Parallel/multi-threaded race condition - can kernel tweaks help?

[OpenSource Ghost]

Assume that Linux-based router with Alpine 4.19 kernel (and shell acces) suffers from parallel/multi-threaded race condition triggered by concurrent user-initiated WireGuard (UDP) bandwidth tests and abrupt termination of connection in the middle of such tests. What can be done aside from switching routers and not running bandwidth tests? Are there any SysCTL tweaks that can help? The goal is not to improve bandwidth, but to obtain stability and reduce chances of race condition occurence. Changing behavior that results in such a condition is not the best solution because bandwidth tests are sometimes necessary.

 

[Nobu]

Might need more context.

Race condition in the program? On the network? In the kernel?

What commands could I execute to reproduce the condition?

Would it be possible to reproduce it with a more simple test command?