Race condition in SmartThreadPool causes not processing tasks

[jeernest]

Hi,

I found an issue in using groups and I would like to check if it is known issue or not.

steps to reproduce:

1. Start a ThreadPool with a WorkGroup - set MinThreads = 0, MaxThreads =1, IdleTimeout = 60 seconds
2. Send a task to WorkItemGroup every 60 seconds
3. after some time, tasks stop to process

Here is what happens:
Thread#1 [STP Worker thread] timeouts, because it has empty queue for 60 seconds and starts terminating (there's still some work to do after the timeout check and before the thread is freed)
In meantime
Thread#2 [new thread] enqueues a task to the WorkItemGroup,
Thread#2 WorkItemGroup enqueues a task to SmartThreadPool expecting that STP withdraw the task,
Thread#2 STP enqueues a task and doesn't start a thread, because there's one already running (it doesn't know that it's terminating)
Thread#1 [STP Worker thread] finishes termination and removes itself from STP
Thread#3 [new thread] [Receiver] enqueues a task to Work Items Group
Thread#3 WorkItemGroup doesn't do anything, because it thinks that STP has already enough tasks in queue

since then the STP has no running thread and all new tasks are enqueued to the WorkItemGroup and the queue is growing and never processed

[MichelZ]

Have you found a solution for this? I think I am experiencing something similar

[jeernest]

Unfortuntely, no. We stopped using work item groups and everything started to work.

[jirkapok]

Which versions are affected by this issue?

[jeernest]

Which versions are affected by this issue?

As far as I remember it was 2.0