Upgrading from SLE HA 15 to SLE HA 16.0

Perform the following steps before starting the upgrade:

1. Make sure the latest system updates are installed by running zypper patch.

2. Read the release notes. Find the list of all changes, new features and known issues in the release notes.

   Consult the release notes to check whether the following applies:

   • Your hardware needs special considerations

   • Any currently used software packages have changed significantly

   • Your installation requires special precautions

3. Back up your data by copying the existing configuration files to a separate medium (such as a tape device or removable hard disk). This primarily applies to files stored in /etc and some directories and files in /var and /opt.

4. Check the available disk space. If you suspect you are running short of disk space, free up disk space or resize partitions.

5. If your machine serves as a VM Host Server for KVM, properly shut down all running VM Guests prior to the upgrade.

6. Root login by ssh will be disabled by default in SLE 16. You might want to enable it back before starting the migration. To do so, create /etc/ssh/sshd_config.d/50-permit-root-login.conf with the following content:

   PermitRootLogin yes

   Once the migration is complete, you can install openssh-server-config-rootlogin package and remove the previously added file.

1. Log in to the node either as the root user or as a user with sudo privileges.

2. Run the cluster health check:

   > sudo crm cluster health sles16

   This command reports issues in two categories:

   • The WARN category indicates cluster configuration that isn't recommended, such as deprecated features. Issues in this category can be fixed during or after the upgrade, so the migration check passes. For example:

     [WARN] Corosync transport "udpu" will be deprecated in favor of "knet" in corosync 3.

   • The FAIL category indicates cluster configuration that can't be fixed during or after the upgrade, such as removed resource agents. Issues in this category must be fixed before you can upgrade the cluster, so the migration check fails. The command's output will tell you what to fix. For example:

     [FAIL] The following resource agents will be removed in SLES 16.
            * ocf:heartbeat:IPaddr: please replace it with ocf:heartbeat:IPaddr2

3. Fix any FAIL issues, then run the cluster health check again to make sure you didn't miss anything.

1. Log in either as the root user or as a user with sudo privileges.

2. Check the status of the cluster:

   > sudo crm status

   • If the resources are in maintenance mode, they will continue to run even while the cluster services are offline.

     

     This is only true if the nodes keep running. If you need to shut down or reboot the nodes, take the cluster out of maintenance mode first.

   • If the resources aren't in maintenance mode, they will all stop when you stop the cluster services.

3. Stop the cluster services on all nodes at once:

   > sudo crm cluster stop --all

4. Check the status of the cluster services:

   > sudo crm cluster status

   Both Pacemaker and Corosync should be inactive.

1. Use zypper packages to check for orphaned and unneeded packages.

   1. Orphaned packages are no longer available in any of the configured package repositories. They can no longer get updated and become unsupported.

      For a list of orphaned packages, run:

      > zypper packages --orphaned

   2. Unneeded packages are dependencies of packages that have been installed either explicitly by the user or implicitly as part of a pattern or product, and that have been removed in the meantime. They are usually no longer needed and should be removed, too.

      For a list of unneeded packages, run:

      > zypper packages --unneeded

      

      Tip

      To avoid unneeded packages, use zypper rm with the --clean-deps option.

   3. You can combine both lists into one:

      > zypper packages --orphaned --unneeded

   4. Use these lists to determine which packages are still needed and which can be safely removed.

      

      Warning: Do not remove packages you need

      If packages are renamed or removed from a pattern or product, zypper may no longer consider them explicitly installed and mark them as unneeded, even though they are still crucial for your installation.

      Carefully review the list of packages you are removing.

      To remove all orphaned and unneeded packages with a single command, run:

      > sudo zypper rm $(zypper --no-refresh packages --orphaned --unneeded | gawk '{print $5}' | tail -n +5)

      Exclude a single package or pattern from being uninstalled:

      > sudo zypper rm $(zypper --no-refresh packages --orphaned --unneeded | gawk '{print $5}' | tail -n +5 | grep -v PACKAGE_TO_EXCLUDE)

      Exclude multiple packages defined in a text file, separated by a newline:

      > sudo zypper rm $(zypper --no-refresh packages --orphaned --unneeded | gawk '{print $5}' | tail -n +5 | grep -v -f /PACKAGES/TO/KEEP.txt)

2. Check for any *.rpmnew and *.rpmsave files. When an upgrade includes changes to a default configuration file that has been altered after package installation, instead of overwriting the file, one of these file types is created. While *.rpmnew contains the new default configuration and leaves your altered file untouched, *.rpmsave is a copy of your altered configuration that has been replaced by the new default file.

   If you find any of these files, examine their content and merge desirable changes. You do not need to search the whole file system, only the /etc directory. Use the following command:

   > find /etc/ -name "*.rpmnew" -o -name "*.rpmsave"

1. Log in either as the root user or as a user with sudo privileges.

2. Start the cluster services on all nodes at once:

   > sudo crm cluster start --all

3. Check the status of the cluster services:

   > sudo crm cluster status

   Both Corosync and Pacemaker should be active.

4. Check the status of the cluster:

   > sudo crm status

   The nodes will have the status UNCLEAN (offline), but will soon change to Online.

1. Log in to the node either as the root user or as a user with sudo privileges.

2. Run the cluster health check with the --fix option to automatically fix certain issues:

   > sudo crm cluster health sles16 --fix

   The output also lists any issues that can't be fixed automatically and tells you how to fix them manually. For example:

   [WARN] The CIB is not validated with the latest schema version.
          * Latest version: 4.0
          * Current version: 3.10
          Please run "crm configure upgrade force" to upgrade to the latest version.

3. Fix any issues, then run the cluster health check again to make you didn't miss anything.