As mentioned in my previous post there is a script supplied to convert CentOS systems to Oracle Linux, available at https://github.com/oracle/centos2ol. As mentioned in that post it is not safe to use at the currect time.
At the end of this post I will list steps that have worked for me that you should use in preference to this script. It’s at the end as this post is primarily on why not to use the supplied conversion script.
Obviously as Oracle Linux relies on RedHat continuing to follow the software license guidelines and make available all the sources they modify it can only be hoped they continue to do so to ensure Oracle Linux remains a viable alternative to CentOS. However as the RedHat announcement makes CentOS unsafe for production work any more Oracle Linux is the currently available free alternative that requires the least conversion work.
As of 26 December 2020 using that script on a fresh centOS8 install from CentOS-8-x86_64-1905-dvd1.iso results in the below.
The error that stopped the converted system booting after the conversion script was run was
Failed to switch root: Specified switch root path '/sysroot' does not seem to be an OS tree. os-release file is missing.
So I built a new CentOS8 KVM instance as a bare server install. Prior to running the conversion script
[mark@c8toOL8-test etc]$ cat os-release NAME="CentOS Linux" VERSION="8 (Core)" ID="centos" ID_LIKE="rhel fedora" VERSION_ID="8" PLATFORM_ID="platform:el8" PRETTY_NAME="CentOS Linux 8 (Core)" ANSI_COLOR="0;31" CPE_NAME="cpe:/o:centos:centos:8" HOME_URL="https://www.centos.org/" BUG_REPORT_URL="https://bugs.centos.org/" CENTOS_MANTISBT_PROJECT="CentOS-8" CENTOS_MANTISBT_PROJECT_VERSION="8" REDHAT_SUPPORT_PRODUCT="centos" REDHAT_SUPPORT_PRODUCT_VERSION="8" [mark@c8toOL8-test etc]$ ls -la *release -rw-r--r--. 1 root root 38 Aug 14 2019 centos-release -rw-r--r--. 1 root root 420 Aug 14 2019 os-release lrwxrwxrwx. 1 root root 14 Aug 14 2019 redhat-release -> centos-release lrwxrwxrwx. 1 root root 14 Aug 14 2019 system-release -> centos-release [mark@c8toOL8-test etc]$ [mark@c8toOL8-test etc]$ ls /etc/yum.repos.d CentOS-AppStream.repo CentOS-CR.repo CentOS-fasttrack.repo CentOS-Sources.repo CentOS-Base.repo CentOS-Debuginfo.repo CentOS-Media.repo CentOS-Vault.repo CentOS-centosplus.repo CentOS-Extras.repo CentOS-PowerTools.repo [mark@c8toOL8-test etc]$
After running the supplied conversion script we are left with
[root@c8toOL8-test centos2ol]# cd /etc [root@c8toOL8-test etc]# cat os-release cat: os-release: No such file or directory [root@c8toOL8-test etc]# ls *release ls: cannot access '*release': No such file or directory [root@c8toOL8-test etc]# ls /etc/yum.repos.d switch-to-oraclelinux.repo
So the conversion script seems to get only part of the way through, far enough
to wipe out CentOS but not far enough to make OL usable.
So do not use the oracle conversion script.
This does not mean that the script may not become usable at some future point, it is just not safe to use at present.
Unrelated information
On a fresh Oracle Linux 8 install these are the files created
[mark@oracle8-freshinstall etc]$ cat os-release NAME="Oracle Linux Server" VERSION="8.3" ID="ol" ID_LIKE="fedora" VARIANT="Server" VARIANT_ID="server" VERSION_ID="8.3" PLATFORM_ID="platform:el8" PRETTY_NAME="Oracle Linux Server 8.3" ANSI_COLOR="0;31" CPE_NAME="cpe:/o:oracle:linux:8:3:server" HOME_URL="https://linux.oracle.com/" BUG_REPORT_URL="https://bugzilla.oracle.com/" ORACLE_BUGZILLA_PRODUCT="Oracle Linux 8" ORACLE_BUGZILLA_PRODUCT_VERSION=8.3 ORACLE_SUPPORT_PRODUCT="Oracle Linux" ORACLE_SUPPORT_PRODUCT_VERSION=8.3 [mark@oracle8-freshinstall etc]$ ls -la *release -rw-r--r--. 1 root root 32 Nov 5 12:58 oracle-release -rw-r--r--. 1 root root 479 Nov 5 12:58 os-release -rw-r--r--. 1 root root 45 Nov 5 12:58 redhat-release lrwxrwxrwx. 1 root root 14 Nov 5 12:58 system-release -> oracle-release [mark@oracle8-freshinstall etc]$ [mark@oracle8-freshinstall etc]$ ls /etc/yum.repos.d oracle-linux-ol8.repo uek-ol8.repo
An alternative upgrade method that seems to work
This method was suggested in a response to the origional CentOS moving to be stream only post at https://blog.centos.org/2020/12/future-is-centos-stream/ by commenter “Phil”. I have edited it to what works for me (which was basically correcting ‘centos-release’ and highlighting the importance of needing yum-utils installed before disabling the centos repositories; plus the reset module step I needed to use.
setenforce 0
dnf -y install yum-utils # <---- this MUST be installed
repobase=http://yum.oracle.com/repo/OracleLinux/OL8/baseos/latest/x86_64/getPackage
wget \
${repobase}/redhat-release-8.3-1.0.0.1.el8.x86_64.rpm \
${repobase}/oracle-release-el8-1.0-1.el8.x86_64.rpm \
${repobase}/oraclelinux-release-8.3-1.0.4.el8.x86_64.rpm \
${repobase}/oraclelinux-release-el8-1.0-9.el8.x86_64.rpm
rpm -e centos-release --nodeps
dnf --disablerepo='*' localinstall ./*rpm
:> /etc/dnf/vars/ociregion
dnf remove centos-linux-repos
dnf --refresh distro-sync --allowerasing --nobest
The last dnf command results in
[root@oracle8 ~]# dnf --refresh distro-sync --allowerasing --nobest Oracle Linux 8 BaseOS Latest (x86_64) 28 kB/s | 2.7 kB 00:00 Oracle Linux 8 Application Stream (x86_64) 36 kB/s | 3.1 kB 00:00 Oracle Instant Client for Oracle Linux 8 (x86_64) 28 kB/s | 2.5 kB 00:00 Latest Unbreakable Enterprise Kernel Release 6 for Oracle Linux 8 (x86_64) 26 kB/s | 2.5 kB 00:00 Dependencies resolved. The operation would result in switching of module 'llvm-toolset' stream 'rhel8' to stream 'ol8' Error: It is not possible to switch enabled streams of a module. It is recommended to remove all installed content from the module, and reset the module using 'dnf module reset' command. After you reset the module, you can install the other stream.
Easily fixed with
[root@oracle8 ~]# dnf module reset llvm-toolset Last metadata expiration check: 0:01:22 ago on Sat 26 Dec 2020 11:41:27 NZDT. Dependencies resolved. ================================================================================================================================================================================ Package Architecture Version Repository Size ================================================================================================================================================================================ Resetting modules: llvm-toolset Transaction Summary ================================================================================================================================================================================ Is this ok [y/N]: y Complete!
After which rerunning the “dnf –refresh distro-sync –allowerasing –nobest” works.
There was one failure
Failed: pcp-webapi-4.3.0-3.el8.x86_64
This left “package-cleanup –problems” reporting problems for that; resolved with “rpm -e pcp pcp-webapi pcp-mnager –nopreun” (noreun was required as the pre-uninstall script kept erroring).
“package-cleanup –orphans” showed quite a few entries, these are packages installed that are not available from the configured repositories; and included things like buildah and podman that certainly should be.
“dnf autoremove” also listed a few packages I would want to keep as available for removal.
The supprising issue was that “shutdown -r now” refused to shutdown, “init 6” has no effect. Neither did “halt” or “reboot”; too many interface changes ?. I had to trigger a ‘force-off’ poweroff from the virt-manager interface and start the KVM again from there. However after that first forced restart reboots were behaving normally again.
So that would be my recomended upgrade path, not the script on github.
Future posts planned
While conversion to Oracle Linux is the least disruptive as far as compatibility goes, I will also be looking at getting tools like docker-ce, puppet-agent, bacula-client etc running on Debian; which if successful may result in a lot less rhel based posts here and a lot more debian based ones.
