Guidance for Red Hat Enterprise Linux 8.x on IBM Storage Scale nodes

Red Hat Enterprise Linux® 8.x is supported starting with IBM Storage Scale release 5.0.4.

Remember: Red Hat Enterprise Linux 8.0 is not an Extended Update Support (EUS) release, so the errata stream stops when Red Hat Enterprise Linux 8.1 is released. Do not use Red Hat Enterprise Linux 8.0 in production. Upgrade Red Hat Enterprise Linux to 8.1 to stay within the IBM Storage Scale support matrix.

The following list contains some general guidance when installing Red Hat Enterprise Linux 8.x on IBM Storage Scale nodes.

After Red Hat Enterprise Linux 8.x is installed, IBM Storage Scale on x86 requires elfutils-devel to be installed for successfully building the GPFS portability layer (GPL) by using the mmbuildgpl command.
Review the Red Hat Enterprise Linux 8.x documentation topic Considerations for adopting RHEL8 for all hardware and driver support that was previously supported but has since been removed.
Review the IBM Storage Scale FAQ in IBM® Documentation for more information and for the supported kernel versions.

Red Hat® provides a leapp upgrade utility to upgrade the OS from Red Hat Enterprise Linux 7.x to 8.x but it has some limitations. It is recommended to provision a new Red Hat Enterprise Linux 8.x node instance instead of using a leapp upgrade if possible. If you are using the leapp upgrade utility, there are some details to be aware of that are as follows.

A leapp upgrade might upgrade the IBM Storage Scale node to the latest release of Red Hat Enterprise Linux 8.x. Ensure that leapp upgrade does not upgrade the node to an unsupported OS version. For supported OS and kernel versions, see IBM Storage Scale FAQ in IBM Documentation.
Note: On s390x, a leapp upgrade from Red Hat Enterprise 7.x to 8.x is not supported. Using the leapp utility for the upgrade can lead to an unbootable system as described in the Red Hat bugzillas RH1890215 and RH1881428. Contact IBM support if you plan to upgrade from Red Hat Enterprise 7.x to 8.x on s390x.
Using the leapp upgrade utility to upgrade from Red Hat Enterprise Linux 7.6 to 8.x with IBM Storage Scale 5.0.4.x or later installed might result in the removal of some IBM Storage Scale packages. This package removal might potentially happen with any third-party software. It is highly recommended to use leapp-0.8.1-1 or later. The leapp utility is supported by Red Hat and any issues that are encountered with the leapp upgrade utility must be reported to Red Hat. Before you run leapp upgrade, save the list of IBM Storage Scale packages that are installed on the system by using the rpm -qa | grep gpfs command and then compare it to the packages installed after the upgrade. The /var/log/leapp directory also contains the packages that are to be removed.
If Cluster Export Services (CES) are enabled on an IBM Storage Scale node running on a version earlier than 5.1.0 and you run a leapp upgrade from Red Hat Enterprise Linux 7.x to 8.x, then you must first stop object and remove its associated packages. For more information, see Instructions for removing object protocol packages when upgrading protocol nodes to Red Hat Enterprise Linux 8.x.
Note: With certain versions of IBM Storage Scale, the object protocol is not supported on Red Hat Enterprise Linux 8.x. Refer to IBM Storage Scale FAQ in IBM Documentation for determining the IBM Storage Scale version on which the object protocol is supported on Red Hat Enterprise Linux 8.x and if you can reinstall object or not.
If Cluster Export Services (CES) are enabled on an IBM Storage Scale 5.0.4.x or later node and you run a leapp upgrade from Red Hat Enterprise Linux 7.x to 8.x, the NFS el7 packages are removed. The NFS el8 packages (gpfs.nfs-ganesha-*.rpm, gpfs.nfs-ganesha-gpfs-*.rpm, gpfs.nfs-ganesha-utils-*.rpm) need to be installed after the leapp upgrade. The NFS el8 packages are located in the /usr/lpp/mmfs/5.x.x.x/ganesha_rpms/rhel8 directory. You can use yum localinstall or rpm -ivh commands to install these packages.
If Cluster Export Services (CES) are enabled on an IBM Storage Scale 5.0.4.x or later node and you run a leapp upgrade from Red Hat Enterprise Linux 7.x to 8.x, the SMB el7 package is removed. The SMB el8 package needs to be installed after the leapp upgrade. The SMB el8 package is located in the /usr/lpp/mmfs/5.x.x.x/smb_rpms/rhel8 directory. You can use yum localinstall or rpm -ivh commands to install this package.
If performance collection or GUI functionalities are enabled on an IBM Storage Scale 5.0.4.x or later node and you run a leapp upgrade from Red Hat Enterprise Linux 7.x to 8.x, the associated gpfs.gss.pmcollector or gpfs.gui packages are removed and they need to be reinstalled after the leapp upgrade. For more information, see Reinstalling performance monitoring collector on Red Hat Enterprise Linux 8.x nodes.
- Performance collection package can be installed from: /usr/lpp/mmfs/5.x.x.x/zimon_rpms/rhel8/gpfs.gss.pmcollector
- If NFS is also installed, the NFS Ganesha performance monitoring package can be installed from: /usr/lpp/mmfs/5.x.x.x./zimon_rpms/rhel8/gpfs.pm-ganesha
- Management GUI package can be installed from: /usr/lpp/mmfs/5.x.x.x/gpfs_rpms/gpfs.gui*
  After installing the management GUI package, start the management GUI by using the systemctl start gpfsgui command.
A leapp upgrade from Red Hat Enterprise Linux 7.x to 8.x removes the python-ldap package, which is prerequisite for the mmadquery command. Ensure that you install python3-ldap after leapp upgrade, if you plan to use mmadquery on Red Hat Enterprise Linux 8.x.
The CLI utility mmadquery that is bundled with IBM Storage Scale 5.1.0 for Red Hat Enterprise Linux 7.x is based on python2. The CLI utility mmadquery that is bundled with IBM Storage Scale 5.1.0 for Red Hat Enterprise Linux 8.x is based on python3. When you install IBM Storage Scale 5.1.0 on Red Hat Enterprise Linux 7.x and upgrade an operating system from Red Hat Enterprise Linux 7.x to Red Hat Enterprise Linux 8.x, mmadquery still points to python2.
To point to the correct mmadquery, which is based on python3, perform the following steps:
1. Remove existing /usr/lpp/mmfs/bin/mmadquery.
2. Copy /usr/lpp/mmfs/bin/mmadquery_py3 to /usr/lpp/mmfs/bin/mmadquery.
After all packages that were removed by leapp upgrade are reinstalled as described in the preceding text, complete the steps in Guidance for upgrading the operating system on IBM Storage Scale nodes to finish bringing the node online after the OS is upgraded to Red Hat Enterprise Linux 8.x.
If Cluster Export Services (CES) are enabled on an IBM Storage Scale node, an in-place upgrade from Red Hat Enterprise Linux 8.x by using leapp upgrade utility might fail.
The leapp utility does not support upgrading Red Hat Enterprise Linux 8.x on the file protocols nodes because of the winbind strings in the /etc/nsswitch.conf file. When either SMB or NFS or both SMB and NFS protocols are configured for Active Directory (AD) based authentication schemes by using the mmuserauth command, this command adds winbind in the /etc/nsswitch.conf file of each protocol node. The in-place upgrade with winbind strings in the /etc/nsswitch.conf is not allowed because of a known OS upgrade limitation.
To do the in-place upgrade, you need to complete the steps from 1 through 4 in the Troubleshooting known issues documentation for upgrading Red Hat Enterprise Linux 8.x. After the OS upgrade, when you resume the protocol nodes, the IBM Storage Scale adds the removed winbind entries in the /etc/nsswitch.conf file on each protocol node. You can verify the successful addition of the removed winbind entries for one of the CES nodes by issuing the following command:
```
# mmuserauth service check -N cesNodes
```
After the OS upgrade, do the following steps to confirm correctness of the /etc/nsswitch.conf file on the nodes:
1. Check whether the current configuration is valid by issuing the following command:
```
# authselect check
```
2. Check whether the /etc/nsswitch.conf file is linked to the current authselect profile nsswitch.conf file by issuing the following command:
```
# ls -l /etc/nsswitch.conf
```
3. If either the current configuration is incorrect or the /etc/nsswitch.conf file is not linked to the current authselect profile, apply forcefully the current authselect profile on the nodes.
  1. Check the current authselect profile by issuing the following command:
```
# authselect current
```
  2. Apply forcefully the authselect profile again by issuing the following command:
```
# authselect select ProfileNameIdentifiedFromPreviousCommand --force
```

The packages that leapp upgrade might remove can vary depending on packages that are installed and dependencies that might exist. It might be required to reinstall any removed packages, which can typically be determined by viewing the log /var/log/leapp/leapp-upgrade.log and looking for Removing dependent packages after the leapp upgrade completes and the node restarts.