fix silent failures in dispatch loop from stalling the pipeline #32922
Conversation
|
Checks are failing. Will not request review until checks are succeeding. If you'd like to override that behavior, comment |
…ker harness to crash the JVM
|
Stopping reviewer notifications for this pull request: review requested by someone other than the bot, ceding control. If you'd like to restart, comment |
| }); | ||
|
|
||
| try { | ||
| dispatchLoopFuture.get(); |
There was a problem hiding this comment.
this is changing start() to be blocking.
If we want that we shoudl have clearer method name and also just get rid of the workProviderExecutor and use this thread directly.
| SingleSourceWorkerHarness.GetWorkSender getWorkSender = | ||
| SingleSourceWorkerHarness.GetWorkSender.forAppliance( | ||
| () -> { | ||
| throw expected; |
There was a problem hiding this comment.
I'm confused why the test of this isn't causing the runtime to halt. It makes me think the unhandled exception handler isn't being called and this is just working because of the changes to make start() blocking and call get on the future().
There was a problem hiding this comment.
added test to ensure that the runtime is being terminated
| private static ThreadFactory terminatingThreadFactory( | ||
| ThreadFactoryBuilder threadFactoryBuilder, Logger logger) { | ||
| return threadFactoryBuilder | ||
| .setUncaughtExceptionHandler(new WorkerUncaughtExceptionHandler(logger)) |
There was a problem hiding this comment.
I'm not sure this is working if the executor thread is catchign the exception and putting it in the future.
I think we might need instead to wrap the executor so that scheduled tasks are wrapped and then catch exceptions and call termination method like com/google/common/util/concurrent/WrappingExecutorService.java
There was a problem hiding this comment.
added to test to ensure that we are halting the JVM
|
back to you thank you! |
| @@ -157,6 +157,19 @@ private static Optional<StreamingConfigTask> fetchConfigWithRetry( | |||
| } | |||
| } | |||
|
|
|||
| private static Optional<ComputationConfig> createComputationConfig(StreamingConfigTask config) { | |||
There was a problem hiding this comment.
can we revert the unrelated change, to keep the diff small and on topic?
| @@ -228,6 +228,22 @@ private static void shutdownExecutor(ScheduledExecutorService executor) { | |||
| } | |||
| } | |||
|
|
|||
| // Calculates the PerWorkerMetrics reporting frequency, ensuring alignment with the | |||
There was a problem hiding this comment.
can we revert the unrelated change, to keep the diff small and on topic?
| @@ -346,10 +348,7 @@ public static StreamingDataflowWorker fromOptions(DataflowWorkerHarnessOptions o | |||
| .setSizeMb(options.getWorkerCacheMb()) | |||
| .setSupportMapViaMultimap(options.isEnableStreamingEngine()) | |||
| .build(); | |||
| Function<String, ScheduledExecutorService> executorSupplier = | |||
| threadName -> | |||
| Executors.newSingleThreadScheduledExecutor( | |||
There was a problem hiding this comment.
I think changing Executors.newSingleThreadScheduledExecutor to Executors.newSingleThreadExecutor should fix the problem? We don't need to introduce a new Executor.
fix silent failures in dispatch loop from stalling the pipeline
changed all executors used by the Streaming harness to use
TerminatingExecutorswhich attaches aWorkerUncaughtExceptionHandlerto all of the thread factories used to create anExecutor/ExecutorService. This means that any uncaught exception will terminate the JVM. This seems to be the original intention inDataflowWorkerHarnessHelper.initializeLogging()since it attaches an uncaught exception handler that will terminate the JVM on uncaught exceptions, however that was only applied to raw threads.added test in
SingleSourceWorkerHarnessto ensure exception is propogated to main thread.WorkerUncaughtExceptionHandleralready has a unit test to ensure that it works.Thank you for your contribution! Follow this checklist to help us incorporate your contribution quickly and easily:
addresses #123), if applicable. This will automatically add a link to the pull request in the issue. If you would like the issue to automatically close on merging the pull request, commentfixes #<ISSUE NUMBER>instead.CHANGES.mdwith noteworthy changes.See the Contributor Guide for more tips on how to make review process smoother.
To check the build health, please visit https://github.com/apache/beam/blob/master/.test-infra/BUILD_STATUS.md
GitHub Actions Tests Status (on master branch)
See CI.md for more information about GitHub Actions CI or the workflows README to see a list of phrases to trigger workflows.