Striim 3.9.8 documentation

Preparing to upgrade and exporting the metadata
  1. Back up the cluster as described in Backing up and restoring a cluster.

  2. On each server in the cluster, copy any .jar files you have added to Striim's lib directory to another location. These should be easy to identify since the files installed by Striim all have the same date.

    Note

    Starting in 3.9.6, a custom Cassandra driver is bundled with Striim, so do not copy any cassandra-cassandra-jdbc-wrapper-<version>.jar you installed in an earlier release.

  3. Applications that have recovery enabled will pick up from the recovery checkpoint when they are restarted after the upgrade. In order for this to work correctly, after the upgrade they must be deployed on the same, unchanged deployment groups with the same options. Make notes as necessary to duplicate your deployment scenarios.

    In releases prior to 3.8.6, a flow that was deployed to a group with more than one agent was always deployed to all agents in the group, even when ON ONE was specified. This bug was fixed in 3.8.6, and deployment to agents will be ON ONE or ON ALL as specified in the DEPLOY command. Consequently, after the upgrade is complete, any applications with recovery enabled that were deployed to groups with more than one agent must be deployed ON ALL in order to pick up from the recovery checkpoint correctly.

  4. Quiesce and undeploy all running applications with persisted streams. Stop and undeploy all other running and deployed applications.

  5. Open the Striim console and enter the following commands to stop the hidden monitoring applications:

    stop application Global.MonitoringSourceApp;
    undeploy application Global.MonitoringSourceApp;
    stop application Global.MonitoringProcessApp;
    undeploy application Global.MonitoringProcessApp;
    exit;
  6. On each server in the cluster, stop the striim-node process:

    • In CentOS 6 or Ubuntu 14.04:

      sudo stop striim-node
    • In CentOS 7 or Ubuntu 16.04 or later:

      sudo systemctl stop striim-node
  7. If the metadata repository is hosted on Derby, on the server where Derby is installed, stop it:

    • In CentOS 6 or Ubuntu 14.04:

      sudo stop striim-dbms
    • In CentOS 7 or Ubuntu 16.04 or later:

      sudo systemctl stop striim-dbms
  8. Export the metadata.

    If the metadata repository is hosted on Derby, on the server that hosts Derby, enter the following commands to export the metadata:

    cd /opt/striim
    sudo bin/tools.sh -A export -F export.json
    

    If the metadata repository is hosted on Oracle, enter the following commands on any server:

    cd /opt/striim
    sudo bin/tools.sh -A export -F export.json -r oracle
    

    If the metadata repository is hosted on PostgreSQL, enter the following commands on any server:

    cd /opt/striim
    sudo bin/tools.sh -A export -F export.json -r postgres
    
  9. Move the Striim configuration file to /opt/striim/conf-backup (or, if you prefer, some other directory) so it will not be removed when you uninstall the old version of Striim:

    sudo mkdir conf-backup
    sudo mv conf/startUp.properties conf-backup
    
  10. If the metadata repository is hosted on Derby, skip this step. 

    Remove the old repository tables (which you backed up in step 1):

    If the metadata repository is hosted on Oracle, log in to sqlplus as the user created in Configuring a DBMS to host Striim's metadata repository and run /opt/striim/conf/DropMetadataReposOracle.sql.

    If the metadata repository is hosted on PostgreSQL, log in to psql as the user created in Configuring a DBMS to host Striim's metadata repository and run /opt/striim/conf/DropMetadataReposPostgressql.sql.

  11. Continue with Upgrading in CentOS or Upgrading in Ubuntu.