgreSQL database-LOB claim-check backend (#​3507)
• New Google Cloud Storage claim-check backend (#​3505)
• Size-threshold auto-offload — large payloads offload to the claim-check store automatically (#​3504)
• Per-message / per-endpoint store selection (#​3508)
• Honor a DI-registered IClaimCheckStore (#​3564)
• New claim-check backends are now packed for NuGet (#​3565)

Distributed agents & durability

• Recover exclusive listener inboxes on the listening node (#​3590)
• Reverse the reset-clears-queue-tables behavior; add IHost.ClearAllWolverineStorageAsync() (#​3592)
• Subscription health check uses the tenant-scoped high water mark for per-tenant subscription agents (#​3582)
• Isolate agent start failures so one wedged agent no longer skips its siblings (#​3519)
• Restart a projection agent whose shard wedged out from under it (#​3519)
• EventSubscriptionAgent restores continuous execution after Rebuild/Rewind (#​3520)
• Stamp the assigned node onto owned event-store daemons (running_on_node, marten#​5001) (#​3578)
• Solo-mode health-check loop no longer leaks an unbounded OpenTelemetry trace (#​3518)
• Warn loudly when a host's application assembly is pinned by an earlier host (#​3521)

HTTP & OpenAPI

• Describe chain-bound OpenAPI params from produced variables; narrow re-reflection to postprocessors (#​3601)
• Bind [FromQuery] on arrays/collections instead of misrouting to complex-flattening (#​3602)
• Bind scalar [FromQuery] decimal correctly + shape-test OpenAPI parameter description across type families (#​3586)
• Don't describe a route-bound [FromQuery]/[FromHeader] parameter twice in OpenAPI (#​3586)
• Marten 9.18 streaming types — StreamPaged, StreamPagedByCursor, ETag support (#​3593)
• Honour opt-in discovery filters for HTTP endpoint discovery (#​3525)
• Never infer a DbContext parameter as the HTTP request body (#​3538)

Transports

• Add a parallel WolverineFx.Mqtt5 package (MQTTnet 5) (#​3517)
• Pulsar: don't claim native scheduling without a retry-letter topic (#​3491)
• Azure Service Bus: pin session listeners to specific session identifiers (#​3533)

Persistence

• Clear transport queue tables on message-store reset — PostgreSQL (#​3529), SQLite (#​3553), MySQL (#​3551), Oracle (#​3552)
• Oracle: durable inbox mark-as-handled binds RAW(16) Guid correctly (#​3581)
  ... (truncated)

6.21.0

Wolverine 6.21.0 is a big one: conjoined multi-tenancy for EF Core, and a measured messaging-performance wave across Kafka and RabbitMQ.

Conjoined multi-tenancy for EF Core (#​3465)

Mark an EF Core entity with ITenanted (the marker shared critter-stack-wide from JasperFx.MultiTenancy) and register your DbContext with AddDbContextWithWolverineManagedConjoinedTenancy<T>(), and Wolverine gives you what Marten users have had for years: a mapped tenant_id column, a tenant-bound global query filter you can't forget, stamp-on-insert, CrossTenantWriteException on cross-tenant writes, conjoined sagas, opt-in Weasel-managed physical tenant partitioning (PostgreSQL list partitions + SQL Server tenant-ordinal), and an authoritative wolverine_tenants registry that doubles as a dynamic tenant source and feeds CritterWatch tenant management. The behavior is checked against a port of Marten's conjoined-tenancy compliance battery. See the new ConjoinedMultiTenantedEfCore sample app and docs.

Messaging performance (GH-3490 / GH-3492)

A client-reported "Wolverine-over-Kafka is 3-12x slower than native" investigation turned into a measured optimization wave (methodology, rigs, and full ledgers are in the repo):

• Sender batching now flushes within the batch timeout (JasperFx 2.30.1). The batch timer was a quiet-period debounce — a steady 8 msg/s stream at the default (100, 250ms) settings measured 5.8 seconds publish-to-consume p50; it now measures 136ms, bounded by the timeout. Affects every transport that sends through the batched sender (RabbitMQ routes were unaffected — they don't).
• Durable (inbox-backed) listeners batch their inbox writes. Kafka: a 2,000 msg/s stream went from unbounded backlog to a steady 32ms delivery p50, max sustained durable throughput +83%. RabbitMQ: same load went from 14.7s-and-climbing to 0.8ms delivery p50, max sustained durable throughput +186% (1,086 → 3,101 msg/s). New MaximumMessagesToReceive listener knob on both (default 100; 1 restores strict message-at-a-time persistence).
• Kafka hot path: incoming envelope mapping ~21% faster / ~28% less allocation per message; outgoing ~17% faster; inline senders no longer issue a blocking full-producer flush per send. RabbitMQ's mapper gets the same incoming fix.
• Batch-arrival correctness: the Envelope[] arrival path now applies the same per-envelope guards as single-message arrival (interop serializer unwrap, dead-lettering of unidentifiable messages, expiry, drain latching), and batched inbox writes route to ancillary message stores correctly.
• SQS: SendMessageBatch per-entry failures (throttling, oversize) are now routed to Wolverine's retry machinery instead of being silently dropped — a silent message-loss fix (GH-3493).
• Out-of-the-box RabbitMQ consumption is Inline — one message at a time, which measures identically to an equivalent single-consumer raw-client loop. The new RabbitMQ "Performance Tuning" docs page covers scaling with BufferedInMemory() / ListenerCount() with measured numbers, and a matching page exists for Kafka.

Behavior changes to note

• The per-message "successfully processed" log now defaults to Debug (was Information) — restore with opts.Policies.MessageSuccessLogLevel(LogLevel.Information).
• wolverine-execution-time is now a floating-point histogram and no longer silently drops sub-millisecond executions (same name/unit; the point type changes).

Transports and messaging

• Azure Service Bus: PrefetchCount on listeners and transport defaults (GH-3471)
• SQS: native DelaySeconds for short scheduled sends on standard queues (GH-3472)
• Pulsar: native scheduled delivery via DeliverAt (GH-3470)
• Kafka: KIP-848 next-generation consumer rebalance protocol support (GH-3473)
• Standardized diagnostic headers on dead-letter and retry moves across all transports (GH-3474)
• SignalR cascading responses are no longer dropped when the HTTP context has already flushed (GH-3499)
• The dead-letter metric only counts genuine moves to the error queue (#​3495)
• Global partitioning docs now cover all eight sharded transports (GH-3466)
• Tolerate empty agent URI lists in agent command serialization (GH-3460)

HTTP / gRPC

• Code-first gRPC client streaming: IAsyncEnumerable<TRequest> -> Task<TResponse> handler shape (#​3500)
• Proto-first gRPC client streaming via the new StreamAsync<TRequest, TResponse> overload (#​3459)

Dependencies

JasperFx 2.30.1 (sender-batching max-age fix), Weasel 9.18.1, Marten 9.16.1.

6.20.0

Wolverine 6.20.0

Dependency upgrades (critter stack)

• Marten 9.16.0, Polecat 5.2.0, JasperFx 2.29.0, Weasel 9.17.0 (aligned across the stack).

Multi-tenancy & connection footprint

• Scope tenant scheduled-job polling to the owning node (GH-3376) — under Wolverine-managed distribution, nodes no longer open a scheduled-job polling connection to every tenant database; the footprint scales with databases, not nodes × databases. Plus daemon tracker-subscription leak hygiene.
• Polecat primary-store database-per-tenant (GH-3445) — IntegrateWithWolverine() now honors a database-per-tenant Polecat store and reads MainDatabaseConnectionString.
• DatabaseServerId sourced from DatabaseDescriptor.Port instead of re-parsing (jasperfx#​514).

CritterWatch / connection state

• Degrade-only connection state for Azure Service Bus, GCP Pub/Sub (GH-3237) and Kafka (GH-3454) listeners — heuristics only degrade to Reconnecting/Disconnected from real SDK callbacks; they never synthesize Connected, and resting state is Unknown.

Kafka

• Read record timestamps into SentAt and expose record headers on raw-JSON listeners (GH-3407).
• JsonSerializerOptions on raw-JSON endpoints now actually applies; PublishRawJson mapper registration fixed.
• Bounded the listener shutdown drain instead of awaiting forever (GH-3434).

Sagas

• SagaConcurrencyException now inherits JasperFx.ConcurrencyException (GH-3444) — existing OnException<ConcurrencyException>() policies now catch saga concurrency failures.
• Lightweight RDBMS saga provider treated as a catch-all so Marten/EF Core win precedence in mixed-storage hosts (GH-3443).

Other

• Ancillary Marten store now registers the event-subscription family interface aliases (GH-3438).
• F# code generation improvements (#​3437); TrackedSession not-tracked vs not-routed fix (#​3435).
• Buffered circuit-breaker tests re-scoped to the transport's non-durable contract (GH-3137); durability docs corrections.

6.19.0

CosmosDB

• Add optimistic concurrency (ETag compare-and-swap) to saga persistence (GH-3414) @​mysticmind
• CosmosDbConfiguration.PartitionSagasById(): opt-in, saga id becomes the document partition key (GH-3415) @​mysticmind
• Refuse a CosmosClient whose serializer would drop a saga's id at host start; document the camelCase requirement (GH-3416) @​mysticmind

HTTP / OpenAPI

• Describe the whole route table on a pre-start ApiExplorer read; minimal API and MVC endpoints were silently omitted (GH-3421) @​mysticmind
• Type a Marten/Polecat aggregate-id route parameter as uuid instead of falling back to string (GH-3420) @​mysticmind

Durability / persistence

• Dead letter recovered envelopes whose transport can't be resolved, instead of losing the rest of the batch and rethrowing forever (GH-3413) @​mysticmind
• Measure and surface per-server database connection budgets: OTel gauges + IWolverineObserver.ConnectionBudget (GH-3397) @​jeremydmiller

Test infrastructure only

• Stop IntegrationContext from disposing a class fixture it doesn't own; pin ApplicationAssembly in the CoreTests harness (GH-3423) @​jeremydmiller
• Give the modular monolith fixture its own message storage schema (GH-3413) @​mysticmind
• Stop using_dynamic_multi_tenancy from poisoning its own next run @​mysticmind

Milestone: https://github.com/JasperFx/wolverine/issues?q=is%3Aissue%20state%3Aclosed%20milestone%3A6.19.0
Full Changelog: JasperFx/wolverine@V6.18.0...V6.19.0

6.18.0

Wolverine 6.18.0

A security-relevant serialization fix, a startup-fatal codegen fix, a silently-dead-listener fix in RabbitMQ, the first F# saga codegen support of any persistence provider, and the CI split that makes "merge when green" mean something again.

If you use MassTransit interop over a durable listener, take this release. See the first section.

⚠️ Security-relevant: reserved envelope headers could be spoofed through the durable inbox

#​3408 — fixed in #​3411

EnvelopeSerializer wrote the typed envelope properties to the wire format and then appended every Envelope.Headers entry verbatim, with no reserved-key filter — and the appended entries came last. Because the reader parses reserved keys straight back into typed properties, a Headers entry under a reserved key silently overwrote the real property on the next read.

A value in envelope.Headers["tenant-id"] is inert while the envelope is in memory. It stops being inert the moment the envelope crosses the serializer — any durable listener, the inbox/outbox, or the scheduled-message store:

1. Something puts tenant-id into envelope.Headers.
2. The durable inbox persists the envelope; the header is appended after the (null) typed property.
3. On read back, env.TenantId is set from it.

saga-id reaches another saga's state, and id rewrites Envelope.Id — the inbox's dedupe identity.

This was live, not theoretical. MassTransitEnvelope.TransferData already copies every incoming MassTransit header into envelope.Headers unfiltered (and by assignment, not TryAdd). Any Wolverine app doing MassTransit interop over a durable listener has had this path open. If that describes you, this release is the one to take.

The fix filters reserved keys on the write side, so the typed property stays authoritative and a reserved key sitting in Headers becomes a no-op. causation-id is deliberately not filtered — DeliveryOptions intentionally carries it as a loose header for Wolverine.Marten's OutboxedSessionFactory, and it is never promoted by the reader.

Startup-fatal codegen fix

#​3399 — fixed in #​3406 — invalid generated class name for batched (array) message types. This one prevents the application from starting.

Fixes

• #​3388 (#​3400) — refuse a competing Marten daemon under Wolverine-managed event subscription distribution. A DaemonMode.Solo/HotCold daemon alongside managed distribution is now an actionable startup exception instead of two schedulers quietly fighting over the same shards.
• CritterWatch #​698 (#​3396) — IAgentRuntime.ApplyRestrictionsAsync persisted the restriction and then never dispatched the commands it computed, so pausing an agent had no immediate effect. Reported by @​erdtsieck against a live cluster.
• #​3385 (#​3403) — a header-identified saga invoked over a gRPC hop failed with an opaque Internal status. It now returns an actionable diagnostic telling you to put the saga identity on the request DTO.
• #​3398 (#​3404) — [AsParameters] now rejects unparseable values in collection query parameters, closing the gap left by the scalar fix in #​3372.
• #​3365 (#​3412) — the Polecat primary IEventStore bridge registered twice, so GetServices<IEventStore>() returned the same store instance two times and anything iterating it double-counted. Polecat's own AddPolecat() had started registering IEventStore and Wolverine was still bridging it as well.
• #​3391 (#​3419) — RabbitMQ: a successful eager channel restart never re-consumed. A callback-exception restart could leave an open channel with zero consumers while reporting State = Connected — a silently dead listener. The listener now defers to ReconnectedAsync(), which re-declares and re-consumes. Also pins the ConnectionMonitor tracking invariant that #​3370 fixed but nothing guarded.

OpenAPI

#​3380 (#​3418) — OpenAPI parameters are now derived from the full binding chain rather than the handler signature alone. Two real defects closed:

• Query/header values bound only by an After/Finally postprocessor were omitted from the operation entirely.
• Route parameter types were read off resolved binding variables, so they degraded to the route constraint (or string) whenever the description was assembled before those frames resolved — which is exactly the build-time OpenAPI / openapi CLI path, because ASP.NET caches the first ApiExplorer read.

More importantly, this ships the OpenAPI shape-test harness that was missing. Adding a shape assertion is now one endpoint plus one [Fact], which is why this class of omission kept shipping unnoticed.

New: Azure Service Bus emulator support

#​3366 (#​3409) — the docs told you to call UseAzureServiceBusTesting(), which only ever existed in Wolverine's own test suite. It is now a real, shipping API:

... (truncated)

6.17.3

Bug-fix and scale release, following the 6.17.2 community sweep. Every item below came from a community report or a review finding — thank you all.

Closed issues

• #​3375 — durability metrics polling pinned a connection per database per node (PR #​3384 by @​erdtsieck). Each durability agent ran its own in-phase PeriodicTimer, so at high database counts the metrics polling itself became significant connection pressure. Agents now register their store with a node-wide sequential sweeper that walks the node's databases one at a time across the UpdateMetricsPeriod window — at most one metrics connection in flight per node, regardless of database count. The registration set is re-read every pass, so databases join and leave the sweep as agents start and stop without a restart.
• #​3332 — CIAWS was disabled by a broken SNS per-tenant LocalStack setup (PR #​3364 by @​Steve-XYZ). Test-only; re-enables the AWS CI job. Partially addresses #​3350 (the CIPolecat half remains open).
• RabbitMQ listeners ghosted after a broker restart on a channel callback exception (PR #​3370 by @​kconfesor). The agent stayed latched after a channel-only shutdown and was never rebuilt, so a listener came back "connected" but dead. Reviewed specifically against the #​3171/#​3187 channel-only-shutdown work to confirm it does not reintroduce the latched-Disconnected state that #​3187 fixed. Two follow-ups are tracked in #​3391.
• ProductSupport#​33 — CritterWatch telemetry was caught by tracked-session waits (PR #​3390, reported by @​uniquelau). A monitored host publishes telemetry that a TrackedSession would pick up and then sit waiting on messages the test never sent. The default ignore rule now covers all of INotToBeRouted (agent commands and framework telemetry), with a deliberate carve-out for Acknowledgement / FailureAcknowledgement, which the session's own acknowledgement APIs depend on. If you are on an older version, IgnoreMessagesMatchingType(t => t.CanBeCastTo<INotToBeRouted>()) is the workaround.

Fixes from review

• Metrics sweeper: unregistration race and a hot-spin guard (PR #​3393, follow-ups from the #​3384 post-merge review). The sweeper removed a registration by URI alone, so when an agent for a database stopped after a replacement agent for the same database had registered — exactly what agent redistribution does — the stopping agent evicted the live registration, and that database silently stopped being polled until the node restarted. Unregistration now removes only the exact registration instance it created. Separately, UpdateMetricsPeriod = TimeSpan.Zero would hot-spin the sweep loop (the pre-#​3384 PeriodicTimer threw); it is now rejected at configuration time, with DurabilityMetricsEnabled = false as the way to turn polling off.

Marten test-helper: PauseThenCatchUpOnMartenDaemonActivity

• #​3388 — cold first catch-up appeared to stall (PR #​3394, reported by @​uniquelau). Investigated in depth. The reported mechanism — that coordinator.ResumeAsync() does not start never-started shards — does not hold: under Wolverine-managed distribution the coordinator is WolverineProjectionCoordinator, whose ResumeAsync builds the daemon lazily and starts every shard, bypassing agent assignment entirely. The cold path works, and there are now four tests proving it (including with a second subscription-agent consumer sharing the agent family).

  The real defect was a timeout mismatch, and it explains the reported symptom exactly. The stage runs inside a child TrackedSession whose token cancels at TrackedSession.Timeout — 5 seconds by default — while the catch-up ignored that token and waited on an internal 60-second budget. The session gave up first and left the catch-up envelope started-but-never-finished, which reads as a hang. This is a genuine 6.16 → 6.17 behavior change: the old active ForceAll finished inside 5 seconds; resume-and-wait on a cold daemon or a busy machine does not. The catch-up now honors the session's token and raises an actionable TimeoutException naming the store and pointing at TrackActivity().Timeout(...).

  If you hit this on 6.17.0–6.17.2, raising the tracked-session timeout is the fix.

Docs

• New page: gRPC + Sagas (PR #​3389, following @​erikshafer's coverage in PR #​3386). gRPC services can start and continue sagas with no gRPC-specific code — the saga identity must be on the message body, because a gRPC method is a thin shim in front of IMessageBus.InvokeAsync<T> and the chain that runs is the handler's. The header-identified gap is tracked as #​3385, with a clear diagnostic planned.
• Testing guide (PR #​3395): tracked sessions ignore framework telemetry by default as of this release, and — the trap behind #​3388 — Timeout() bounds the whole session including its stages, so a slow stage like PauseThenCatchUpOnMartenDaemonActivity() is capped by the session's 5-second default, not by any budget internal to the stage.

Full changelog: JasperFx/wolverine@V6.17.2...V6.17.3

Commits viewable in compare view.

Updated WolverineFx.Http from 6.17.2 to 6.23.1.

Release notes

Sourced from WolverineFx.Http's releases.

6.23.1

Agent distribution

TL;DR: if you pause a projection from CritterWatch on 6.23.0, restarting it appears to do nothing for a full minute. This fixes that.

• A paused projection or subscription agent now resumes immediately when you restart it (#​3663). On 6.23.0 the restart was accepted, the pause restriction was cleared, and then nothing happened until the pending-assignment ledger's TTL expired — 2 × CheckAssignmentPeriod, so 60 seconds with the defaults. Long enough that an operator reasonably concludes the agent is never coming back.

  The ledger introduced in 6.23.0 (#​3622) only counted an assignment as confirmed if a later evaluation saw the agent running and still assigned to the same node. A pause makes those two conditions mutually exclusive: the first evaluation that can observe the delivered assignment is the same one that detaches the agent. The entry was never confirmed, nothing on the stop path cleared it, and the restart's AssignAgent was suppressed as a duplicate still in flight. Delivery alone now confirms the entry, which is the only question the ledger was ever asking.

• Pausing an agent no longer briefly starts it first (#​3666). ApplyRestrictionsAsync kickstarted a health check before persisting the operator's restriction change, so that evaluation ran against the old restrictions and could act against the very intent being applied — for a pause, re-assigning and starting the agent one beat before the merged evaluation stopped it again. Besides the wasted daemon start/stop cycle, this is what armed the stale ledger entry behind #​3663.

PostgreSQL

• Advisory-lock sessions stay invisible to Marten's async-daemon gap detection (#​3664). Marten 9.16.1+ will not skip a stale event-sequence gap while any session whose open transaction predates that gap is still alive (marten#​4953). A session parked in an open transaction for the life of the process therefore reads as a permanent "possible reserver" and can hold the high-water mark — and every async projection — behind a gap that is genuinely dead.

  Wolverine's long-held locks were already shaped correctly: leader election and node coordination hold session-scoped advisory locks on a dedicated connection with no transaction, so they show up as state='idle' with a NULL xact_start. Those sessions are now also tagged application_name = 'wolverine-advisory-lock:<database>', which turns a pg_stat_activity investigation from guesswork into something you can read at a glance. The constraints are pinned in tests and in the Postgres durability docs, including the trap worth knowing in your own code: never add a keepalive query inside a long-lived open transaction — it bumps state_change, makes the session look active, and re-promotes it to candidate reserver.

  Note for combined Marten + Wolverine deployments: older guidance suggested Postgres's idle_in_transaction_session_timeout as a dead-gap backstop. Prefer upgrading Marten and using SkipStaleGapsDespiteLiveTransactionsAfter instead.

6.23.0

Projection & agent distribution

TL;DR: this prevents Wolverine from going into a panic doing agent assignments and churning crazily hard during Kubernetes rollouts or cluster starts and that's a very good thing

The bulk of this release. A wave of fixes (WO-1..8) to the agent assignment plane that together remove the re-assignment churn and livelock that could leave projection agents flapping or wedged.

• Node heartbeats no longer block behind command execution (#​3620)
• Node resurrection restores the node's real identity rather than a skeleton record (#​3621)
• A pending-assignment ledger suppresses duplicate AssignAgent floods (#​3622)
• Agent batch starts are chunked and bounded-parallel, so they stay safe at scale (#​3623)
• Ejection hysteresis plus leader protection (#​3624)
• The assignment row is re-upserted for agents that are already running (#​3619)
• Local agents drain with bounded parallelism on shutdown (#​3636)
• Paused shards now say why. IEventSubscriptionAgent.Failure surfaces a ShardFailure — category, the failing event's sequence and type, and the root exception type — through health checks and a new IWolverineObserver.AgentPaused hook plus a NodeRecordType.AgentPaused record. Failures bound to a specific event (ApplyEvent, EventSerialization, UnknownEventType) or to two processes racing one shard (ProgressionOutOfOrder) are no longer auto-restarted, since they would die on the identical event every time (#​3637, #​3638)
• Agent starts retry locally, bounded by Durability.AgentStartRetryAttempts / AgentStartRetryDelay, instead of idling a full CheckAssignmentPeriod after a startup race (#​3519)
• Store-scoped FindAgentUriAsync overload (#​3647) and a store-aware TryRebuildRegisteredProjectionAsync overload (#​3618)

Bumps JasperFx.Events to 2.36.1, Marten to 9.20.0 and Polecat to 5.7.0.

HTTP

• Endpoints that advertise a request body their HTTP method cannot carry now fail fast at startup instead of silently 404ing out of the route matcher (#​3648)
• A missing Content-Type on an [AcceptsContentType] route returns 415 rather than 404 (#​3649)
• Query-only [AsParameters] endpoints no longer advertise a form body, which had been dropping them from route matching entirely (#​3630)
• Explicitly-routed chains — PublishMessage<T> and SendMessage<T> — now describe the message they read from the body. They had been advertising no request body at all in OpenAPI (#​3646)
• Enum values in query strings and form posts parse case-insensitively (#​3629)
• Compiled queries returned as a Wolverine.Http resource no longer execute twice (#​3625)

Transports

• Conventional routing respects named brokers (#​3633)
• Google Pub/Sub: named-broker support for sharded and partitioned topics (#​3632), and ListenToPubsubSubscriptionOnNamedBroker (#​3631)
• Kafka: topic-already-exists error codes are handled rather than thrown (#​3640)
• Redis: scheduled retries survive an unreadable timestamp (#​3613), and scheduled entries that repeatedly fail to deserialize are dead-lettered instead of looping (#​3644)
• EnvelopeMapper reads both timestamp header formats (#​3645)
• NServiceBus interop: the EnclosedMessageTypes header is split before the message type is resolved, so interop works across Azure Service Bus, SNS, SQS and the database transports (#​3628)

Persistence

• Oracle runs the durability agent through the shared batching mechanics (#​3614). Note that ODP.NET exposes no DbBatch, so the work splits per statement.

Other

• Resource discovery no longer eager-connects broker transports (#​3617)

Contributors

Thank you to everyone who contributed to this release:

... (truncated)

6.22.0

Wolverine 6.22.0 rolls up the claim-check backend wave, distributed-agent and durability hardening, HTTP/OpenAPI binding fixes, and the Marten 9.18 / JasperFx 2.34 critter-stack alignment.

Dependency alignment

• JasperFx family → 2.34.0
• Marten → 9.18.0 (with Marten.AspNetCore / Marten.Newtonsoft), Polecat [5.5.0,6.0.0)
• Weasel family → 9.18.1

Claim-check offloading

• New NATS JetStream Object Store claim-check backend (#​3506)
• New PostgreSQL database-LOB claim-check backend (#​3507)
• New Google Cloud Storage claim-check backend (#​3505)
• Size-threshold auto-offload — large payloads offload to the claim-check store automatically (#​3504)
• Per-message / per-endpoint store selection (#​3508)
• Honor a DI-registered IClaimCheckStore (#​3564)
• New claim-check backends are now packed for NuGet (#​3565)

Distributed agents & durability

• Recover exclusive listener inboxes on the listening node (#​3590)
• Reverse the reset-clears-queue-tables behavior; add IHost.ClearAllWolverineStorageAsync() (#​3592)
• Subscription health check uses the tenant-scoped high water mark for per-tenant subscription agents (#​3582)
• Isolate agent start failures so one wedged agent no longer skips its siblings (#​3519)
• Restart a projection agent whose shard wedged out from under it (#​3519)
• EventSubscriptionAgent restores continuous execution after Rebuild/Rewind (#​3520)
• Stamp the assigned node onto owned event-store daemons (running_on_node, marten#​5001) (#​3578)
• Solo-mode health-check loop no longer leaks an unbounded OpenTelemetry trace (#​3518)
• Warn loudly when a host's application assembly is pinned by an earlier host (#​3521)

HTTP & OpenAPI

• Describe chain-bound OpenAPI params from produced variables; narrow re-reflection to postprocessors (#​3601)
• Bind [FromQuery] on arrays/collections instead of misrouting to complex-flattening (#​3602)
• Bind scalar [FromQuery] decimal correctly + shape-test OpenAPI parameter description across type families (#​3586)
• Don't describe a route-bound [FromQuery]/[FromHeader] parameter twice in OpenAPI (#​3586)
• Marten 9.18 streaming types — StreamPaged, StreamPagedByCursor, ETag support (#​3593)
• Honour opt-in discovery filters for HTTP endpoint discovery (#​3525)
• Never infer a DbContext parameter as the HTTP request body (#​3538)

Transports

• Add a parallel WolverineFx.Mqtt5 package (MQTTnet 5) (#​3517)
• Pulsar: don't claim native scheduling without a retry-letter topic (#​3491)
• Azure Service Bus: pin session listeners to specific session identifiers (#​3533)

Persistence

• Clear transport queue tables on message-store reset — PostgreSQL (#​3529), SQLite (#​3553), MySQL (#​3551), Oracle (#​3552)
• Oracle: durable inbox mark-as-handled binds RAW(16) Guid correctly (#​3581)
  ... (truncated)

6.21.0

Wolverine 6.21.0 is a big one: conjoined multi-tenancy for EF Core, and a measured messaging-performance wave across Kafka and RabbitMQ.

Conjoined multi-tenancy for EF Core (#​3465)

Mark an EF Core entity with ITenanted (the marker shared critter-stack-wide from JasperFx.MultiTenancy) and register your DbContext with AddDbContextWithWolverineManagedConjoinedTenancy<T>(), and Wolverine gives you what Marten users have had for years: a mapped tenant_id column, a tenant-bound global query filter you can't forget, stamp-on-insert, CrossTenantWriteException on cross-tenant writes, conjoined sagas, opt-in Weasel-managed physical tenant partitioning (PostgreSQL list partitions + SQL Server tenant-ordinal), and an authoritative wolverine_tenants registry that doubles as a dynamic tenant source and feeds CritterWatch tenant management. The behavior is checked against a port of Marten's conjoined-tenancy compliance battery. See the new ConjoinedMultiTenantedEfCore sample app and docs.

Messaging performance (GH-3490 / GH-3492)

A client-reported "Wolverine-over-Kafka is 3-12x slower than native" investigation turned into a measured optimization wave (methodology, rigs, and full ledgers are in the repo):

• Sender batching now flushes within the batch timeout (JasperFx 2.30.1). The batch timer was a quiet-period debounce — a steady 8 msg/s stream at the default (100, 250ms) settings measured 5.8 seconds publish-to-consume p50; it now measures 136ms, bounded by the timeout. Affects every transport that sends through the batched sender (RabbitMQ routes were unaffected — they don't).
• Durable (inbox-backed) listeners batch their inbox writes. Kafka: a 2,000 msg/s stream went from unbounded backlog to a steady 32ms delivery p50, max sustained durable throughput +83%. RabbitMQ: same load went from 14.7s-and-climbing to 0.8ms delivery p50, max sustained durable throughput +186% (1,086 → 3,101 msg/s). New MaximumMessagesToReceive listener knob on both (default 100; 1 restores strict message-at-a-time persistence).
• Kafka hot path: incoming envelope mapping ~21% faster / ~28% less allocation per message; outgoing ~17% faster; inline senders no longer issue a blocking full-producer flush per send. RabbitMQ's mapper gets the same incoming fix.
• Batch-arrival correctness: the Envelope[] arrival path now applies the same per-envelope guards as single-message arrival (interop serializer unwrap, dead-lettering of unidentifiable messages, expiry, drain latching), and batched inbox writes route to ancillary message stores correctly.
• SQS: SendMessageBatch per-entry failures (throttling, oversize) are now routed to Wolverine's retry machinery instead of being silently dropped — a silent message-loss fix (GH-3493).
• Out-of-the-box RabbitMQ consumption is Inline — one message at a time, which measures identically to an equivalent single-consumer raw-client loop. The new RabbitMQ "Performance Tuning" docs page covers scaling with BufferedInMemory() / ListenerCount() with measured numbers, and a matching page exists for Kafka.

Behavior changes to note

• The per-message "successfully processed" log now defaults to Debug (was Information) — restore with opts.Policies.MessageSuccessLogLevel(LogLevel.Information).
• wolverine-execution-time is now a floating-point histogram and no longer silently drops sub-millisecond executions (same name/unit; the point type changes).

Transports and messaging

• Azure Service Bus: PrefetchCount on listeners and transport defaults (GH-3471)
• SQS: native DelaySeconds for short scheduled sends on standard queues (GH-3472)
• Pulsar: native scheduled delivery via DeliverAt (GH-3470)
• Kafka: KIP-848 next-generation consumer rebalance protocol support (GH-3473)
• Standardized diagnostic headers on dead-letter and retry moves across all transports (GH-3474)
• SignalR cascading responses are no longer dropped when the HTTP context has already flushed (GH-3499)
• The dead-letter metric only counts genuine moves to the error queue (#​3495)
• Global partitioning docs now cover all eight sharded transports (GH-3466)
• Tolerate empty agent URI lists in agent command serialization (GH-3460)

HTTP / gRPC

• Code-first gRPC client streaming: IAsyncEnumerable<TRequest> -> Task<TResponse> handler shape (#​3500)
• Proto-first gRPC client streaming via the new StreamAsync<TRequest, TResponse> overload (#​3459)

Dependencies

JasperFx 2.30.1 (sender-batching max-age fix), Weasel 9.18.1, Marten 9.16.1.

6.20.0

Wolverine 6.20.0

Dependency upgrades (critter stack)

• Marten 9.16.0, Polecat 5.2.0, JasperFx 2.29.0, Weasel 9.17.0 (aligned across the stack).

Multi-tenancy & connection footprint

• Scope tenant scheduled-job polling to the owning node (GH-3376) — under Wolverine-managed distribution, nodes no longer open a scheduled-job polling connection to every tenant database; the footprint scales with databases, not nodes × databases. Plus daemon tracker-subscription leak hygiene.
• Polecat primary-store database-per-tenant (GH-3445) — IntegrateWithWolverine() now honors a database-per-tenant Polecat store and reads MainDatabaseConnectionString.
• DatabaseServerId sourced from DatabaseDescriptor.Port instead of re-parsing (jasperfx#​514).

CritterWatch / connection state

• Degrade-only connection state for Azure Service Bus, GCP Pub/Sub (GH-3237) and Kafka (GH-3454) listeners — heuristics only degrade to Reconnecting/Disconnected from real SDK callbacks; they never synthesize Connected, and resting state is Unknown.

Kafka

• Read record timestamps into SentAt and expose record headers on raw-JSON listeners (GH-3407).
• JsonSerializerOptions on raw-JSON endpoints now actually applies; PublishRawJson mapper registration fixed.
• Bounded the listener shutdown drain instead of awaiting forever (GH-3434).

Sagas

• SagaConcurrencyException now inherits JasperFx.ConcurrencyException (GH-3444) — existing OnException<ConcurrencyException>() policies now catch saga concurrency failures.
• Lightweight RDBMS saga provider treated as a catch-all so Marten/EF Core win precedence in mixed-storage hosts (GH-3443).

Other

• Ancillary Marten store now registers the event-subscription family interface aliases (GH-3438).
• F# code generation improvements (#​3437); TrackedSession not-tracked vs not-routed fix (#​3435).
• Buffered circuit-breaker tests re-scoped to the transport's non-durable contract (GH-3137); durability docs corrections.

6.19.0

CosmosDB

• Add optimistic concurrency (ETag compare-and-swap) to saga persistence (GH-3414) @​mysticmind
• CosmosDbConfiguration.PartitionSagasById(): opt-in, saga id becomes the document partition key (GH-3415) @​mysticmind
• Refuse a CosmosClient whose serializer would drop a saga's id at host start; document the camelCase requirement (GH-3416) @​mysticmind

HTTP / OpenAPI

• Describe the whole route table on a pre-start ApiExplorer read; minimal API and MVC endpoints were silently omitted (GH-3421) @​mysticmind
• Type a Marten/Polecat aggregate-id route parameter as uuid instead of falling back to string (GH-3420) @​mysticmind

Durability / persistence

• Dead letter recovered envelopes whose transport can't be resolved, instead of losing the rest of the batch and rethrowing forever (GH-3413) @​mysticmind
• Measure and surface per-server database connection budgets: OTel gauges + IWolverineObserver.ConnectionBudget (GH-3397) @​jeremydmiller

Test infrastructure only

• Stop IntegrationContext from disposing a class fixture it doesn't own; pin ApplicationAssembly in the CoreTests harness (GH-3423) @​jeremydmiller
• Give the modular monolith fixture its own message storage schema (GH-3413) @​mysticmind
• Stop using_dynamic_multi_tenancy from poisoning its own next run @​mysticmind

Milestone: https://github.com/JasperFx/wolverine/issues?q=is%3Aissue%20state%3Aclosed%20milestone%3A6.19.0
Full Changelog: JasperFx/wolverine@V6.18.0...V6.19.0

6.18.0

Wolverine 6.18.0

A security-relevant serialization fix, a startup-fatal codegen fix, a silently-dead-listener fix in RabbitMQ, the first F# saga codegen support of any persistence provider, and the CI split that makes "merge when green" mean something again.

If you use MassTransit interop over a durable listener, take this release. See the first section.

⚠️ Security-relevant: reserved envelope headers could be spoofed through the durable inbox

#​3408 — fixed in #​3411

EnvelopeSerializer wrote the typed envelope properties to the wire format and then appended every Envelope.Headers entry verbatim, with no reserved-key filter — and the appended entries came last. Because the reader parses reserved keys straight back into typed properties, a Headers entry under a reserved key silently overwrote the real property on the next read.

A value in envelope.Headers["tenant-id"] is inert while the envelope is in memory. It stops being inert the moment the envelope crosses the serializer — any durable listener, the inbox/outbox, or the scheduled-message store:

1. Something puts tenant-id into envelope.Headers.
2. The durable inbox persists the envelope; the header is appended after the (null) typed property.
3. On read back, env.TenantId is set from it.

saga-id reaches another saga's state, and id rewrites Envelope.Id — the inbox's dedupe identity.

This was live, not theoretical. MassTransitEnvelope.TransferData already copies every incoming MassTransit header into envelope.Headers unfiltered (and by assignment, not TryAdd). Any Wolverine app doing MassTransit interop over a durable listener has had this path open. If that describes you, this release is the one to take.

The fix filters reserved keys on the write side, so the typed property stays authoritative and a reserved key sitting in Headers becomes a no-op. causation-id is deliberately not filtered — DeliveryOptions intentionally carries it as a loose header for Wolverine.Marten's OutboxedSessionFactory, and it is never promoted by the reader.

Startup-fatal codegen fix

#​3399 — fixed in #​3406 — invalid generated class name for batched (array) message types. This one prevents the application from starting.

Fixes

• #​3388 (#​3400) — refuse a competing Marten daemon under Wolverine-managed event subscription distribution. A DaemonMode.Solo/HotCold daemon alongside managed distribution is now an actionable startup exception instead of two schedulers quietly fighting over the same shards.
• CritterWatch #​698 (#​3396) — IAgentRuntime.ApplyRestrictionsAsync persisted the restriction and then never dispatched the commands it computed, so pausing an agent had no immediate effect. Reported by @​erdtsieck against a live cluster.
• #​3385 (#​3403) — a header-identified saga invoked over a gRPC hop failed with an opaque Internal status. It now returns an actionable diagnostic telling you to put the saga identity on the request DTO.
• #​3398 (#​3404) — [AsParameters] now rejects unparseable values in collection query parameters, closing the gap left by the scalar fix in #​3372.
• #​3365 (#​3412) — the Polecat primary IEventStore bridge registered twice, so GetServices<IEventStore>() returned the same store instance two times and anything iterating it double-counted. Polecat's own AddPolecat() had started registering IEventStore and Wolverine was still bridging it as well.
• #​3391 (#​3419) — RabbitMQ: a successful eager channel restart never re-consumed. A callback-exception restart could leave an open channel with zero consumers while reporting State = Connected — a silently dead listener. The listener now defers to ReconnectedAsync(), which re-declares and re-consumes. Also pins the ConnectionMonitor tracking invariant that #​3370 fixed but nothing guarded.

OpenAPI

#​3380 (#​3418) — OpenAPI parameters are now derived from the full binding chain rather than the handler signature alone. Two real defects closed:

• Query/header values bound only by an After/Finally postprocessor were omitted from the operation entirely.
• Route parameter types were read off resolved binding variables, so they degraded to the route constraint (or string) whenever the description was assembled before those frames resolved — which is exactly the build-time OpenAPI / openapi CLI path, because ASP.NET caches the first ApiExplorer read.

More importantly, this ships the OpenAPI shape-test harness that was missing. Adding a shape assertion is now one endpoint plus one [Fact], which is why this class of omission kept shipping unnoticed.

New: Azure Service Bus emulator support

#​3366 (#​3409) — the docs told you to call UseAzureServiceBusTesting(), which only ever existed in Wolverine's own test suite. It is now a real, shipping API:

... (truncated)

6.17.3

Bug-fix and scale release, following the 6.17.2 community sweep. Every item below came from a community report or a review finding — thank you all.

Closed issues

• #​3375 — durability metrics polling pinned a connection per database per node (PR #​3384 by @​erdtsieck). Each durability agent ran its own in-phase PeriodicTimer, so at high database counts the metrics polling itself became significant connection pressure. Agents now register their store with a node-wide sequential sweeper that walks the node's databases one at a time across the UpdateMetricsPeriod window — at most one metrics connection in flight per node, regardless of database count. The registration set is re-read every pass, so databases join and leave the sweep as agents start and stop without a restart.
• #​3332 — CIAWS was disabled by a broken SNS per-tenant LocalStack setup (PR #​3364 by @​Steve-XYZ). Test-only; re-enables the AWS CI job. Partially addresses #​3350 (the CIPolecat half remains open).
• RabbitMQ listeners ghosted after a broker restart on a channel callback exception (PR #​3370 by @​kconfesor). The agent stayed latched after a channel-only shutdown and was never rebuilt, so a listener came back "connected" but dead. Reviewed specifically against the #​3171/#​3187 channel-only-shutdown work to confirm it does not reintroduce the latched-Disconnected state that #​3187 fixed. Two follow-ups are tracked in #​3391.
• ProductSupport#​33 — CritterWatch telemetry was caught by tracked-session waits (PR #​3390, reported by @​uniquelau). A monitored host publishes telemetry that a TrackedSession would pick up and then sit waiting on messages the test never sent. The default ignore rule now covers all of INotToBeRouted (agent commands and framework telemetry), with a deliberate carve-out for Acknowledgement / FailureAcknowledgement, which the session's own acknowledgement APIs depend on. If you are on an older version, IgnoreMessagesMatchingType(t => t.CanBeCastTo<INotToBeRouted>()) is the workaround.

Fixes from review

• Metrics sweeper: unregistration race and a hot-spin guard (PR #​3393, follow-ups from the #​3384 post-merge review). The sweeper removed a registration by URI alone, so when an agent for a database stopped after a replacement agent for the same database had registered — exactly what agent redistribution does — the stopping agent evicted the live registration, and that database silently stopped being polled until the node restarted. Unregistration now removes only the exact registration instance it created. Separately, UpdateMetricsPeriod = TimeSpan.Zero would hot-spin the sweep loop (the pre-#​3384 PeriodicTimer threw); it is now rejected at configuration time, with DurabilityMetricsEnabled = false as the way to turn polling off.

Marten test-helper: PauseThenCatchUpOnMartenDaemonActivity

• #​3388 — cold first catch-up appeared to stall (PR #​3394, reported by @​uniquelau). Investigated in depth. The reported mechanism — that coordinator.ResumeAsync() does not start never-started shards — does not hold: under Wolverine-managed distribution the coordinator is WolverineProjectionCoordinator, whose ResumeAsync builds the daemon lazily and starts every shard, bypassing agent assignment entirely. The cold path works, and there are now four tests proving it (including with a second subscription-agent consumer sharing the agent family).

  The real defect was a timeout mismatch, and it explains the reported symptom exactly. The stage runs inside a child TrackedSession whose token cancels at TrackedSession.Timeout — 5 seconds by default — while the catch-up ignored that token and waited on an internal 60-second budget. The session gave up first and left the catch-up envelope started-but-never-finished, which reads as a hang. This is a genuine 6.16 → 6.17 behavior change: the old active ForceAll finished inside 5 seconds; resume-and-wait on a cold daemon or a busy machine does not. The catch-up now honors the session's token and raises an actionable TimeoutException naming the store and pointing at TrackActivity().Timeout(...).

  If you hit this on 6.17.0–6.17.2, raising the tracked-session timeout is the fix.

Docs

• New page: gRPC + Sagas (PR #​3389, following @​erikshafer's coverage in PR #​3386). gRPC services can start and continue sagas with no gRPC-specific code — the saga identity must be on the message body, because a gRPC method is a thin shim in front of IMessageBus.InvokeAsync<T> and the chain that runs is the handler's. The header-identified gap is tracked as #​3385, with a clear diagnostic planned.
• Testing guide (PR #​3395): tracked sessions ignore framework telemetry by default as of this release, and — the trap behind #​3388 — Timeout() bounds the whole session including its stages, so a slow stage like PauseThenCatchUpOnMartenDaemonActivity() is capped by the session's 5-second default, not by any budget internal to the stage.

Full changelog: JasperFx/wolverine@V6.17.2...V6.17.3

Commits viewable in compare view.

Dependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting @dependabot rebase.

---

Dependabot commands and options

You can trigger Dependabot actions by commenting on this PR:

• @dependabot rebase will rebase this PR
• @dependabot recreate will recreate this PR, overwriting any edits that have been made to it
• @dependabot show <dependency name> ignore conditions will show all of the ignore conditions of the specified dependency
• @dependabot ignore this major version will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)
• @dependabot ignore this minor version will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)
• @dependabot ignore this dependency will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)

[socket-security]

Review the following changes in direct dependencies. Learn more about Socket for GitHub.

Diff	Package	Supply Chain Security	Vulnerability	Quality	Maintenance	License
	nuget/​wolverinefx@​6.23.1
	nuget/​wolverinefx.http@​6.17.2 ⏵ 6.23.1

View full report

[Powerworks]

Superseded by e3f9593 — WolverineFx/.Http/.Marten/.RabbitMQ/.RuntimeCompilation and Marten/Marten.AspNetCore are consolidated into one manual bump instead of merging this PR individually, since these packages release in lockstep and several of these Dependabot PRs disagreed with each other on the same version lines (e.g. this PR vs #24/#28 both touching Marten to different targets, or #27/#29/#30 all touching WolverineFx independently). Verified: build clean, all 383 non-integration tests pass.

[dependabot]

OK, I won't notify you again about this release, but will get in touch when a new version is available. You can also ignore all major, minor, or patch releases for a dependency by adding an ignore condition with the desired update_types to your config file.

If you change your mind, just re-open this PR and I'll resolve any conflicts on it.