stack info

stack info

A large number of threads are stuck on read-write locks, preventing the processing of HTTP requests. The state of the lock is incorrect; normally, there shouldn’t be such a lock state. First reader = XNIO task-5, first reader hold counter = 2, state = 131072. Other threads’ ThreadLocal records show hold counters that are either 1 or 2. The AQS linked list structure is as follows: XNIO task-5 (waiting for write lock, holding read lock!! It’s very strange, I don’t know where this read lock came from, could it be a release error?) -> XNIO task-3 (waiting for write lock) -> XNIO task-6 (waiting for read lock) -> XNIO task-16 (waiting for read lock).

锁状态dump

There is an error in the middle, but the actual error is concealed, and it is suspected that this is the cause.
error log

help， thanks!