DPOD software updates are available from time to time and include enhancements, security updates, component upgrade, bug fixes, etc.

The following sections describe the process of upgrading an existing installation.

Upgrade Paths

The following table describes the upgrade paths between versions:

¹ Up to version 1.0.8.6, the fix pack file does not include previous fix packs, so each fix pack should be applied one by one.

² Version 1.0.12.0 requires at least 1.0.10.0 because of Store engine upgrade and data upgrade issues.

³ Version 1.0.14.0 requires at least 1.0.13.0 because of an architecture change of the Store engine.

Upgrading a Cell Environment

Note: This procedure has changed since 1.0.12.0. In previous versions the process of upgrading a cell environment was different.

To upgrade a cell environment:

1. Make sure the entire cell is up and running. Upgrading is possible only if the cell is healthy.

2. In the cell manager, execute the following command:

   curl -X PUT "montier-es-http:9200/_cluster/settings" -H 'Content-Type: application/json' -d'{"persistent":{"cluster.max_shards_per_node":"20000"}}'
   
   Expected response:
   {"acknowledged":true,"persistent":{"cluster":{"max_shards_per_node":"20000"}},"transient":{}}

3. Upgrade the cell manager.

4. When the cell manager upgrade is complete, only the following services should be up and running - MonTier-Derby, MonTier-es-raw-trans-Node-1/2/3/4 (the Web Console will be unavailable).
   Do not start the other services manually on the cell manager unless requested explicitly.

5. Copy /app/keys directory (including any sub-directories) from the cell manager to all cell members (this will share the encryption key and generated certificates between the cell components).

6. Upgrade all cell members (can be done in parallel).

7. When the cell members upgrade is complete, start the cell manager services using the Admin Console or app-util.sh ("Start All").

Copy the Software Update Files to DPOD Server

1. DPOD software updates are available through IBM Fix Central. The download consists of two files:

   1. The update file: DPOD-fixpack-<version number>.sfs

   2. The md5 hash calculation of the update file: DPOD-fixpack-<version number>.md5

2. Copy the update files to the following directory on DPOD server: /installs/update/fix/TODO

3. If the directory does not exist, execute the following command to create it:

   mkdir -p /installs/update/fix/TODO

Make Sure there is Enough Free Disk Space

Inspect the free space of each mount point. You may use the following command:

df -h

The following table lists the required available space:

If there is not enough free space on the /installs mount point, you may free some from the following directories:

Follow Special Steps

Some versions require special steps. Make sure to follow them:

Install the Software Update

1. If this DPOD server is an all-in-one or a cell manager, make sure DPOD is up and running, and that the Store status is green (under DPOD Health → Store). Upgrading is possible only if the current installation is healthy.

2. Execute the following commands as root (using sudo is not supported):

cd /installs/update/fix/TODO
MonTierUpdateInstaller.sh -u DPOD-fixpack-<version number>.sfs -s DPOD-fixpack-<version number>.md5

For example:
MonTierUpdateInstaller.sh -u DPOD-fixpack-1_0_14_0.sfs -s DPOD-fixpack-1_0_14_0.md5

Please review the upgrade log file if the console displays any error messages.

Optional Manual Steps during Software Update

Some versions require manual steps in order to complete the upgrade. These steps are described below.

Rerunning the Installer after Installer Update

Some versions make changes to the installer file itself (MonTierUpdateInstaller.sh). In that case, the following message will be displayed, and the installer should be rerun manually:

Found new version of the software update installer... updating
Software update installer was updated successfully.

Important !!! Please run the software update installer again.

Rerunning the Installer after Operating System Update

Some versions upgrade the operating system (CentOS) when DPOD is installed in Appliance Mode.

In that case, the installer should be rerun manually after the OS upgrade is complete and the server is restarted.

Running Store Data Migration Tool

Some versions include new versions of the Store engine, which might require some of the existing data to be migrated to a newer format or structure.

If the upgrade process detects data that should be migrated before applying the upgrade, the following message will be displayed, and the data migration tool should be run manually:

Some of the existing data needs to be migrated to a newer format or structure.
The data migration tool has been deployed in /installs/data-migration-tool.
Further information can be found in the documentation at: Admin Guide -> Upgrade -> Upgrade

Configuring the Data Migration Tool

You may edit the configuration file /installs/data-migration-tool/data-migration-tool.conf before running the tool.
The following entries are of interest (other settings should not change unless advised otherwise by IBM support):

1. duration.limit (default: 999999) - limits the execution time in minutes.
   This option is useful if you want to schedule the tool to run during off-peak time.
   In such a case, you can limit the tool to run for a few hours each time and schedule it to start when off-peak time starts, so performance will not be impacted during peak time hours.

2. delete.kibana_indices (default: true) - determines whether Kibana indices should be deleted or not.
   true - Delete old Kibana indices - all exiting Kibana dashboards and other settings will be deleted.
   false - Keep Kibana indices and try to migrate them to the new store engine version - however, it is up to you to fix the indices manually according to Kibana's documentation.

Running the Data Migration Tool

1. Make sure DPOD is up and running. The data migration tool is designed to work while DPOD is up and running, so users may sign in to DPOD and use it while the data migration tool is upgrading the data in the background.

2. Since the tool is a long-running process, it is recommended to run it in a "no hang-up" mode, which will cause the process to continue running even after the SSH session is closed. In this mode, the console output is written to nohup.log file in the local directory. Use the following command to run the tool and inspect the console output:

   nohup /installs/data-migration-tool/data-migration-tool.sh &
   tail -f nohup.out

3. The data migration process may take anywhere between a few minutes and a few days, depending on the amount of data, the server load and the server hardware. A rough estimation of the time left will be presented while the process is running.

4. Please review the data migration tool log file (/installs/logs/store_data_migration.log) if the console displays any error message.

Interrupting the Data Migration Tool

Pressing Ctrl+C or setting duration.limit in the configuration file will stop the tool during the migration process.

Stopping the tool will cause it to re-process the last index that was in the middle of the migration on the next run.

While this is usually not an issue, note that on some cases it may cause complications, for example:

1. The user wants the tool to run during a nightly maintenance window, between 2-4 AM.

2. The tool is scheduled using cron to 2 AM and the duration.limit setting is set to 120 minutes.

3. For this specific user, depending on its hardware and data sizes, processing of each index takes about 3 hours.

4. Since the tool is interrupted after 2 hours, on the next night, the tool will try to migrate the same index again and will never advance to the next index.