Plesk is stable when it works, but when a site shows 502 Bad Gateway, a panel login hangs, or the whole control panel refuses to load, the number of moving parts (nginx, Apache, PHP-FPM, MariaDB, the panel’s own services) makes it hard to know where to start. The good news is that most failures fall into a handful of patterns, and Plesk ships with a powerful command line repair tool. This Plesk troubleshooting guide gives you a repeatable method, then covers the specific errors you will meet most often.

Understand the Plesk Request Path

On a default Linux Plesk server, a website request travels through several layers, and a failure at any one produces a different error:

Layer	Service name	Fails with
Front-end proxy	nginx	Connection refused, wrong certificate
Back-end web server	apache2 (Debian/Ubuntu) or httpd (RHEL)	502 from nginx when Apache is down
PHP handler	plesk-php83-fpm (version varies)	502, 503 or 504, blank page
Database	mariadb or mysql	“Error establishing a database connection”
Panel itself	sw-engine, psa	Cannot log in to port 8443
Mail	postfix, dovecot	Delivery and login failures

Always start by establishing which layer is broken before changing anything.

Plesk Troubleshooting Step 1: Check the Services

# Status of all Plesk-managed services
plesk bin service --status

# Individual checks
systemctl status nginx
systemctl status sw-engine
systemctl status mariadb
systemctl list-units --type=service --state=failed

Restart a failed service and read why it failed:

systemctl restart nginx
journalctl -u nginx --since "30 minutes ago" --no-pager | tail -50

Step 2: Run Plesk Repair

The plesk repair utility checks and fixes the most common problems automatically. Take a snapshot or backup first, then run:

# Repair everything (interactive; answer y to apply fixes)
plesk repair all -y

# Targeted repairs
plesk repair web -y            # web server configs, vhosts
plesk repair db -y             # Plesk database consistency
plesk repair mail -y           # mail server configuration
plesk repair installation -y   # missing files, wrong permissions, packages
plesk repair dns -y            # DNS zones

plesk repair web is the one that solves a large share of 502 and 404 problems, because it regenerates Apache and nginx configuration for every domain. If a single domain is broken, regenerate just that one:

plesk repair web -domain example.com -y

Fixing 502 Bad Gateway

A 502 means nginx could not get a valid response from the upstream (Apache or PHP-FPM). Work through these in order:

1. Is Apache running? systemctl status apache2 (or httpd). If it crashed, check journalctl and /var/log/apache2/error.log.
2. Is PHP-FPM for that domain’s PHP version running? Run systemctl list-units | grep plesk-php. Restart the specific one, for example systemctl restart plesk-php83-fpm.
3. Did the PHP-FPM pool run out of workers? Look for server reached pm.max_children in /var/log/plesk-phpfpm/error.log. Raise it under Domains » example.com » PHP Settings » Performance settings.
4. Out of memory? The kernel may have killed Apache or PHP. Search with dmesg -T | grep -i "killed process". See our guide to fixing high CPU and memory usage on a Linux VPS.
5. Broken configuration after a change? Test with nginx -t and apachectl configtest, then plesk repair web.

Fixing 504 Gateway Timeout

A 504 means the upstream was reachable but too slow. The usual causes are a long-running script, a slow database query or an overloaded server. In Plesk, increase the timeouts for the domain under Apache & nginx Settings:

# Additional nginx directives
proxy_read_timeout 300;
proxy_connect_timeout 60;
fastcgi_read_timeout 300;

Raising a timeout hides the symptom, so also find the slow request. Enable the PHP-FPM slow log in the domain’s PHP settings or check the MariaDB slow query log. Our MySQL and MariaDB performance tuning guide covers this.

Fixing “Unable to Connect to Database Server”

This error in the Plesk login page almost always means the Plesk psa database or MariaDB is down.

systemctl status mariadb
plesk db "SELECT 1"
tail -50 /var/log/mariadb/mariadb.log

Common causes are a full disk (df -h), an InnoDB crash after power loss, or a corrupted psa database. Do not delete InnoDB log files blindly. If the disk is full, free space first, restart MariaDB and then run plesk repair db. For corrupted tables, restore only from a verified backup rather than experimenting on the live psa database.

Panel Not Loading on Port 8443

If sites work but the panel does not:

systemctl restart sw-engine sw-cp-server
ss -tlnp | grep 8443
tail -30 /var/log/sw-cp-server/error_log
tail -30 /var/log/plesk/panel.log

Also check that your firewall allows 8443, and that you have not been banned by Fail2Ban after failed logins. Check with plesk bin ip_ban --list and unban an address with plesk bin ip_ban --unban 203.0.113.10.

Where the Important Logs Live

Problem area	Log file
Panel	/var/log/plesk/panel.log, /var/log/sw-cp-server/error_log
Domain web errors	/var/www/vhosts/example.com/logs/error_log
Apache global	/var/log/apache2/error.log or /var/log/httpd/error_log
nginx	/var/log/nginx/error.log
PHP-FPM	/var/log/plesk-phpfpm/
Mail	/var/log/maillog or /var/log/mail.log
Updates	/var/log/plesk/install/
Repair history	/var/log/plesk/repair.log

Fixing Permissions and Ownership

Wrong ownership after a manual migration causes 403 and 500 errors. Plesk can reset it for you:

plesk repair fs -y
# or one domain only
plesk repair web -domain example.com -y

Avoid running chmod -R 777. It does not fix anything permanently and opens a security hole. For background see Linux file permissions explained.

A Quick Troubleshooting Checklist

1. Check disk (df -h, df -i) and memory (free -m)
2. List failed services (systemctl --failed)
3. Read the relevant log for the last 30 minutes
4. Run plesk repair web or plesk repair all
5. Reproduce on a single domain to separate site problems from server problems
6. Only then restart services or reboot

If you are planning to move from another panel, see our cPanel to Plesk migration guide, and for a fresh install, How to Install and Secure Plesk Obsidian.

Frequently Asked Questions

Is it safe to run plesk repair all?

Yes, it is designed to be safe and asks before each change unless you use -y. Still, take a backup or snapshot first on production servers.

Why does nginx show 502 only for one domain?

A single-domain 502 usually means that domain’s PHP-FPM pool is stopped, misconfigured or out of workers. Check its PHP version and handler, then run plesk repair web -domain for that domain.

How do I restart Plesk completely?

Use plesk restart or systemctl restart psa. This restarts all Plesk services without rebooting the server.

Does Plesk have a built-in log viewer?

Yes. Domains » example.com » Logs shows access and error logs for each site, and Tools & Settings » Logs covers the panel and server components.