Below are the recommended steps taken to update to CentOS 7.4 without breaking the ZFS packages.
- Uninstall ZFS and spl packages
yum remove zfs* -y
yum remove libzfs2 -y
yum remove spl* -y
- Remove your preconfig script if you have it, and download the most up to date version. Run the script with the below options to remove and install all relevant packages
rm -f preconfig
wget images .45d rives.com/gtools/preconfig
chmod +x preconfig
./preconfig -zf
- The script will auto reboot the server, when it comes back you’ll need to run the first two commands to clean the mirror cache. After running yum clean all, you’ll be prompted to remove
some old files, do so. Then re-run preconfig to remove the old kernel.
yum clean all
yum makecache fast
./preconfig -af
- When that finishes the server will reboot once more. When it comes back everything should be updated and back to normal.
Note: This will require some downtime if the servers are in production. If you are running a cluster, you have the option to do one server at a time to avoid complete downtime of the cluster
storage.
Below are the recommended steps taken if you’ve updated to CentOS 7.4 and broken the ZFS packages.
- Uninstall ZFS and spl packages, as well as removing a few modules
yum remove zfs* -y
yum remove libzfs2 -y
yum remove spl* -y
rm -rf /lib/modules/ * /extra/ *
rm -rf /lib/modules/ * /weak-updates/ *
- Remove your preconfig script if you have it, and download the most up to date version. Run the script with the below options to remove and install all relevant packages
rm -f preconfig
wget images .45d rives.com/gtools/preconfig
chmod +x preconfig
./preconfig -zf
- The script will auto reboot the server. When it comes back, you’ll need to run the first two commands to clean the mirror cache. After running yum clean all, you’ll be prompted to remove
some old files, do so. Then re-run preconfig to remove the old kernel.
yum clean all
yum makecache fast
./preconfig -af
- When that finishes the server will reboot once more. When it comes back everything should be updated and back to normal.
Note: This will require some downtime if the servers are in production. If you are running a cluster, you have the option to do one server at a time to avoid complete downtime of the cluster
storage.