How to Update a CentOS Server

This guide explains how to update the packages on your CentOS server, how to tell whether a reboot is needed and how to have updates installed automatically. If your server runs CentOS Linux 6, 7 or 8, read the section on versions that are no longer supported first.

Which version are you using?

cat /etc/redhat-release

The output looks like CentOS Stream release 9 or CentOS Linux release 7.9.2009 (Core).

Support status of the versions

CentOS Stream is the version where the next minor release of Red Hat Enterprise Linux is developed; it receives updates continuously.

Before you start

• Run the commands as root or with sudo.
• Take a backup. Backups are the customer's responsibility; for your server you can use our backup services or take a copy of the whole server with the System Copy add-on.
• Choose a quiet time. Some services restart during the update, and a kernel update needs a reboot.
• If you lose your SSH connection, you can still reach a VDS server from the console in the client area: Using the Console on a VDS Server.

Installing updates

1. See which updates are available:

   dnf check-update

2. Install them:

   dnf upgrade

   The command shows what it will do and asks for confirmation; press y and Enter. (dnf update is an older name for the same command.)

Security updates cannot be installed on their own on CentOS: the CentOS repositories do not publish security advisory data, so dnf upgrade --security reports that there is nothing to install even when there is. Always install all updates with dnf upgrade.

Is a reboot needed?

A new kernel or core libraries (such as glibc or systemd) only take effect after a reboot. Check with:

dnf needs-restarting -r

The command says whether a reboot is needed. If it is not found, install it with dnf install dnf-plugins-core. To list the services still using old files: dnf needs-restarting -s. To reboot the server: reboot. You can also use Restart in the client area.

Installing updates automatically

dnf-automatic downloads and installs updates on a schedule. Install it:

dnf install dnf-automatic

Set these lines in the [commands] section of /etc/dnf/automatic.conf:

[commands]
upgrade_type = default
apply_updates = yes

Leave upgrade_type = default: these repositories do not publish security advisories, so the security setting would install nothing.

Turn on the timer:

systemctl enable --now dnf-automatic.timer

By default the server is not rebooted. If you want, reboot = when-needed in the same section reboots it when an update needs it.

Versions that are no longer supported

CentOS Linux 6, 7 and 8 and CentOS Stream 8 no longer receive any updates, including security fixes. Their packages have been moved to the vault.centos.org archive, so yum update / dnf upgrade fails on these servers with errors such as "Cannot find a valid baseurl" (CentOS 7) or "Failed to download metadata" (CentOS 8). The archive lets you install packages, but no new fixes arrive there.

On CentOS 7 the commands are yum check-update and yum update; whether a reboot is needed is checked with needs-restarting -r (yum-utils package), and automatic updates are done with yum-cron (/etc/yum/yum-cron.conf, apply_updates = yes). Since no updates are published any more, these tools will not bring new fixes either.

For a server that stays on the internet, the lasting solution is to move to a supported system. AlmaLinux and Rocky Linux are close relatives of CentOS; their version 10 is supported until 2035. If you would like your server reinstalled with a current system, you can open a support ticket from your client area.

Troubleshooting

• "Waiting for process with pid … to finish": another dnf process is running (often the automatic update). Wait for it to finish.
• "Cannot download repomd.xml" or "Failed to download metadata": check the server's internet and DNS settings, then try again with dnf clean all and dnf upgrade.
• The server does not come back after a reboot: open the console in the client area and check the screen (Using the Console on a VDS Server). The boot menu usually still has the previous kernel. If you cannot solve it, you can open a support ticket from your client area.

The information in this guide is taken from official documentation: CentOS Linux EOL, CentOS Linux vs CentOS Stream, CentOS Stream 10, CentOS Vault, dnf command reference, dnf needs-restarting, dnf-automatic, needs-restarting(1), yum-cron.conf.