Bump the nuget-minor-patch group with 10 updates

[dependabot]

Updated AWSSDK.DynamoDBv2 from 4.0.103.7 to 4.0.105.

Release notes

Sourced from AWSSDK.DynamoDBv2's releases.

No release notes found for this version range.

Commits viewable in compare view.

Updated AWSSDK.Kinesis from 4.0.103.2 to 4.0.104.

Release notes

Sourced from AWSSDK.Kinesis's releases.

No release notes found for this version range.

Commits viewable in compare view.

Updated AWSSDK.S3 from 4.0.103.1 to 4.0.103.4.

Release notes

Sourced from AWSSDK.S3's releases.

No release notes found for this version range.

Commits viewable in compare view.

Updated AWSSDK.SecretsManager from 4.0.100.13 to 4.0.100.14.

Release notes

Sourced from AWSSDK.SecretsManager's releases.

No release notes found for this version range.

Commits viewable in compare view.

Updated AWSSDK.SimpleNotificationService from 4.0.100.13 to 4.0.100.15.

Release notes

Sourced from AWSSDK.SimpleNotificationService's releases.

No release notes found for this version range.

Commits viewable in compare view.

Updated AWSSDK.SQS from 4.0.100.13 to 4.0.100.14.

Release notes

Sourced from AWSSDK.SQS's releases.

No release notes found for this version range.

Commits viewable in compare view.

Updated Azure.Messaging.ServiceBus from 7.20.2 to 7.21.0.

Release notes

Sourced from Azure.Messaging.ServiceBus's releases.

7.21.0

7.21.0 (2026-10-06)

Features Added

• Added SqlFilterCount and CorrelationFilterCount properties to TopicRuntimeProperties, exposing the total number of SQL filters and correlation filters across all of a topic's subscriptions. These are populated by GetTopicRuntimePropertiesAsync and GetTopicsRuntimePropertiesAsync.
• Added ServiceBusAdministrationClientOptions.ServiceVersion.V2024_05 and made it the default service version. The topic filter counts above are served by the 2024-05 service API version, so the administration client now sends api-version=2024-05 by default.
• Added GetMessageSessionsAsync overloads on ServiceBusClient for queues and subscriptions. The no-filter overload returns the IDs of sessions that have active messages or session state, and the sessionStateUpdatedAfter overload returns session IDs whose session state was updated after the specified timestamp. Implements the com.microsoft:get-message-sessions AMQP management operation. (#​58761)
• Added opt-in support for non-exclusive session locking on ServiceBusSessionReceiver, allowing a session to be cooperatively taken over by another receiver. Set ServiceBusSessionReceiverOptions.EnableNonExclusiveSession to accept a session non-exclusively, then read the token from ServiceBusSessionReceiver.SessionLockToken and pass it as ServiceBusSessionReceiverOptions.SessionLockToken = Guid.Parse(token) to take that session over. ServiceBusSessionReceiver.IsSessionExclusive reports the mode the session was established under. Dispositions for a non-exclusive session are routed over the management link so that settlement keeps working across a takeover, which lowers settlement throughput compared to an exclusive session. This applies to ServiceBusSessionReceiver only; ServiceBusSessionProcessor continues to lock sessions exclusively. Accepting a session with EnableNonExclusiveSession set throws NotSupportedException when the endpoint declines it, either by refusing the request outright or by accepting it without assigning a lock token, which is how a caller detects whether the feature is available for a namespace. An endpoint that declines in some other way surfaces the exception its own error maps to. (#​60060)

Bugs Fixed

• Fixed retry classification for web socket failures with nested causes. On modern .NET, a transient network failure during a web socket connection attempt surfaces as a WebSocketException that wraps an HttpRequestException, which wraps the meaningful IOException or SocketException. The retry policy previously inspected only one level of nesting and treated these failures as terminal. The policy now unwraps nested wrapper exceptions to a bounded depth, so transient failures such as a connection reset use the configured retries. Terminal socket failures, such as host-not-found and host-unreachable, are not retried at any supported depth. A host-unreachable failure on an established connection is now terminal. Earlier versions retried it. (#​61868)

• Fixed a bug where canceling ServiceBusReceiver.CloseAsync left the receiver unable to close its own links. The receiver was marked as closed, and its set of locked messages disposed, before the cancellation was observed, so every later call to CloseAsync returned immediately without doing any work and the links stayed open until the owning ServiceBusClient was disposed. The receiver is now left open and closable when a close does not complete, so the operation can be retried. (#​59309)

Other Changes

• The default ServiceBusAdministrationClient service version is now 2024-05 (previously 2021-05). Existing operations are unaffected in behavior; the change is required to surface the new topic filter count properties.

Commits viewable in compare view.

Updated dotnet-diagnostics-benchmarkdotnet from 0.25.0 to 0.27.0.

Release notes

Sourced from dotnet-diagnostics-benchmarkdotnet's releases.

0.27.0

HTTP destination attribution and trustworthy networking evidence

v0.27.0 adds opt-in outbound HttpClient destination attribution without target changes and corrects networking correlation, failed-operation latency, availability, and partial-capture reporting.

New: HTTP destination evidence

Enable includeHttpDestination=true in MCP, --include-http-destination in the CLI, or IncludeHttpDestination = true in BenchmarkDotNet. Defaults remain false.

The DiagnosticSource bridge captures only scheme/host/port plus real W3C trace/span identity, including runtime-only .NET 8 Activities with empty native HTTP tags. Separate destination availability/provenance flows through activity and trace queries, distributed collection, CLI one-shot/session, and BenchmarkDotNet. Missing, ambiguous, capped, or transport-incomplete evidence is not assigned a backend.

This is the requested URI authority, not necessarily the physical endpoint, proxy, DNS address, or redirect-final destination. Configured redaction applies to structured authority. The new path does not export full URLs, paths, queries or credentials; existing native tags remain unchanged and may already contain URLs.

Networking corrections

• Conservative bounded identity correlation prevents silent concurrent latency misattribution.
• Failed Start/Failed/Stop operations retain their start until completion rather than losing their latency.
• Completion, nullable transport loss and parse errors qualify partial captures throughout summaries and focused views.
• Per-metric availability distinguishes unavailable latency from genuinely measured zero. Accepted sample counts remain distinct from reservoir-retained percentile counts.

Migration

Networking latency population is version 2: percentiles include all accepted completions, including those with observed failure. Inspect latencyPopulationVersion and paired outcome/sample counts before comparing old captures. HTTP 503 is a response status, not automatically a transport failure.

Existing nonnullable duration scalars remain compatibility fields; consult latencyAvailability and sample counts before interpreting them. Missing legacy metadata remains unknown. Duration is still requested time; stream-read elapsed is not proof of target coverage.

Distribution

The MCP server, standalone CLI, Core library and BenchmarkDotNet diagnoser share version 0.27.0. Release assets include NuGet/symbol packages and MCP/CLI self-contained binaries for Linux x64/arm64, Windows x64/arm64 and macOS arm64.

Container images: ghcr.io/pedrosakuma/dotnet-diagnostics:0.27.0 and :0.27.0-lean, each for Linux amd64/arm64.

See the changelog, HTTP destination guide, and networking evidence guide.

Changes: #​963, #​964, #​965, #​966, #​967, #​968, #​970, #​971.

Full comparison: pedrosakuma/dotnet-diagnostics@v0.26.0...v0.27.0

0.26.0

Highlights

v0.26.0 improves diagnostic evidence integrity and adds concurrent GC/activity acquisition in the standalone CLI, alongside explicit OS-backed CoreCLR CPU profiling.

• Concurrent GC/activity capture: dotnet-diagnostics-cli collect --kind gc-activities captures both streams against one target, returns an inline overlay, and provides real handles for session follow-up queries. Independent retention budgets and explicit partial outcomes preserve useful evidence without claiming complete coverage.
• Trace-targeted retention and corrected timing: bounded target-trace budgets survive noisy traffic and distributed collection. Interval union avoids double-counting overlapping child spans or GC intervals.
• Explicit on-CPU profiling: select --cpu-backend os in the CLI or cpuBackend=Os in MCP for Linux perf or Windows kernel ETW. Prerequisites remain explicit; there is no silent fallback from an explicitly selected backend.
• Safer gcdump shutdown: the owned auxiliary reader is cancelled and quiesced before forced pipe disposal, without suppressing unrelated parser failures or publishing incomplete captures as complete.
• More truthful diagnostic quality: ThreadPool, gcdump, CPU, and GC consumers distinguish observed facts, inference, loss, and unavailable evidence. ClrMD is updated to 4.1.745802 with Linux DAC-unload revalidation.

GC migration notes

Collection elapsed time is not application pause. Legacy GcEvent.PauseDuration and GcSummary.TotalPauseTime / MaxPauseTime retain their collection-elapsed meanings.

Use GcSummary.Suspension for version-2 measurements of the fully-suspended GC-related phase, from GCSuspendEEStop to GCRestartEEStart for reasons 1 and 6. This is not the wider suspension envelope or exact per-thread lost execution time.

GC query payloads now use GcMeasurementView. Unavailable overlay pause/overlap values and unassociated GC generation can be nullable; appended record parameters also change generated deconstruction arity.

Corrected comparison metrics are fullySuspendedTimeMs.v2, fullySuspendedPercent.v2, and maxFullySuspendedTimeMs.v2; the signal is gc.fully-suspended-share.v2. Do not reinterpret old collection-elapsed baselines as these measurements.

See the CLI reference, MCP tool reference, and full release notes.

Scope and remaining limitations

The advisory agent harness and human-review infrastructure are included, but calibration remains incomplete under #​921. They are not a model-quality gate or a claim of independently validated diagnostic correctness.

Concurrent aggregate-suite reliability remains tracked in #​879; the separate Linux culture-lookup restriction remains subject to #​929. No detected loss is not a universal completeness guarantee.

Distributions

The shared release version covers dotnet-diagnostics-mcp, dotnet-diagnostics-cli, dotnet-diagnostics-benchmarkdotnet, and dotnet-diagnostics-core.

Self-contained MCP/CLI archives target Linux x64/arm64, Windows x64/arm64, and macOS arm64. Container variants are ghcr.io/pedrosakuma/dotnet-diagnostics:0.26.0 and :0.26.0-lean, each for Linux amd64/arm64.

Full changelog: pedrosakuma/dotnet-diagnostics@v0.25.0...v0.26.0

Commits viewable in compare view.

Updated Microsoft.Azure.Cosmos from 3.63.0 to 3.63.1.

Release notes

Sourced from Microsoft.Azure.Cosmos's releases.

No release notes found for this version range.

Commits viewable in compare view.

Updated Microsoft.NET.Test.Sdk from 18.10.0 to 18.10.1.

Release notes

Sourced from Microsoft.NET.Test.Sdk's releases.

18.10.1

What's Changed

• Isolate code coverage files in published output by @​azat-msft in [rel/18.10] Isolate code coverage files in published output microsoft/vstest#16476

Full Changelog: microsoft/vstest@v18.10.0...v18.10.1

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