These scripts are designed to aid a user to migrate from one Enterprise Linux system (such as CentOS, RHEL) to Rocky Linux.
Use migrate2rocky to convert Enterprise Linux 8 systems to Rocky Linux 8.
Use migrate2rocky9 to convert Enterprise Linux 9 systems to Rocky Linux 9.
Upgrades are not supported.
Migration from non-RHEL derived distributions such as Fedora is not supported.
./migrate2rocky.sh -h
├── -h # --> Display this help
├── -r # --> Convert to Rocky
└── -V # --> Verify switch
[!! USE WITH CAUTION !!]Please note the following disk space requirements. These requirements may vary from one system to another. Failure to have adequate disk space available may result in migrate2rocky leaving the system in an unstable state:
/usr 250M
/var 1.5G
/boot 50M
When running this script, especially via a remote session, it is highly recommended to enter a screen or tmux session before running. If a standard ssh or terminal session, such as the Cockpit Terminal window, is disrupted, the script will die and leave the system in a potential unrecoverable state. For more on tmux sessions, please see: https://github.com/tmux/tmux/wiki
Do not run this script through the Terminal screen built into Cockpit. As the script runs the upgrades, Cockpit will be restarted and Terminal connection will disconnect, thus stopping the script and leaving the system in an unrecoverable state. It may be possible to launch a screen or tmux session from the Cockpit Terminal, but USE AT YOUR OWN RISK.
If you are attempting to migrate a system that has not been updated since 8.0
then you must run dnf update before attempting the migration.
If you are migrating from CentOS 8.0 then you must manually fix the baseurls of
the CentOS repositories before running dnf update:
sed -i -r \
-e 's!^mirrorlist=!#mirrorlist=!' \
-e 's!^#?baseurl=http://(mirror|vault).centos.org/\$contentdir/\$releasever/!baseurl=https://dl.rockylinux.org/vault/centos/8.5.2111/!i' \
/etc/yum.repos.d/CentOS-*.repo
This script expects the original repository configuration being present, as
well as enabled (i.e. for CentOS the baseos repo configuration in the
/etc/yum.repos.d/CentOS-Linux-BaseOS.repo file has to be present and enabled).
Also make sure that there are no other repositories which could interfere
with the original configuration.
Any distribution that has had its core repositories altered, removed, duplicated or overridden may cause migrate2rocky to break or corrupt the system when run. Any attempt to migrate such systems, even after reversing the changes made by such software, is not supported in any way. In all cases you should backup your system before using migrate2rocky and USE AT YOUR OWN RISK.
This especially happens on systems configured with a centralized package management like Katello (RedHat Satellite 6) or Uyuni (RedHat Satellite 5, SUSE Manager).
Installing : rocky-release-8.3-13.el8.noarch 2/5Error unpacking rpm package rocky-release-8.3-13.el8.noarch
...
error: unpacking of archive failed on file /usr/share/redhat-release: cpio: File from package already exists as a directory in system
error: rocky-release-8.3-13.el8.noarch: install failed
...
Error: Transaction failed
This results from conflicts in the directory structure of RHEL with that of RockyLinux. migrate2rocky will detect the issue and go on to remove the conflicting directory and install rocky-release with the rpm command.
The above error message is a display bug in dnf. It does not affect the actual dnf command or the migration. You may safely ignore this message. (RHBZ#1980712)
This is normal. The running kernel cannot be safely removed when migrate2rocky is run. The RockyLinux kernel should come up as the default highlighted kernel on reboot but the other ones will remain until they are removed or replaced by newer kernels. If you want you can manually remove the old kernels after reboot with dnf or rpm.
After migrating a system with an OpenJDK package installed you might encounter
that java does not work any more. There is a bug in the OpenJDK packages that
leads to losing the symbolic links in /etc/alternatives during the migration.
A bug report against the upstream packages has been filed
here. As a workaround
you can use the following script to recreate the package default alternatives:
rpm -qa --scripts java-{1.8.0,11}-openjdk-{headless,devel} | sed -n '/postinstall/, /exit/{ /postinstall/! { /exit/ ! p} }' | sh
This issue is caused by a version mismatch due to the way that modules work that trick ipa into thinking that the package was downgraded even if it was not. To fix this issue run the following command after migration:
ipa-server-upgrade --skip-version-check
Note: Since ipa-server-upgrade is a java program you will likely have to run the command to mitigate the "Symbolic links to Java programs..." issue above before running this command.
This is because the centos-release-* packages that contain the .repo files for the individual repositories depend on centos-release. Storage sig and related release packages should be available soon from RockyLinux. In the meantime you can use a command like the following to install the .repo files and continue to use the repository from CentOS (note please substitute the URL to the release package for the repo that you need):
rpm2cpio <(curl http://mirror.centos.org/centos/8/extras/x86_64/os/Packages/centos-release-gluster9-1.0-1.el8.noarch.rpm) | cpio -iD/ \*.repo
The latest version of this script can be found here.
The migrate2rocky script pipes everything shown on stdout and stderr to
/var/log/migrate2rocky.log.
If you run in to issues executing this script, please submit an issue here.
Make sure to include the output log, and remove any sensitive information. (if any)
Feel free to create a pull request if you think you've got the fix.