Database crashes when its semaphores get removed.
Whenever any command is run against the database using as the same user account that started the databases, semaphores get removed when the user logs off and the database crashes.
The database crashes with semaphore errors every time a user account that runs a database administrative task cron session finishes
The database is started with a
user account where the corresponding semaphores are owned by this user:
$ sudo -u sysadmin proserve dbname
$ ipcs -s
------ Semaphore Arrays --------
key semid owner perms nsems
0x00000000 983040 sysadmin 666 7
0x00000000 1015809 sysadmin 666 39
0x00000000 1048578 sysadmin 666 39
When someone logs on to the system with this user account and connects to the database, for example: using ssh or sudo
$ proshut dbname -C list
When that user logs off from the system, semaphore errors occur and the database crashes
/var/log/messages reports the following:
systemd: Created slice user-1000.slice.
systemd: Starting user-1000.slice.
systemd: Started Session 759 of user sysadmin.
systemd-logind: New session 759 of user sysadmin.
systemd: Starting Session 759 of user sysadmin.
systemd-logind: Removed session 759.
systemd: Removed slice user-1000.slice.
systemd: Stopping user-1000.slice.