[dependabot]

Pinned Aaron.Akka.Reminders at 1.5.73.4.

Release notes

Sourced from Aaron.Akka.Reminders's releases.

1.5.73.4

1.5.73.4 November 3rd 2026

New Features

• SQL Server storage can now use a connection factory instead of a connection string. Pass a Func<CancellationToken, ValueTask<DbConnection>> to the new WithSqlServerStorage(connectionFactory, ...) overload, or set SqlServerReminderStorageSettings.ConnectionFactory, to supply connections yourself, for example with an Azure AD access token in SqlConnection.AccessToken. The factory is called once per storage operation, so tokens stay fresh, and the storage still opens and disposes each connection. Set exactly one of ConnectionString or ConnectionFactory. Existing connection-string overloads are unchanged (#​172, #​176).

Dependency Updates

• Pinned System.Security.Cryptography.Pkcs to 9.0.21. Microsoft.Data.SqlClient 6.1.4 brought in 9.0.11, which has GHSA-2fmv-rv2c-59x7 (moderate) and GHSA-846f-f764-76mx (high) (#​175).

Commits viewable in compare view.

Pinned Akka.Cluster.Sharding at 1.5.74.

Release notes

Sourced from Akka.Cluster.Sharding's releases.

1.5.74

1.5.74 November 5th, 2026

Akka.NET v1.5.74 backports a batch of Akka.Streams, Cluster.Sharding, Cluster.Tools, Persistence and Hosting bug fixes from dev, plus a large set of de-flaked tests. It also moves NuGet publishing to GitHub Actions.

Akka.Streams

• Ignore Dispose() on the ActorSystem's shared default materializer - system.Materializer() and ActorMaterializer.Create(system) with no custom settings return one materializer that the ActorSystem owns. Disposing it, for example in a using block, through a scoped DI registration, or in an actor's PostStop, shut it down for every caller until the system restarted. Dispose() on that shared instance is now ignored. The ActorSystem still shuts it down on termination, and an explicit Shutdown() still stops it. Materializers created with custom settings or on an IActorContext behave as before.
• Fix Flow.Setup input forwarding - Flow.Setup grabbed upstream elements but never pushed them into the flow it built, so they were lost.
• Fix inherited attributes in deferred stream operators - LazyFlow, LazySink and Source.FromTaskSource now materialize their inner graph with the attributes of the enclosing graph, so dispatcher, supervision and other attributes apply to it.
• Honor StreamRefAttributes on SourceRef - The FinalTerminationSignalDeadline, BufferCapacity and DemandRedeliveryInterval attributes now take effect on a SourceRef. Before, only SubscriptionTimeout did. Fixes #​8878.
• Read final-termination-signal-deadline from HOCON - StreamRefSettings.FinalTerminationSignalDeadline was never loaded from configuration and was always zero.
• Stage correctness fixes:
  • UnfoldFlow no longer hangs on its initial seed (#​7867, #​7906).
  • KeepAlive (IdleInject) no longer injects over an element that upstream already has ready, and its idle deadline starts at the first pull instead of at materialization (#​7863, #​7881).
  • MergePrioritized keeps the configured priority ratios (#​7872).
  • RestartWithBackoff sets its reset deadline before starting the inner graph, so a synchronous completion doesn't see a stale deadline (#​7861).
  • Source.Lazily fails its materialized task with the factory's exception instead of a misleading "stopped without completing" error (#​8843).
  • PartitionHub consumers report the real failure when a hub callback task is cancelled, instead of failing with a null exception (#​8064).
  • The outgoing TCP stage completes cleanly when upstream finishes before the connection is established, instead of logging a NullReferenceException (#​8628).

Akka.Cluster.Sharding

• Defer ShardingProducerController cleanup until the delivery child terminates - A message that arrived while an idle entity's ProducerController was stopping tried to create a new child with the same name, threw InvalidActorNameException and restarted the controller. Messages for that entity are now buffered until the old child has terminated. Fixes #​8850.
• Preserve owner termination during shard handoff - When the region that owned a shard terminated mid-rebalance, the handoff kept sending HandOff to the dead region and stalled until handoff-timeout. It now remembers the termination and finishes the handoff.

Akka.Cluster.Tools

• Handle a failed lease release when ClusterSingletonManager leaves Oldest - A faulted Lease.Release() became an unhandled Status.Failure, so nobody saw that the release failed. It is now handled and logged.

Akka.Persistence

• Fix race condition in ReceivePersistentActor async handlers - The RunTask continuation could run on the thread pool and touch actor state off the actor thread. It now runs synchronously, as intended.

Akka

• Fix ByteString.CopyTo(Span/Memory) throwing "Destination is too short" - The Span<byte> and Memory<byte> overloads threw whenever a copy stopped partway through a segment, including when the destination was sized exactly to count. Fixes #​8469.
• Add Match extension methods for Option<T> - New Match overloads in Akka.Util.Extensions take a function or action for the Some and None cases.

Akka.Hosting

• MAUI detection no longer throws under InvariantGlobalization - AddAkka threw CultureNotFoundException at startup when InvariantGlobalization was on and a satellite assembly, such as one from Microsoft.Data.SqlClient, was loaded.

Build and tests

• Publish to NuGet via GitHub Actions trusted publishing - Release packages are now pushed from GitHub Actions using NuGet trusted publishing (OIDC) instead of from Azure DevOps with a long-lived API key. Package contents are unchanged.
• De-flaked Remote, Hosting, core, Streams, Persistence, DI, Cluster.Sharding, Cluster.Tools, DistributedData and multi-node specs (#​8881, #​8882, #​8883, #​8884). These change tests only.

4 contributors since release 1.5.73

COMMITS	LOC+	LOC-	AUTHOR
65	3030	1265	Aaron Stannard
4	229	87	Gregorius Soedharmo
3	317	12	Ariel Fischer
2	48	9	Matt Kotsenas

To see the full set of changes in Akka.NET v1.5.74, click here.
... (truncated)

Commits viewable in compare view.

Pinned Akka.Persistence.Sql.Hosting at 1.5.73.

Release notes

Sourced from Akka.Persistence.Sql.Hosting's releases.

1.5.73

1.5.73 November 2nd 2026

Akka.NET 1.5.73 and Microsoft.Extensions 9.0

• Bump Akka.NET to 1.5.73
• Bump Akka.Hosting to 1.5.73 - Akka.Hosting 1.5.73 requires Microsoft.Extensions 9.0 or later (previously 8.0), which applies to Akka.Persistence.Sql.Hosting consumers.

LinqToDB 6 and opt-in tag-table insert batching

• Upgrade LinqToDB to 6.5.1 and add opt-in batched journal inserts - Backport of #​580. Upgrades LinqToDB from 5.4 to 6.5.1 and adds optional batched journal inserts for SQL Server, PostgreSQL, and SQLite in TagTable mode. Batching remains off by default; FK installation remains a separate operator-applied migration.
• Retain Akka 1.5 and the existing .NET Standard 2.0/.NET 6 library targets. Preserve legacy MySql.Official provider selection and select batching from the actual provider supplied through DataOptions.
• Preserve independent cloned connections, external-transaction separation, configured retry/provider instances, mapping, command timeout, and query hints.
• Compatibility: the LinqToDB upgrade applies to all users, even with batching off. Custom LinqToDB 5 IDataProvider implementations must be adapted and recompiled. Configure interceptors through DataOptions.UseInterceptor(s); direct post-construction interceptor additions/removals are not carried over to cloned connections. See the batching and upgrade notes.
• Skip the tag-table fast insert for untagged batches and single tagged rows - When batched inserts are enabled, untagged batches use BulkCopy and single tagged rows use the regular row insert; batches containing tagged events keep the fast path.

Bug Fixes

• Cancel SQL queries when journal streams stop - Cancel in-flight database queries when a read-journal or journal-replay stream stops, including Offset.FromEnd and current-query boundary lookups. Preserve the configured scheduler for refresh delays.
• Complete the journal write queue on actor shutdown - Stopping the journal actor now explicitly completes its write queue and cancels pending write-receipt waits instead of leaving buffered writes awaiting promises that never complete.
• Rate-limit SQL initialization retries - Failed journal and snapshot store initialization now waits asynchronously before restarting, controlled by the new initialization-retry-delay setting (default 1s). An unreachable database no longer causes a tight initialization/restart loop.
• Fix SQL materializer lifetime - Journal, snapshot store, and sequence actor materializers are now owned by their actors, so stream supervisors and journal queues no longer accumulate across actor restarts (for example, during repeated initialization failures).

Operations and Documentation

• Add operator-run tag cleanup and opt-in FK migrations - PostgreSQL and SQL Server scripts for auditing and cleaning up orphaned tag rows, and for installing or rolling back an optional ON DELETE CASCADE foreign key from the tag table to the journal. Schema auto-initialization is unchanged; FK adoption remains optional. Includes reproducible FK insert/delete benchmarks.
• Document database-side journal retention - Guidance and example SQL for PostgreSQL, SQL Server, MySQL, and SQLite to purge old journal events with your own SQL without leaving orphaned tag rows behind.

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 <dependency name> major version will close this group update PR and stop Dependabot creating any more for the specific dependency's major version (unless you unignore this specific dependency's major version or upgrade to it yourself)
• @dependabot ignore <dependency name> minor version will close this group update PR and stop Dependabot creating any more for the specific dependency's minor version (unless you unignore this specific dependency's minor version or upgrade to it yourself)
• @dependabot ignore <dependency name> will close this group update PR and stop Dependabot creating any more for the specific dependency (unless you unignore this specific dependency or upgrade to it yourself)
• @dependabot unignore <dependency name> will remove all of the ignore conditions of the specified dependency
• @dependabot unignore <dependency name> <ignore condition> will remove the ignore condition of the specified dependency and ignore conditions