Upgrading the SAP HANA Database

Prerequisites:

• The protected SAP HANA database is running on node1 and node2.

• System replication is enabled with node1 registered as the primary replication site and node2 registered as a secondary replication site.

*If using the ‘logreplay_readaccess’ operation mode for an active/active (read-enabled secondary) configuration, you must temporarily switch to the ‘logreplay’ operation mode. This is because the ‘logreplay_readaccess’ operation mode requires both the primary and secondary systems to be running the same version of SAP HANA and hence cannot be used during a rolling upgrade procedure. The original active/active (read-enabled secondary) configuration may be reestablished after the upgrade process is complete by manually executing an hdbnsutil -sr_register command to set --operationMode=logreplay_readaccess on the desired read-enabled secondary.

Follow the steps below to update SAP HANA database.

1. Disable resource health monitoring (quickCheck processes) on node1:
   1. Make a note of the value of LKCHECKINTERVAL in /etc/default/LifeKeeper on node1. This value will be used in step 8.
   2. Set LKCHECKINTERVAL=0 in /etc/default/LifeKeeper on node1.
   3. Restart the lkcheck daemon by running ‘killall lkcheck’ on node1.
      
      
2. Stop LifeKeeper on node2:
   /opt/LifeKeeper/bin/lkstop -f
   
   
3. Stop the SAP HANA database on node2:
   su - <sid>adm -c "sapcontrol -nr <InstNum> -function StopSystem HDB"
   
   
4. Verify that the SAP HANA database is stopped on node2 (the HDB Daemon process status should be ‘Stopped’):
   su - <sid>adm -c "sapcontrol -nr <InstNum> -function GetProcessList"
   
   
5. Perform SAP HANA database patching procedures on node2. For example, from the directory containing the new SAP HANA software:
   ./hdblcm --action=update
   
   
6. Start the SAP HANA database on node2:
   su - <sid>adm -c "sapcontrol -nr <InstNum> -function StartSystem HDB"
   
   
7. Start LifeKeeper on node2:
   /opt/LifeKeeper/bin/lkstart
   
   
8. Re-enable resource health monitoring (quickCheck processes) on node1:
   1. Set LKCHECKINTERVAL=<Previous value from step 1(a)> in /etc/default/LifeKeeper on node1.
   2. Restart the lkcheck daemon by running ‘killall lkcheck’ on node1.
      
      
9. Disable resource health monitoring (quickCheck processes) on node2:
   1. Make a note of the value of LKCHECKINTERVAL in /etc/default/LifeKeeper on node2. This value will be used in step 17.
   2. Set LKCHECKINTERVAL=0 in /etc/default/LifeKeeper on node2.
   3. Restart the lkcheck daemon by running ‘killall lkcheck’ on node2.
      
      
10. Wait for the system replication from node1 to node2 to synchronize. Once in sync, the LifeKeeper SAP HANA resource state will transition to ‘Standby - In Sync’ on node2.
    
    
11. Once system replication from node1 to node2 is in-sync, perform a switchover of the SAP HANA resource hierarchy to node2. Wait for the switchover to complete before proceeding to the next step.

*SAP recommends the use of “Takeover with Handshake” in this step to minimize downtime during the switchover.

12. Stop LifeKeeper on node1:
    /opt/LifeKeeper/bin/lkstop -f
    
    
13. Stop the SAP HANA database on node1:
    su - <sid>adm -c "sapcontrol -nr <InstNum> -function StopSystem HDB"
    
    
14. Verify that the SAP HANA database is stopped on node1 (the HDB Daemon process status should be ‘Stopped’):
    su - <sid>adm -c "sapcontrol -nr <InstNum> -function GetProcessList"
    
    
15. Perform SAP HANA database patching procedures on node1. Use the --hdbupd_server_nostart parameter of hdblcm to prevent the database from automatically restarting after patching is complete. Note that if the database is automatically restarted in this step, then it will have to be stopped again in order for LifeKeeper to register node1 as a secondary system replication site.
    ./hdblcm --action=update --hdbupd_server_nostart
    
    
16. Start LifeKeeper on node1:
    /opt/LifeKeeper/bin/lkstart
    
    
17. Re-enable resource health monitoring (quickCheck processes) on node2:
    
    1. Set LKCHECKINTERVAL=<Previous value from step 9(a)> in /etc/default/LifeKeeper on node2.
    2. Restart the lkcheck daemon by running ‘killall lkcheck’ on node2.
       
       
18. LifeKeeper will automatically register node1 as a secondary replication site and restart the database. Once this happens, wait for system replication from node2 to node1 to synchronize. Once in sync, the LifeKeeper SAP HANA resource state will transition to ‘Standby - In Sync’ on node1, at which point the cluster has returned to a highly-available state.
    
    
19. If desired, perform a switchover of the SAP HANA resource hierarchy back to the original primary server node1.