Follow

CONF-0008: Upgrading Kinetica

Overview:

Keeping the Kinetica application up-to-date in terms of versioning is very important in order to keep up with bug fixes, enhancements, as well as security. This article will cover how to upgrade Kinetica against the various supported OS platforms. 

 

Action:

 

CentOS

  1. Retrieve the needed RPM file from the Public Repository:
    wget http://repo.kinetica.com/path/to/needed/upgrade/file.rpm
  2. Stop ALL services on the head node:

    /opt/gpudb/core/bin/gpudb stop all
  3. Ensure ALL services have been successfully stopped. DO NOT proceed to the next step until verifying the services are fully stopped on the Head Node:
    /opt/gpudb/core/bin/gpudb status 
    
    Host Manager     : Stopped
    Tomcat           : Stopped
    Stats            : Stopped
    GPUdb            : Stopped
    Text             : Internal
    Httpd            : Disabled
    ODBC connector   : Stopped
    Reveal           : Stopped
  4. If the environment is clustered, use the stop all command from Step 2 to then stop the services on the worker node(s). 
  5. Use the Status command from Step 3 to ensure the services are fully stopped on the worker node(s).
  6. Perform the upgrade on all nodes using the below command. Please note, the order in which the nodes are upgraded has no impact on the process:
    sudo yum install gpudb-<gpuhardware>-<licensetype>-<version>-<release>.<architecture>.rpm
  7. The upgrade process should start Host Manager autonomously on each node. Using the status command, verify Host Manager is in fact running on all nodes before proceeding:
    /opt/gpudb/core/bin/gpudb status
    
    Host Manager : Running (PID)
    Tomcat : Running (PID)
    Stats : Running (PID)
    GPUdb : Stopped
    Text : Internal
    Httpd : Disabled
    ODBC connector : Stopped
    Reveal : Stopped
  8. ONLY after we have Host Manager running on all nodes, do we then start the remaining services. If clustered, issue the below command on the head node ONLY:
    /opt/gpudb/core/bin/gpudb start all
  9. Using the status command, verify all services are up and running:
    /opt/gpudb/core/bin/gpudb status 
    
    Host Manager : Running (PID)
    Tomcat : Running (PID)
    Stats : Running (PID)
    StatsCollector : Running (PID)
    GPUdb : Running (PID)
    Text : Internal
    Httpd : Disabled
    ODBC connector : Running (PID)
    Reveal : Running (PID)
  10. Kinetica is now ready for use!

 

Ubuntu/Debian 

  1. Retrieve the needed RPM file from the Public Repository:
    wget http://repo.kinetica.com/path/to/needed/upgrade/file.rpm
  2. Stop ALL services on the head node:

    /opt/gpudb/core/bin/gpudb stop all
  3. Ensure ALL services have been successfully stopped. DO NOT proceed to the next step until verifying the services are fully stopped on the Head Node:
    /opt/gpudb/core/bin/gpudb status 
    
    Host Manager     : Stopped
    Tomcat           : Stopped
    Stats            : Stopped
    GPUdb            : Stopped
    Text             : Internal
    Httpd            : Disabled
    ODBC connector   : Stopped
    Reveal           : Stopped
  4. If the environment is clustered, use the stop all command from Step 2 to then stop the services on the worker node(s). 
  5. Use the Status command from Step 3 to ensure the services are fully stopped on the worker node(s).
  6. Perform the upgrade on all nodes using the below command. Please note, the order in which the nodes are upgraded has no impact on the process:
    sudo dpkg install gpudb-<gpuhardware>-<licensetype>-<version>-<release>.<architecture>.rpm
  7. The upgrade process should start Host Manager autonomously on each node. Using the status command, verify Host Manager is in fact running on all nodes before proceeding:
    /opt/gpudb/core/bin/gpudb status
    
    Host Manager : Running (PID)
    Tomcat : Running (PID)
    Stats : Running (PID)
    GPUdb : Stopped
    Text : Internal
    Httpd : Disabled
    ODBC connector : Stopped
    Reveal : Stopped
  8. ONLY after we have Host Manager running on all nodes, do we then start the remaining services. If clustered, issue the below command on the head node ONLY:
    /opt/gpudb/core/bin/gpudb start all
  9. Using the status command, verify all services are up and running:
    /opt/gpudb/core/bin/gpudb status 
    
    Host Manager : Running (PID)
    Tomcat : Running (PID)
    Stats : Running (PID)
    StatsCollector : Running (PID)
    GPUdb : Running (PID)
    Text : Internal
    Httpd : Disabled
    ODBC connector : Running (PID)
    Reveal : Running (PID)
  10. Kinetica is now ready for use!

 

SLES

  1. Retrieve the needed RPM file from the Public Repository:
    wget http://repo.kinetica.com/path/to/needed/upgrade/file.rpm
  2. Stop ALL services on the head node:

    /opt/gpudb/core/bin/gpudb stop all
  3. Ensure ALL services have been successfully stopped. DO NOT proceed to the next step until verifying the services are fully stopped on the Head Node:
    /opt/gpudb/core/bin/gpudb status 
    
    Host Manager     : Stopped
    Tomcat           : Stopped
    Stats            : Stopped
    GPUdb            : Stopped
    Text             : Internal
    Httpd            : Disabled
    ODBC connector   : Stopped
    Reveal           : Stopped
  4. If the environment is clustered, use the stop all command from Step 2 to then stop the services on the worker node(s). 
  5. Use the Status command from Step 3 to ensure the services are fully stopped on the worker node(s).
  6. Perform the upgrade on all nodes using the below command. Please note, the order in which the nodes are upgraded has no impact on the process:
    sudo zypper install gpudb-<gpuhardware>-<licensetype>-<version>-<release>.<architecture>.rpm
  7. The upgrade process should start Host Manager autonomously on each node. Using the status command, verify Host Manager is in fact running on all nodes before proceeding:
    /opt/gpudb/core/bin/gpudb status
    
    Host Manager : Running (PID)
    Tomcat : Running (PID)
    Stats : Running (PID)
    GPUdb : Stopped
    Text : Internal
    Httpd : Disabled
    ODBC connector : Stopped
    Reveal : Stopped
  8. ONLY after we have Host Manager running on all nodes, do we then start the remaining services. If clustered, issue the below command on the head node ONLY:
    /opt/gpudb/core/bin/gpudb start all
  9. Using the status command, verify all services are up and running:
    /opt/gpudb/core/bin/gpudb status 
    
    Host Manager : Running (PID)
    Tomcat : Running (PID)
    Stats : Running (PID)
    StatsCollector : Running (PID)
    GPUdb : Running (PID)
    Text : Internal
    Httpd : Disabled
    ODBC connector : Running (PID)
    Reveal : Running (PID)
  10. Kinetica is now ready for use!

 

Additional Information:

Upgrading Kinetica will not remove any user/group data that has been configured in the environment or affect any passwords. Also, persist data will not be deleted or affected.

Was this article helpful?
1 out of 1 found this helpful
Have more questions? Submit a request

0 Comments

Article is closed for comments.