MariaDB will not start, or the site is up and the database is not.
A full disk, a crashed table, an InnoDB log, a connection limit. We read the error log before deleting ib_logfile. The data stays on your server.
Disk first. Then the log. Then the table.
Disk at 100%, a corrupted InnoDB redo log, or an upgrade that changed the data format. Free space comes before any repair flag. A dump from yesterday is the rollback.
PHP-FPM workers above what MariaDB will accept. Raising max_connections on a 2 GB VPS is how it gets killed. We match the pool to the RAM.
The slow log names it. An index is added only if the table is yours to change. We do not rewrite the application around it.
Oracle MySQL is the same class of job. The package name is not.
Hestia and DirectAdmin usually install MariaDB. A server that is actually MySQL stays on MySQL. We do not convert one to the other in the middle of an outage. A dropped table is a restore from the last dump, not a repair of an empty file.
€55 an hour. Database down is emergency, if someone is free.
Minimum 1 hour. Emergency €95, subject to availability. A major version upgrade is agreed after a dump, not started because the panel offered it.
MariaDB refused connections, or will not start?
Tell us the panel and whether the disk is full. No database password in the first message.