[concurrency-interest] ForkJoinPool deadlock - bug?

Doug Lea dl at cs.oswego.edu
Wed Oct 21 11:44:10 EDT 2020


Thanks for the report. It appears to be a bug. I'll further investigate.

On 10/21/20 10:52 AM, Matthias Schmalz via Concurrency-interest wrote:
> Dear experts,
>
> We are using ForkJoinPool as part of a software project. In our 
> integration tests, we occasionally observe deadlocks with the 
> following characteristics:
> - The ForkJoinPool stops executing new tasks.
> - All worker threads are waiting inside of ForkJoinPool.managedBlock.
> - The number of running threads is 0, the number of active threads is 1.
>
> After increasing the "parallelism" parameter, the problem still 
> occurred. Increasing the "minimumRunnable" parameter from 1 to 2 made 
> the problem disappear. So we have a workaround for the moment.


More information about the Concurrency-interest mailing list