FirebirdSQL
Performing QA test for core-4236 can hang if concurrent FB instance is active
Ticket has been created after talk with Alex. Two FB snapshots LI-6.0.0.336 were deployed on Linux-QA test machine ("Debian 6.5.13-1 (2023-11-29) x86_64"), with different values of RemoteServicePort and FIREBIRD_TMP in envirionment. Value of FIREBIRD_LOCK was not set (i.e. remained default).
Then two QA sessions started (at the same time, i.e. in parallel), each against appropriate FB instance: pytest ... --server <FB_instance_1> pytest ... --server <FB_instance_2>
Both FB instances had to pass two ServerMode values: 1) SS; 2) CS. Both of them initially worked fine, w/o any problem (i.e. they took subsequent *_test.py, run it and display output). But then one of instances hanged up on test for core-4236 ("Database shutdown is reported as successfully completed before all active connections are in fact interrupted"), checked ServerMode was 'SS'.
Problem with this test raised also when FB 6.x and 5.x have been checked (5.x has been affected at that time). All dumps, stack traces and lock-print logs have beeen sent to Alex privately.
