I think I will need more lines from the end.
I would expect the reason, why he is not continuing after starting the upgrade.
The last entry from the start command showed the “mysql_fix_privilege_tables” phase.
I would not expect the database to need several minutes for that phase.
There are round about 1 hour difference between the output of systemd and the mysqld.log. Are they from the same start operation ?
The mysql socket error may be misleading.
But you should check, whether the path is available and the user who is running mariadb has the appropiate privileges.
On my openSUSE system I do have a file for the running mariadb instance
srwxrwxrwx 1 mysql mysql 0 Jun 17 17:00 mysql.sock
in directory /run/:
drwxr-xr-x 2 mysql mysql 60 Jun 17 17:00 mysql
It will be created in the start phase.
Do you have a backup of your database ?
If the instance is stopped you can also just tar the files.
In /var/log/messages we may find also some information.
If the start is not working via systemctl, we can also start the instance manually and then execute mariadb_upgrade manually (you should have created a backup of the my.cnf and the data directory / database directory!).
So first systemctl stop mariadb
Then start mysql as user mysql (mysqld_safe --defaults-file=/my.cnf --datadir=<path where to find the data)
Then execute mariadb_uprade
But I would recommend first looking for more information in the existing logs.
systemd is doing this automatically for you. - If it is working.