MariaDB Restart on Linux

To mariadb restart on a modern Linux system, run sudo systemctl restart mariadb. If the service is called mysql or mysqld on your distribution, swap the name. The legacy service mariadb restart wrapper still works — it forwards to systemctl. This guide covers every form, when to reload instead of restart, and how to verify the service came back up.

Contents

TL;DR
Pick the right service name
Restart vs reload
Verify it’s back up
When a restart isn’t enough
Frequently asked questions
Related guides
References

Last verified: 2026-04-23 on Ubuntu 22.04, AlmaLinux 9, and CentOS 7 (EOL). Originally published 2022-12-31, rewritten and updated 2026-04-23.

TL;DR

# Modern (systemd)
sudo systemctl restart mariadb

# If the service is named differently
sudo systemctl restart mariadb.service
sudo systemctl restart mysql
sudo systemctl restart mysqld

# Legacy wrapper (still works, forwards to systemctl)
sudo service mariadb restart
sudo service mysql restart
sudo service mysqld restart

Pick the right service name

Ubuntu / Debian with MariaDB package — mariadb.
Ubuntu / Debian with MySQL package — mysql.
CentOS / RHEL / AlmaLinux / Rocky — mariadb (or mysqld on older installs).
Amazon Linux 2 — mariadb.

Not sure which name to use? List all three:

systemctl list-unit-files | grep -E 'mariadb|mysql'

The one marked enabled is the service that starts on boot — that’s the one to restart.

mariadb restart — systemctl commands across Linux distributions

Restart vs reload

# Full stop + start — drops connections, ~seconds downtime
sudo systemctl restart mariadb

# Keep connections alive, re-read config — faster, no downtime
sudo systemctl reload mariadb

reload is faster and keeps existing connections alive but only works for config changes that MariaDB can apply live. Most system variables do work; changing the InnoDB buffer pool size, loading/unloading a plugin, or swapping the data directory all require restart. When in doubt, check SHOW VARIABLES LIKE '...' after a reload to confirm the new value took effect — if it didn’t, go back and restart.

Verify it’s back up

# Service-level check
sudo systemctl status mariadb

# Expect "Active: active (running)" in green, no failed units.

# Connection-level check
mysql -u root -p -e 'SELECT VERSION();'

Both checks are quick. If systemctl says active but mysql -u root -p hangs, the server is listening but not accepting queries — usually a data-directory corruption that needs the log: journalctl -u mariadb -n 100 --no-pager.

When a restart isn’t enough

If systemctl restart returns quickly but MariaDB is still unavailable, you’re into troubleshooting territory — config syntax errors, disk full, orphaned PID files, port conflicts. See troubleshooting a MariaDB that won’t start for the triage checklist.

Frequently asked questions

What’s the shortest mariadb restart command?

sudo systemctl restart mariadb. That’s the canonical form on every distribution using systemd (Ubuntu 16.04+, Debian 8+, CentOS 7+, RHEL 7+, AlmaLinux, Rocky). The legacy service mariadb restart wrapper also works and just forwards to systemctl on modern systems.

Why is the service sometimes called mysql or mysqld?

MariaDB shipped as a drop-in MySQL replacement, and some distributions kept the original service names for compatibility. On CentOS/RHEL the service is called mariadb; on some Debian/Ubuntu packages it’s mysql; on older Red Hat derivatives you’ll find mysqld. systemctl status mariadb mysql mysqld — whichever returns ‘loaded’ is the one running.

Should I use restart or reload?

restart fully stops and starts the service — every connection is dropped. reload is lighter: it signals the running process to re-read its config without killing connections. Reload works for most my.cnf changes that don’t require a fresh server instance. Restart is necessary for changes to InnoDB buffer pool size, plugin loading, or anything that needs the engine re-initialized.

How do I check if the restart worked?

sudo systemctl status mariadb — the output shows ‘active (running)’ in green if it’s up. The log excerpt underneath (or journalctl -u mariadb -n 50) reveals any startup errors. For a connection-level check, mysql -u root -p -e 'SELECT VERSION();' confirms the server is accepting queries.

What if systemctl restart fails?

Check the log with journalctl -u mariadb -n 100 --no-pager. Common causes: syntax error in /etc/mysql/my.cnf (typos, missing quotes), disk full (df -h), orphaned PID file (/var/run/mysqld/mysqld.pid), or a port conflict. See troubleshooting MariaDB not starting for the full triage.