Reproducible hang on Debian 9, oracle-java-13 and above

[geeohgeegeeoh]

For the current code, I can reproduce a hang on oracle-java-13, oracle-java-14 and oracle-java-15.

oracle-java-11 is fine.

On FreeBSD 12, more recent OpenJDK to equivalent levels are fine btw.

I can give you a jstack thread dump if you want.

[lolepezy]

I can give you a jstack thread dump if you want.

That would be very helpful! Especially for jdk 14.

[geeohgeegeeoh]

Here's a jthread trace
java-threads.txt

[ties]

Hi,

Two quick questions:

• On what kind of machine (distribution/number of CPU cores) do you encounter this issue?
• How long after startup do you encounter this?

This will help us when attempting to reproduce it :)

[geeohgeegeeoh]

Its a 48 core Dell running Debian, with around 250GB of memory. It happens within 5min. From clean start, attempting to bootstrap state. Sometimes it can pass through, but usually it gets stuck.

…

On Mon, Sep 28, 2020 at 11:18 PM Ties de Kock ***@***.***> wrote: Hi, Two quick questions: - On what kind of machine (distribution/number of CPU cores) do you encounter this issue? - How long after startup do you encounter this? This will help us when attempting to reproduce it :) — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#277 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABORQ7AWZ3727GD4NWL3LRTSICEKRANCNFSM4RVD6AJQ> .

[ties]

I can reproduce this on a machine with a high number of cores (R730xd/28 cores/56 hyperthreads/openjdk 14). It seems like it gets stuck with the high amount of threads created and this is a symptom of an underlying bug.

I could work around it by limiting the number of threads (by setting _JAVA_OPTIONS="-XX:ActiveProcessorCount=16" in an environment variable). Please let me know if that is a suitable workaround for you for now.

@erikrozendaal could you take a look?