Shared object error running Rails 7 with concurrent requests

[headius]

I run into this occasionally attempting to run Rails 7 with Puma on TruffleRuby 22.3. It kills the server.

Caused by: java.lang.UnsupportedOperationException: Thread-safety bug: the object is already shared. This means another thread marked the object as shared concurrently.
	at org.truffleruby.language.objects.shared.ShareObjectNode.createSharedShape(ShareObjectNode.java:129)
	at org.truffleruby.language.objects.shared.ShareObjectNodeGen.executeAndSpecialize(ShareObjectNodeGen.java:99)
	at org.truffleruby.language.objects.shared.ShareObjectNodeGen.executeShare(ShareObjectNodeGen.java:69)
	at org.truffleruby.language.objects.shared.WriteBarrierNode.writeBarrierCached(WriteBarrierNode.java:69)
	at org.truffleruby.language.objects.shared.WriteBarrierNodeGen.executeWriteBarrier(WriteBarrierNodeGen.java:80)
	at org.truffleruby.core.array.library.ObjectArrayStore$ShareElements.shareElements(ObjectArrayStore.java:102)
	at org.truffleruby.core.array.library.ObjectArrayStoreGen$ArrayStoreLibraryExports$Cached.shareElements(ObjectArrayStoreGen.java:90)
	at org.truffleruby.core.array.library.ObjectArrayStore.makeShared(ObjectArrayStore.java:87)
	at org.truffleruby.core.array.library.ObjectArrayStoreGen$ArrayStoreLibraryExports$Cached.makeShared(ObjectArrayStoreGen.java:333)
	at org.truffleruby.language.objects.shared.ShareInternalFieldsNode.shareArray(ShareInternalFieldsNode.java:50)
	at org.truffleruby.language.objects.shared.ShareInternalFieldsNodeGen.executeShare(ShareInternalFieldsNodeGen.java:46)
	at org.truffleruby.language.objects.shared.ShareObjectNode.shareCached(ShareObjectNode.java:74)
	at org.truffleruby.language.objects.shared.ShareObjectNodeGen.executeShare(ShareObjectNodeGen.java:51)
	at org.truffleruby.language.objects.shared.WriteBarrierNode.writeBarrierCached(WriteBarrierNode.java:69)
	at org.truffleruby.language.objects.shared.WriteBarrierNodeGen.executeWriteBarrier(WriteBarrierNodeGen.java:80)
	at org.truffleruby.language.objects.shared.PropagateSharingNode.propagateShared(PropagateSharingNode.java:39)
	at org.truffleruby.language.objects.shared.PropagateSharingNodeGen.executePropagate(PropagateSharingNodeGen.java:39)
	at org.truffleruby.core.hash.library.PackedHashStoreLibrary$Set.setFirst(PackedHashStoreLibrary.java:180)
	at org.truffleruby.core.hash.library.PackedHashStoreLibraryGen$HashStoreLibraryExports$Cached.set(PackedHashStoreLibraryGen.java:102)
	at org.truffleruby.core.hash.library.HashStoreLibraryGen$CachedDispatch.set(HashStoreLibraryGen.java:999)
	at org.truffleruby.core.hash.library.EmptyHashStore.set(EmptyHashStore.java:49)
	at org.truffleruby.core.hash.library.EmptyHashStoreGen$HashStoreLibraryExports$Cached.set(EmptyHashStoreGen.java:86)
	at org.truffleruby.core.hash.HashNodes$SetIndexNode.set(HashNodes.java:239)
	at org.truffleruby.core.hash.HashNodesFactory$SetIndexNodeFactory$SetIndexNodeGen.execute(HashNodesFactory.java:981)
	at org.truffleruby.language.RubyCoreMethodRootNode.execute(RubyCoreMethodRootNode.java:48)

The app is a trivial generated "blog" scaffold app with Puma configured for 32 threads and siege doing 32 connections. If I drive the app with a few requests before starting benchmarking, it seems to fail less often.

[andrykonchin]

Thank you for the report!

[eregon]

@headius Could you share the application?
That would make it a lot easier to reproduce, e.g., we don't have to guess which scaffolding commands were used, database, etc.
Besides it would be interesting to reproduce your benchmark numbers.

[headius]

The application is the same setup as in #2804, but running on Linux on i7 in native mode.

[jedld]

I am encountering the same error as well with 1000 concurrent request. Even used the docker image:

FROM ghcr.io/graalvm/truffleruby:debian-22.3
RUN apt-get update && apt-get --fix-broken install && apt-get update --fix-missing \
    && apt-get install -y wget \
        nginx \
        mariadb-client \
        vim \
        nano \
        procps \
        nodejs \
        make \
        git \
        build-essential \
        curl \
        libcurl4 \
        libcurl4-openssl-dev \
        default-libmysqlclient-dev
RUN mkdir -p  /usr/src/app/
WORKDIR /usr/src/app/
COPY . /usr/src/app/
RUN bundle
ENV RAILS_ENV=production
CMD rails s -p 8080 -b 0.0.0.0

below is the error encountered:

Puma starting in single mode...
* Puma version: 5.6.5 (truffleruby 22.3.0 - ruby 3.0.3) ("Birdie's Version")
*  Min threads: 75
*  Max threads: 75
*  Environment: production
*          PID: 1
* Listening on http://0.0.0.0:8080
Use Ctrl-C to stop
java.lang.RuntimeException: Ruby Thread id=93 from /opt/truffleruby-22.3.0/lib/gems/gems/puma-5.6.5/lib/puma/thread_pool.rb:104 terminated with internal error:
        at org.truffleruby.core.thread.ThreadManager.printInternalError(ThreadManager.java:305)
        at org.truffleruby.core.thread.ThreadManager.threadMain(ThreadManager.java:293)
        at org.truffleruby.core.thread.ThreadManager.lambda$initialize$2(ThreadManager.java:263)
        at java.base@17.0.5/java.lang.Thread.run(Thread.java:833)
        at org.graalvm.truffle/com.oracle.truffle.polyglot.PolyglotThread.access$001(PolyglotThread.java:53)
        at org.graalvm.truffle/com.oracle.truffle.polyglot.PolyglotThread$1.execute(PolyglotThread.java:100)
        at org.graalvm.truffle/com.oracle.truffle.polyglot.PolyglotThread$ThreadSpawnRootNode.executeImpl(PolyglotThread.java:134)
        at org.graalvm.truffle/com.oracle.truffle.polyglot.PolyglotThread$ThreadSpawnRootNode.execute(PolyglotThread.java:125)
        at jdk.internal.vm.compiler/org.graalvm.compiler.truffle.runtime.OptimizedCallTarget.executeRootNode(OptimizedCallTarget.java:709)
        at jdk.internal.vm.compiler/org.graalvm.compiler.truffle.runtime.OptimizedCallTarget.profiledPERoot(OptimizedCallTarget.java:632)
        at jdk.internal.vm.compiler/org.graalvm.compiler.truffle.runtime.OptimizedCallTarget.callBoundary(OptimizedCallTarget.java:565)
        at com.oracle.svm.truffle.api.SubstrateOptimizedCallTarget.invokeCallBoundary(SubstrateOptimizedCallTarget.java:115)
        at com.oracle.svm.truffle.api.SubstrateOptimizedCallTargetInstalledCode.doInvoke(SubstrateOptimizedCallTargetInstalledCode.java:194)
        at com.oracle.svm.truffle.api.SubstrateOptimizedCallTarget.doInvoke(SubstrateOptimizedCallTarget.java:97)
        at jdk.internal.vm.compiler/org.graalvm.compiler.truffle.runtime.OptimizedCallTarget.callIndirect(OptimizedCallTarget.java:477)
        at jdk.internal.vm.compiler/org.graalvm.compiler.truffle.runtime.OptimizedCallTarget.call(OptimizedCallTarget.java:458)
        at org.graalvm.truffle/com.oracle.truffle.polyglot.PolyglotThread.run(PolyglotThread.java:96)
        at com.oracle.svm.core.thread.PlatformThreads.threadStartRoutine(PlatformThreads.java:775)
        at com.oracle.svm.core.posix.thread.PosixPlatformThreads.pthreadStartRoutine(PosixPlatformThreads.java:203)
Caused by: java.lang.UnsupportedOperationException: Thread-safety bug: the object is already shared. This means another thread marked the object as shared concurrently.