We've recently switched our production server to FB 4.0.2 from FB 3.0.7, both as SuperServer.

Current OS is Debian 11 with 4 CPU and 24GB of RAM running on VMWare.
Previous OS was CentOS 7, same HW config.

We have about 20-25 DBs accessed through this server by an average of 30-40 concurrent remote connections via TCP/IP (the clients are Delphi applications using FireDAC).
Some of these connections are "steady" (the application connects to the DB and stays there until it's closed), some others are short lived ones from a stateless application server, so maybe 15-25 connections are steady ones, the other 10-15 are short lived ones.

With FB 3 we had absolutely no problems, everything worked fine for years.

Now with FB 4.0.2 we're experiencing some serious problems (engine crashes), luckily not so often, but when it happes all the connections and active transactions obviously are gone.

In a week we had 5 of these events and they're almost all different, but I suspect there's a common cause to them.

This is what I see in the firebird.log (I kept only the error lines):

Fatal lock manager error: invalid lock id (94352), errno: 0
Operating system call pthread_mutex_destroy failed. Error code 16
Fatal lock manager error: Process disappeared in LockManager::acquire_shmem, errno: 0
Operating system call pthread_mutex_destroy failed. Error code 16
Operating system call pthread_mutex_lock failed. Error code 22

All these errors led to a restart of the engine by the guardian:

/opt/firebird/bin/fbguard: /opt/firebird/bin/firebird terminated abnormally (-1)
/opt/firebird/bin/fbguard: guardian starting /opt/firebird/bin/firebird

Have you got an idea what could be the cause?

Unfortunately we've changed both FB version and OS, so I can't say which of them is the root of these problems... Maybe there's some config I can change to mitigate or eliminate these problems, or maybe it's a FB bug?

Thanks!