To perform an upgrade from Oracle Key Vault release 18.x, you must upgrade each multi-master cluster node.

There are different steps for upgrading the multi-master cluster depending on your deployment. A 2-node cluster, running Oracle Key Vault release 18.5 or earlier, configured as a single read-write pair would involve running a pre-upgrade script which no other deployment requires. Multi-master cluster nodes deployed in a read-write configuration must follow different upgrade steps than those deployed as read-only nodes.

Oracle does not support direct upgrades from Oracle Key Vault release 18.1 or earlier. You must upgrade to Oracle Key Vault release 18.2 or later before upgrading to release 21.4.

The upgrade process involves performing the upgrade on each multi-master cluster node. After you have begun a cluster upgrade, ensure that you upgrade all the nodes in the cluster one after the other, without too much intervening time between upgrades of two nodes.

Upgrading an Oracle Key Vault multi-master cluster includes upgrading each cluster node to the new later version. You must upgrade all nodes to the same Oracle Key Vault version. You should first upgrade the read-only nodes of the cluster, and then upgrade the read-write pairs. As each cluster node is upgraded, its node version is updated to the new version of the Oracle Key Vault. After you complete the upgrade of all cluster nodes, the cluster version is updated to the new version of the Oracle Key Vault. You can check the node version or the cluster version by selecting the Cluster tab, then in the left navigation bar, selecting Management. Oracle Key Vault multi-master cluster upgrade is considered complete when node version and cluster version at each cluster node is updated to the latest version of Oracle Key Vault.

Before you perform the upgrade, note the following:

• Perform the entire upgrade process on all multi-master cluster nodes, without interruption. That is, after you have started the cluster upgrade process, ensure that you try and upgrade all nodes, individually one after the other or in read-write pairs. Do not perform any critical operations or make configurational changes to Oracle Key Vault until you have completed upgrading all the nodes in your environment.
• Be aware that you cannot use any new features that were introduced in this release until you have completed upgrading all of the multi-master cluster nodes. An error is returned when such features are used from the node that has been upgraded. Oracle recommends that you plan the upgrade of all cluster nodes close to each other to ensure availability of the new features sooner.
• Starting in Oracle Key Vault release 21.2, expiration alerts for deactivated or destroyed objects are not generated. If you are upgrading from Oracle Key Vault release 21.1 or earlier, then the following behavior is expected:
  • As each cluster node is upgraded, Oracle Key Vault deletes all expiration alerts for any certificate and secret objects, as well as for key objects that have been revoked or destroyed.
  • Cluster nodes that have not been upgraded yet will continue to generate alerts for these same objects, and also send email notifications for these alerts. This behavior that results in deletion and recreation of alerts may repeat until the last cluster node is upgraded.
  • After the upgrade is complete, expiration alerts for the certificate and secret objects will have the alert type of Certificate Object Expiration and Secret Object Expiration, respectively.

Similar to a standalone or primary-standby environment, you must perform pre-upgrade tasks such as backing up the Oracle Key Vault server.

1. In the server where Oracle Key Vault is installed, log in as user support, and then switch to the root user.
2. Back up the server so that you can recover data in case the upgrade fails.
3. Ensure that no full or incremental backup jobs are running. Delete all scheduled full or incremental backup jobs before the upgrade.
4. Plan for downtime according to the following specifications:

   Oracle Key Vault Usage	Downtime required
   Wallet upload or download	NO
   Java Keystore upload or download	NO
   Transparent Data Encryption (TDE) direct connect	YES (NO with persistent cache)
   Primary Server Upgrade in a primary-standby deployment	YES (NO with persistent cache)

   If Oracle Key Vault uses an online master encryption key, then plan for a downtime of 15 minutes during the Oracle Database endpoint software upgrades. Database endpoints can be upgraded in parallel to reduce total downtime.

5. If the Oracle Key Vault system has a syslog destination configured, ensure that the remote syslog destination is reachable from the Oracle Key Vault system, and that logs are being correctly forwarded. If the remote syslog destination is not reachable from the Oracle Key Vault system, then the upgrade process can become much slower than normal.
6. Check the disk size before you begin the upgrade. If any of the nodes in question have a disk size that is greater than 2 TB and uses BIOS boot mode, then you cannot upgrade that system to the new release. Oracle recommends that you remove the node from the cluster and if possible, replace it with a node whose disk is less than 2 TB in size.
7. Check the boot partition size. If any of the nodes in question have a boot partition that is less than 500 MB, then you cannot upgrade that system to the new release. You can check this size as follows:
   1. Mount the /boot partition.

      /bin/mount /boot

   2. Check the Size column given by the following command:

      /bin/df -h /boot

   3. Unmount the /boot partition:

      /bin/umount /boot

   If the boot partition given by this command shows less than 488 MB, then you cannot upgrade to the current release. Oracle recommends that you remove the node from the cluster and if possible, replace it with a node that has been freshly installed with the same Oracle Key Vault version as the rest of the cluster nodes.
8. Increase the Maximum Disable Node Duration setting as appropriate so that any disabled cluster nodes have sufficient time to be upgraded then enabled back into the cluster. Note that increasing the Maximum Disable Node Duration setting also increases disk space usage.
9. For each node that had been integrated with the Oracle Audit Vault in Oracle Key Vault release 21.2 or earlier, do the following to disable and remove the Oracle Audit Vault integration:
   1. Disable the Oracle Audit Vault integration: Log into the Oracle Key Vault management console as a System Administrator, select the System tab and then System Settings from the left navigation bar. In the Audit Vault integration pane that appears, disable Oracle Audit Vault. Click Save.
   2. Log in to the Oracle Key Vault server through SSH as user support, switch user su to root and then switch user su to oracle.
   3. Stop the agent by executing the following command:

      agent_installation_directory/bin/agentctl stop

   4. Log in to the Oracle Audit Vault Server console as an Oracle Audit Vault administrator.
   5. Delete the corresponding agent and target.
   6. Log in to the Oracle Key Vault server through SSH as user support, then switch user su to root.
   7. Delete the installation directory for the Oracle Audit Vault agent.
10. Ensure that the Oracle Key Vault server certificate has not expired, nor is close to expiry, before you begin the upgrade.
    You can find how much time the Oracle Key Vault server certificate has before it expires by checking the OKV Server Certificate Expiration setting on the Configure Alerts page in the Oracle Key Vault management console.

Before upgrading to Oracle Key Vault release 21.4 from release 18x, you will need to extend the vg_root to increase disk space.

1. Log in to the server for which you will perform the upgrade and switch user as root.
2. Ensure that the persistent cache settings for Oracle Key Vault have been set.
   You will need to ensure that the persistent cache has been enabled because in a later step in this procedure, you must shut down the server. Shutting down the Oracle Key Vault server will incur downtime. To avoid any downtime, Oracle recommends that you turn on persistent cache.
3. Run the vgs command to determine the free space

   vgs

   The VFree column shows how much free space you have (for example, 21 GB).

4. Disable the node.
   Select the Cluster tab, and then Management in the left navigation bar. Under Cluster Details, select check box for the node to disable, and then click Disable. The node's status will change from DISABLING to DISABLED.
5. Power off the server in order to add a new disk

   /sbin/shutdown -h now

6. Add a new disk to the server with a capacity of 100 GB or greater
7. Start the server.
8. Log in to the Oracle Key Vault server through SSH as user support, then switch user su to root.

   ssh support@okv_server_IP_address
   su - root
   

9. Stop the Oracle Key Vault services.

   service tomcat stop;
   service httpd stop;
   service kmipus stop;
   service kmip stop;
   service okvogg stop;
   service javafwk stop;
   service monitor stop;
   service controller stop;
   service dbfwlistener stop;
   service dbfwdb stop;
   service rsyslog stop;
   

10. Run the fdisk -l command to find if there are any available partitions on the new disk.

    fdisk -l

    At this stage, there should be no available partitions.
11. Run the fdisk disk_device_to_be_added command to create the new partition.
    For example, to create a disk device named /dev/sdb:

    fdisk /dev/sdb

    In the prompts that appear, enter the following commands in sequence:

    • n for new partition
    • p for primary
    • 1 for partition number
    • Accept the default values for cylinder (press Enter twice).
    • w to write and exit
12. Use the pvcreate disk_device_partition command to add the newly added disk to the physical volume.
    For example, for a disk device named /dev/sdb1, which is the name of the disk partition that you created (based on the name used for the disk device that was added).

    pvcreate /dev/sdb1

    Output similar to the following appears:

    Physical volume "/dev/sdb1" successfully created

13. Use the vgextend vg_root disk_device_partition command to extend the logical volume with this disk space that you just added.
    For example, for the partition /dev/sdb1, you would run:

    vgextend vg_root /dev/sdb1

    Output similar to the following appears:

    Volume group "vg_root" successfully extended

14. Run the vgs command again to ensure that VFree shows an increase of 100 GB.

    vgs

    Output similar to the following appears:

    VG      #PV #LV #SN Attr   VSize   VFree
    vg_root   2  12   0 wz--n- 598.75g <121.41g
    

15. Restart the Oracle Key Vault server.

    /sbin/reboot

    After you have successfully added the disk, re-enable the node. After you re-enable the disabled multi-master cluster node, its status changes from DISABLED to ENABLING, then to ACTIVE. The status of the node will remain at ENABLING and will not change to ACTIVE unless bidirectional replication between it and all other nodes is occurring successfully.

Depending on your multi-master cluster configuration, you must follow the steps that are specific to your deployment.

• About Upgrading Multi-Master Clusters
  When upgrading a multi-master cluster, you may upgrade the read-only nodes one after the other and in the case of read-write pairs, you must upgrade both of the nodes simultaneously.
• Upgrading Multi-Master Cluster Read-Only Nodes
  Before upgrading multi-master cluster read-only nodes, ensure that you understand the requirements for performing this kind of upgrade.
• Upgrading Multi-Master Cluster Read-Write Pairs
  Before upgrading multi-master cluster read-write pairs, ensure that you understand the requirements for performing this kind of upgrade.

After you complete the upgrade of at least one node, you can log into any of the upgraded nodes to check the node and cluster versions.

1. Log in to the Oracle Key Vault management console as a user who has the System Administrator role.
2. Select the Cluster tab.
3. In the left navigation bar, select Management.
4. Check the following areas:
   • To find the node version, check the Cluster Details area.
   • To find the cluster version, check the Cluster Information area.

Nodes that were created in Oracle Key Vault releases earlier than release 21.1 use Classic mode, in which only one network interface was used.

If you prefer to use dual NIC network mode, which supports the use two network interfaces, then you can switch the node to use this mode, from the command line.

If necessary, extend the swap space on each node. Oracle Key Vault release 21.5 requires a hard disk size greater than or equal to 1 TB in size with approximately 64 GB of swap space.

1. Log in to the server in which you upgraded Oracle Key Vault and connect as root.
2. Check the current amount of swap space.

   [root@my_okv_server support]# swapon -s

   Output similar to the following appears. This example shows that the system has 4 GB of swap space.

   Filename Type Size Used Priority
   /dev/dm-0 partition 4194300 3368 -1
   

   There must be 64 GB of swap space if the disk is greater than 1 TB in size.

3. Run the vgs command to determine how much free space is available.

   vgs

   The VFree column shows how much free space you have (for example, 21 GB).

4. Power off the server in order to add a new disk.

   /sbin/shutdown -h now

5. Add a new disk to the server of a size that will bring the VFree value to over 64 GB.
6. Start the server.
7. Log in to the Oracle Key Vault server through SSH as user support, then switch user su to root.

   ssh support@okv_server_IP_address
   su - root
   

8. Run the fdisk -l command to find if there are any available partitions on the new disk.

   fdisk -l

   At this stage, there should be no available partitions.

9. Run the fdisk disk_device_to_be_added command to create the new partition.
   For example, to create a disk device named /dev/sdc:

   fdisk /dev/sdc

   In the prompts that appear, enter the following commands in sequence:

   • n for new partition
   • p for primary
   • 1 for partition number
   • Accept the default values for cylinder (press Enter twice).
   • w to write and exit
10. Use the pvcreate disk_device_partition command to add the newly added disk to the physical volume.
    For example, for a disk device named /dev/sdc1, which is the name of the disk partition that you created (based on the name used for the disk device that was added).

    pvcreate /dev/sdc1

    Output similar to the following appears:

    Physical volume "/dev/sdc1" successfully created

11. Use the vgextend vg_root disk_device_partition command to extend the logical volume with this disk space that you just added.
    For example, for the partition /dev/sdc1, you would run:

    vgextend vg_root /dev/sdc1

    Output similar to the following appears:

    Volume group "vg_root" successfully extended

12. Run the vgs command again to ensure that VFree shows an increase of 64 GB.

    vgs

13. Disable swapping.

    [root@my_okv_server support]# swapoff -v /dev/vg_root/lv_swap

14. To extend the swap space, run the lvresize command.

    [root@my_okv_server support]# lvresize -L +60G /dev/vg_root/lv_swap

    Output similar to the following appears:

    Size of logical volume vg_root/lv_swap changed from 4.00 GiB (128 extents) to 64.00 GiB (2048 extents)
    Logical volume lv_swap successfully resized.
    

15. Format the newly added swap space.

    [root@my_okv_server support]# mkswap /dev/vg_root/lv_swap

    Output similar to the following appears:

    mkswap: /dev/vg_root/lv_swap: warning: don't erase bootbits sectors
    on whole disk. Use -f to force.
    Setting up swapspace version 1, size = 67108860 KiB
    no label, UUID=fea7fc72-0fea-43a3-8e5d-e29955d46891
    

16. Enable swapping again.

    [root@my_okv_server support]# swapon -v /dev/vg_root/lv_swap

17. Verify the amount of swap space that is available.

    [root@my_okv_server support]# swapon -s

    Output similar to the following appears:

    Filename Type Size Used Priority 
    /dev/dm-0 partition 67108860 0 -1

18. Restart the Oracle Key Vault server.

    /sbin/reboot

    For primary-standby deployments, ensure that the primary and standby nodes sync up before proceeding further with next steps.

For each multi-master cluster node, Oracle recommends that you clean up the older kernels that were left behind after the upgrade.

1. Log in to the Oracle Key Vault server as the support user.
2. Switch to the root user.

   su - root

3. Mount /boot if it was not mounted on the system.
   1. Check if the /boot is mounted. The following command should display /boot information if it was mounted.

      df -h /boot;

   2. Mount it if /boot is not mounted.

      /bin/mount /boot;

4. Check the installed kernels and the running kernel.
   1. Search for any kernels that are installed.

      rpm -q kernel-uek | sort;

      The following example output shows that two kernels are installed:

      kernel-uek-4.14.35-2047.504.2.el7uek.x86_64 
      kernel-uek-5.4.17-2136.304.4.5.el7uek.x86_64

   2. Check the latest kernel.

      uname -r;

      The following output shows an example of a kernel version that was installed at the time:

      5.4.17-2136.304.4.5.el7uek.x86_64

      # uname-r

      5.4.17-2136.304.4.5.el7uek.x86_64

      This example assumes 5.4.17-2136.304.4.5.el7uek.x86_64 as the latest version (newer versions may be available by now). Based on the output from the commands above, remove the older kernel (kernel-uek-4.14.35-2047.504.2.el7uek.x86_64). You should remove all kernels that are older than the latest kernel.

5. Remove the older kernel and its associated RPMs.

   For example, to remove the kernel-uek-4.14.35-2047.504.2.el7uek.x86_64 kernel:

   yum --disablerepo=* remove `rpm -qa|grep 4.14.35-2047.504.2.el7uek`

   Output similar to the following appears:

     Resolving Dependencies
   -->   Running transaction check
   ---> Package kernel-uek.x86_64 0:4.14.35-2047.504.2.el7uek will be erased
   ---> Package kernel-uek-devel.x86_64 0:4.14.35-2047.504.2.el7uek will be erased
   --> Finished Dependency Resolution
   
   Dependencies Resolved
   
   =================================================================================================================
    Package               Arch        Version                           Repository                          Size
   =================================================================================================================
   Removing:
    kernel-uek          x86_64         4.14.35-2047.504.2.el7uek       @anaconda/7.7                        58 M
    kernel-uek-devel    x86_64         4.14.35-2047.504.2.el7uek       @avs-ol-dependencies                 63 M
   
   Transaction Summary
   =================================================================================================================
   Remove        2 Package(s)
   
   Installed size: 121 M
   Is this ok [y/N]:

6. Enter y to accept the deletion output.
7. Repeat these steps starting with Step 4 for all kernels that are older than the latest kernel.

For each multi-master cluster node, you should remove SSH-related DSA keys left behind after the upgrade, because they can cause problems with some code analysis tools.

1. Log in to the Oracle Key Vault management console as a user with the System Administrator role.
2. Enable SSH.

   Log in to the Oracle Key Vault management console as a user who has the System Administrator role. Select the System tab, then Settings. In the Network Details area, click SSH Access. Select IP address(es) and then enter only the IP addresses that you need, or select All. Click Save.

3. Login to the Oracle Key Vault support account using SSH.

   ssh support@OracleKeyVault_serverIPaddress

4. Switch to the root user.

   su - root

5. Change directory to /etc/ssh.

   cd /etc/ssh

6. Rename the following keys.

   mv ssh_host_dsa_key.pub ssh_host_dsa_key.pub.retire
   mv ssh_host_dsa_key ssh_host_dsa_key.retire

7. Disable SSH access.

   Log in to the Oracle Key Vault management console as a user who has the System Administrator role. Select the System tab, then Settings. In the Network Details area, click SSH Access. Select Disabled. Click Save.