7 Setting up an SAP HANA cluster #

The following procedure needs to be executed on the primary node (also called the “master”). Before proceeding, make sure the prerequisites listed in Section 7.1, “Prerequisites” are fulfilled.

1. Open the YaST control center. In it, click HA Setup for SAP Products in the category High Availability.

2. If an SAP HANA instance has been detected, you can choose between the scale-up scenarios Performance-optimized, Cost-optimized, or Chained (multi-tier). For information about these scale-up scenarios, see Section 1.1.3, “Simplified SAP HANA system replication setup”.

   Continue with Next.

   

3. This step of the wizard presents a list of prerequisites for the chosen scale-up scenario. These prerequisites are the same as those presented in Section 7.1, “Prerequisites”.

   Continue with Next.

4. The next step lets you configure the communication layer of your cluster.

   • Provide a name for the cluster.

   • The default transport mode Unicast is usually appropriate.

   • Under Number of rings, a single communication ring usually suffices.

     For redundancy, it is often better to use network interface bonding instead of multiple communication rings. For more information, see Administration Guide, Part “Configuration and Administration”, Chapter “Network Device Bonding” at https://documentation.suse.com/sles-15.

   • From the list of communication rings, configure each enabled ring. To do so, click Edit selected, then select a network mask (IP address) and a port (Port number) to communicate over.

     Finish with OK.

   • Additionally, decide whether to enable the configuration synchronization service Csync2 and Corosync secure authentication using HMAC/SHA1.

     For more information about Csync2, see Administration Guide Part “Installation, Setup and Upgrade”, Chapter “Using the YaST Cluster Module”, Section “Transferring the Configuration to All Nodes” at https://documentation.suse.com/sles-15.

     For more information about Corosync secure authentication, see Administration Guide, Part “Installation, Setup and Upgrade”, Chapter “Using the YaST Cluster Module”, Section “Defining Authentication Settings” at https://documentation.suse.com/sles-15.

   Proceed with Next.

   

5. The wizard will now check whether it can connect to the secondary machine using SSH. If it can, it will ask for the root password to the machine.

   Enter the root password.

   The next time the primary machine needs to connect to the secondary machine, it will connect using an SSH certificate instead of a password.

6. For both machines, set up the host names and IP address (for each ring).

   Host names chosen here are independent from the virtual host names chosen in SAP HANA. However, to avoid issues with SAP HANA, host names must not include hyphen characters (-).

   If this has not already been done before, such as during the initial installation of SAP HANA, host names of all cluster servers must now be added to the file /etc/hosts. For this purpose, activate Append to /etc/hosts.

   Proceed with Next.

7. If NTP is not yet set up, do so. This avoids the two machines from running into issues because of time differences.

   1. Click Reconfigure.

   2. On the tab General Settings, activate Now and on Boot.

   3. Add a time server by clicking Add. Click Server and Next. Then specify the IP address of a time server outside of the cluster. Test the connection to the server by clicking Test.

      To use a public time server, click Select › Public server and select a time server. Finish with OK.

      Proceed with OK.

   4. On the tab Security Settings, activate Open Port in Firewall.

   5. Proceed with Next.

8. In the next step, choose fencing options. The YaST wizard only supports the fencing mechanism SBD (STONITH block device). To avoid split-brain situations, SBD uses a disk device which stores cluster state.

   The chosen disk must be available from all machines in the cluster under the same path. Ideally, use either by-uuid or by-path for identification.

   The disk must not use host-based RAID, cLVM2 or reside on a DRBD instance. The device can have a small size, for example, 100 MB.

   

   Warning: Data on device will be lost

   All data on the chosen SBD device or devices will be deleted.

   To define a device to use, click Add, then choose an identification method such as by-uuid and select the appropriate device. Click OK.

   To define additional SBD command-line parameters, add them to SBD options.

   If your machines reboot particularly fast, activate Delay SBD start.

   For more information about fencing, see the Administration Guide at https://documentation.suse.com/sles-15.

   Proceed with Next.

9. The following page allows configuring watchdogs which protect against the failure of the SBD daemon itself and force a reboot of the machine in such a case.

   It also lists watchdogs already configured using YaST and watchdogs that are currently loaded (as detected by lsmod).

   To configure a watchdog, use Add. Then choose the correct watchdog for your hardware and leave the dialog with OK.

   For testing, you can use the watchdog softdog. However, we highly recommend using a hardware watchdog in production environments instead of softdog. For more information about selecting watchdogs, see Administration Guide, Part “Storage and Data Replication”, Chapter “Storage Protection”, Section “Conceptual Overview”, Section “Setting Up Storage-based Protection”, Section “Setting up the Watchdog” at https://documentation.suse.com/sles-15.

   Proceed with Next.

10. Set up the parameters for your SAP HANA installation or installations. If you have selected the cost-optimized scenario, additionally fill out details related to the non-production SAP HANA instance.

    Production SAP HANA instance

    • Make sure that the System ID and Instance number match those of your SAP HANA configuration.

    • Replication mode and Operation mode usually do not need to be changed.

      For more information about these parameters, see the HANA Administration Guide provided to you by SAP.

    • Under Virtual IP address, specify a virtual IP address for the primary SAP HANA instance. Under Virtual IP Mask, set the length of the subnetwork mask in CIDR format to be applied to the Virtual IP address.

    • Prefer site takeover defines whether the secondary instance should take over the job of the primary instance automatically (true). Alternatively, the cluster will restart SAP HANA on the primary machine.

    • Automatic registration determines whether primary and secondary machine should switch roles after a takeover.

    • Specify the site names for the production SAP HANA instance on the two nodes in Site name 1 and Site name 2.

    • Having a backup of the database is a precondition for setting up SAP HANA replication.

      If you have not previously created a backup, activate Create initial backup. Under Backup settings, configure the File name and the Secure store key for the backup. The key in the SAP HANA Secure User Store on the primary node must have been created before starting the wizard.

      For more information, see the documentation provided to you by SAP.

    • Cost-optimized scenario only: Within Production system constraints, configure how the production instance of SAP HANA should behave while inactive on the secondary node.

      Setting the Global allocation limit allows directly limiting memory usage. Activating Preload column tables will increase memory usage.

      For information about the necessary global allocation limit, see documentation provided to you by SAP such as How to Perform System Replication for SAP HANA at https://archive.sap.com/documents/docs/DOC-47702.

    Cost-optimized scenario only: non-production SAP HANA instance

    • Make sure that the System ID and Instance number match those of your non-production SAP HANA instance.

      These parameters are needed to allow monitoring the status of the non-production SAP HANA instance using the SAPInstance resource agent.

    • Generate a hook script for stopping the non-production instance and starting the production instance and removing the constraints on the production system. The script is written in Python 2 and can be modified as necessary later.

      Click Hook script and then set up the correct user name and password for the database. Then click OK.

      You can now manually verify and change the details of the generated hook script. When you are done, click OK to save the hook script at /hana/shared/SID/srHook.

      

      Warning: Passwords stored in plain text

      By default, the hook script stores all credentials in plain text. To improve security, modify the script yourself.

    Proceed with Next.

    

    Figure 7.1: SAP HANA options (cost-optimized scenario) #

      

11. On the page High-Availability Configuration Overview, check that the setup is correct.

    To change any of the configuration details, return to the appropriate wizard page by clicking one of the underlined headlines.

    Proceed with Install.

12. When asked whether to install additional software, confirm with Install.

13. After the setup is done, there is a screen showing a log of the cluster setup.

    To close the dialog, click Finish.

14. Multi-tier/chain scenario only: Using the administrative user account for the production SAP HANA instance, register the out-of-cluster node for system replication:

    SIDadm > hdbnsutil -sr_register --remoteHost=SECONDARY_HOST_NAME \
    --remoteInstance=INSTANCE_NUMBER --replicationMode=async \
    --name=SITE_NAME

1. On the production machines with SAP HANA installed, create a configuration file by running the sap_ha YaST module.

2. On the last screen, click the Save configuration button.

3. Decide what you want to do:

   • To review the configuration, upload and validate the configuration on the primary SAP HANA machine and run:

     # yast2 sap_ha readconfig CONFIGURATION_FILE_PATH

     It is possible to start the installation on the review screen.

   • To start the installation based on the provided configuration file unattended, run:

     # yast2 sap_ha readconfig CONFIGURATION_FILE_PATH unattended

4. Import, validate, and install the cluster unattended, based on the provided configuration file:

   # yast2 sap_ha readconfig CONFIGURATION_FILE_PATH unattended