An error is reported during the cluster preinstallation. The disk space in the configuration planning file is set to a large value (the disk space is not multiplied by 0.9). However, clusters are successfully preinstalled on some nodes. In this case, roll back the preinstallation on nodes that are failed to preinstall clusters and generate a configuration file to perform the preinstallation.
To roll back the preinstallation without affecting the normal nodes that with clusters preinstalled successfully, perform the following steps:
1 Uninstall the diskmgt.
[root@tac1 tmp]# cd /usr/local/diskmgt/script/
[root@tac1 script]# ./uninstall.sh
On the remote host, unmount the disk partition.
Delete a disk partition.
fdisk /dev/sdb and press Enter.
Enter d and press Enter.
Enter w and press Enter.
4. Delete the UUID of the disk mounting information.
Run the vi /etc/fstab command to delete the information about the UUID.
The preceding four steps are performed on
all nodes where the error occurs.
5. In the preinstallation configuration file, delete all the IP addresses of the nodes with clusters normally preinstalled from the g_hosts and g_parted_conf values.
[omm@fi1 preinstall]$ vi preinstall.ini
[omm@fi1 preinstall]$ pwd
6. Perform the preinstallation on the active management node.