Centos 7 To Rocky Linux

Seamless Migration: Practical Guide to Transitioning from CentOS 7 to Rocky Linux

After CentOS 7’s EOL (June 2024), system administrators face a forced upgrade path. Rocky Linux, a 1:1 RHEL-compatible rebuild, covers the gap without introducing unpredictable ecosystem changes. On-site workloads, legacy applications, regulatory environments—Rocky provides drop-in continuity.

Assessing the Need

CentOS 7 security updates stop. PCI DSS, ISO 27001, or simple operational sanity? Running unpatched systems is not an option. Consider: is a live in-place migration feasible, or does your workload justify parallel deployment and staged cutover? For most on-prem or VM-based environments, the in-place script-based upgrade is the shortcut. Cloud images—especially AWS/GCP prebuilds—often favor redeployment instead.

Pre-Migration Checklist

A few non-negotiables:

Note: Staging on actual hardware reveals subtle driver/firmware issues virtual testbeds might miss.

1. Backup—No Shortcuts

One rsync or tar command does not equal a verified backup. Besides a compressed archive, ensure:

• Databases (mysqldump, pg_dumpall) live outside the root filesystem.

• Custom mount points (/data, NFS, iSCSI) are snapshotted.

• Example:

  sudo tar -cvpzf /var/backups/centos7-rootfs-$(date +%F).tar.gz --exclude=/var/backups --one-file-system /
  mysqldump -A -u root -p > /var/backups/all_databases.sql
  

2. Update CentOS 7 Completely

Don’t attempt Rocky conversion on a halfway-patched system. Reboot after updating in case any kernel or library upgrade is pending:

sudo yum clean all && sudo yum update -y
sudo reboot

Skip this, and the migration tool may bail out or introduce subtle RPM conflicts.

3. Run migrate2rocky.sh

The official migration tool covers CentOS-to-Rocky without requiring offline boot media or VM redeployment.

• Download and audit the script (security best practice):

  curl -O https://raw.githubusercontent.com/rocky-linux/rocky-tools/main/migrate2rocky/migrate2rocky.sh
  less migrate2rocky.sh   # Quick review, check for curl-to-bash hazards
  chmod +x migrate2rocky.sh
  

• Recommended: Dry-run mode first.

  sudo ./migrate2rocky.sh -d
  

  Watch for warnings such as:

  WARNING: Repo epel-release not found for Rocky Linux 8
  

  Known issue: Custom/third-party RPMs or nonstandard yum/dnf plugins may cause the script to abort.

• Full run:

  sudo ./migrate2rocky.sh -r
  

  What the script does:

  • Swaps out all CentOS repo files for Rocky equivalents.
  • Rebases package sources.
  • Replaces CentOS branding components.
  • Cross-verifies base and update channel RPM signatures.

4. Post-Migration Validation

Quick and dirty check: is /etc/os-release showing Rocky?

cat /etc/os-release

Expected output:

NAME="Rocky Linux"
VERSION="8.9 (Green Obsidian)"
ID="rocky"

But don’t trust banners. Check for ghost packages:

rpm -qa | grep -i centos

Nothing should return—leftovers often signal incomplete migration or custom-built packages.

Restart and verify all stateful services:

sudo systemctl restart httpd
sudo systemctl status httpd
# Repeat for custom services or any workload using systemd overrides

If a service fails, check for dropped SELinux contexts (restorecon -Rv /var/www), missing libraries (ldd /usr/sbin/nginx), or config drift.

5. Final Cleanup

Now that yum is out, switch entirely to DNF. Do a forced sync across all RPMs:

sudo dnf clean all
sudo dnf distro-sync -y
sudo dnf update -y

Tip: Unused (or CentOS-branded) kernel entries often linger in /boot, eating space. Remove with dnf remove kernel-old, but don’t delete your running kernel.

Common Pitfalls and Fixes

• EPEL/Third-party repos: Many don’t ship Rocky 8-compatible RPMs right away. Disable (enabled=0 in .repo files) and enable only after validation.

• Proprietary drivers: NVIDIA, Mellanox, RAID cards—test your workload thoroughly if DKMS modules are installed. Driver breakage often shows as:

  modprobe: ERROR: could not insert 'nvidia': Invalid argument
  

  It may require driver reinstallation or updated headers.

• SELinux headaches: Context mismatches after migration? Re-label:

  sudo touch /.autorelabel
  sudo reboot
  

• RPM conflicts or broken dependencies: Don’t hesitate to remove orphaned packages. Sometimes, a clean install of a service is faster than debugging weird version mismatches.

In Practice

For high-availability clusters (Pacemaker/Corosync, MariaDB Galera), stagger migrations node-by-node to avoid downtime. Always validate on a non-production replica when service-level objectives are tight.

Migrating from CentOS 7 to Rocky Linux is not entirely frictionless, but—with industrial backup discipline, targeted script usage, and methodical validation—downtime can be minimized to a single restart window. Future patching and compliance audits become routine again.