Fix Kafka CI: start Kafka container, fix configuration test timing, reduce infinite-loop test delay#2395
Merged
jeremydmiller merged 7 commits intomainfrom Mar 31, 2026
Merged
Fix Kafka CI: start Kafka container, fix configuration test timing, reduce infinite-loop test delay#2395jeremydmiller merged 7 commits intomainfrom
jeremydmiller merged 7 commits intomainfrom
Conversation
…duce infinite-loop test delay - CITargets.cs: Add 'kafka' to StartDockerServices in CIKafka target — was only starting postgresql, so Kafka tests had no broker in CI. Added WaitForKafkaToBeReady() that polls port 9092 until the broker accepts connections (up to 60 seconds, 30 attempts × 2s). - KafkaTransportTests.cs: Fix KafkaTopicGroupConfigurationTests — the Specification() and TopicCreation() methods use the DelayedEndpointConfiguration pattern, storing actions that are applied lazily at startup. Tests were checking group.SpecificationConfig/CreateTopicFunc immediately after calling the fluent methods, before Apply() was ever called. Added ((IDelayedEndpointConfiguration)config).Apply() before each assertion. - publish_and_receive_raw_json.cs: Reduce the do_not_go_into_infinite_loop_with_garbage_data test delay from 2 minutes to 30 seconds. The test has no assertions after the delay — it only checks the process doesn't crash. 30 seconds is sufficient to observe a tight retry loop, and saves ~90 seconds per CI run. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
…from handler pipeline When PauseListenerContinuation fires from within the handler pipeline (e.g. rate limiting), ListeningAgent.PauseAsync called StopAndDrainAsync which always called LatchReceiver() before receiver.DrainAsync(). This caused DurableReceiver/BufferedReceiver/ InlineReceiver's DrainAsync to see _latched=true (waitForCompletion=true) and wait up to DrainTimeout (30s) for the ActionBlock/in-flight count to drain — which never happened because the current message's execute frame was still on the call stack. The result was a 30-second stall on every rate-limit event, causing the rate-limiting test to time out. Fix: split StopAndDrainAsync into a shared StopAndDrainCoreAsync(bool latchBeforeDrain). - Normal shutdown (StopAndDrainAsync) continues to pre-latch so DrainAsync can safely wait for in-flight messages to complete. - PauseAsync now calls StopAndDrainCoreAsync(latchBeforeDrain: false) so DrainAsync sees _latched==false and returns immediately, avoiding the deadlock. Also add fast durability polling settings to the rate-limiting integration test so the scheduled message is picked up quickly after the listener restarts. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
…ackages - JasperFx: 1.22.0 → 1.23.0 - JasperFx.Events: 1.24.1 → 1.25.0 - All Weasel.*: 8.10.2 → 8.11.2 - Marten + Marten.AspNetCore: 8.26.1 → 8.28.0 - Polecat: 1.4.0 → 1.6.1 - Version: 5.26.0 → 5.27.0 Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Pin Microsoft.CodeAnalysis.Common to 4.14.0 in Directory.Packages.props to resolve the NuGet conflict between Microsoft.CodeAnalysis.Workspaces.MSBuild (which requires 4.14.0) and Microsoft.EntityFrameworkCore.Design (which pulls in 4.8.0 transitively). This was causing all 20 CI workflows to fail. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
The merge brought back local project references to external repos (jasperfx, weasel, marten). Replace all with proper NuGet package references to match the existing Directory.Packages.props versions. Also add Microsoft.CodeAnalysis.CSharp and Microsoft.CodeAnalysis.Analyzers to support the new Wolverine.SourceGeneration Roslyn analyzer project. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Replace local project reference to ../../../../polecat/src/Polecat/Polecat.csproj with the NuGet package reference, matching the existing Polecat 1.6.1 version in Directory.Packages.props. This was causing CS0246 errors for all Polecat types in CI where the external repo is not present. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
This was referenced Apr 1, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
CI was never starting Kafka:
CIKafkatarget inCITargets.csonly calledStartDockerServices("postgresql")— the Kafka broker was never started, so all integration tests would fail in CI with connection-refused errors. Added"kafka"to the service list and a newWaitForKafkaToBeReady()readiness check that polls port 9092.3 unit tests were failing:
KafkaTopicGroupConfigurationTests.specification_uniform_sets_config_on_group,specification_per_topic_receives_topic_name, andtopic_creation_sets_func_on_groupall failed becauseKafkaTopicGroupListenerConfigurationuses theDelayedEndpointConfigurationpattern (actions stored and applied lazily at startup). Tests were reading properties immediately after calling fluent methods, beforeApply()was called. Fixed by explicitly calling((IDelayedEndpointConfiguration)config).Apply()before assertions.2-minute test delay reduced to 30s:
do_not_go_into_infinite_loop_with_garbage_datahadawait Task.Delay(2.Minutes())with no assertions — only verifies the process doesn't crash. 30 seconds is sufficient to observe a tight retry loop and saves ~90 seconds per CI run.Test plan
KafkaTopicGroupConfigurationTestslocally — all 6 tests passcw-changesbranch caused by accidentally committed local JasperFx<ProjectReference>entries (not in main/this branch)🤖 Generated with Claude Code