Data Aggregator / Data Repository Vertica Performance Issues After Upgrade to 2.2.0 or 2.2.1

book

Article ID: 48844

calendar_today

Updated On:

Products

CA Infrastructure Performance

Issue/Introduction

Description:

Customers upgrading to release CA Infrastructure Manager 2.2.0 or 2.2.1 and above might notice issues with the Vertica database:

Symptoms include:

  • Dashboards in Performance Center, that report on Data Aggregator data, progressively slow down over time
  • Data Repository/vertica processes show consistent high CPU utilization
  • Data Repository average CPU usage increases over time.
  • Data Repository memory usage increases over time
  • Failures when synchronizing Performance Center and Data Aggregator

Solution:

To resolve these issues the Vertica database must be upgraded. Please follow the instructions below:

  1. Open an issue with support to get the vertica-6.0.2-0.x86_64.RHEL5.rpm. Then download it to your data repository installation folder previously created by the InstallDR.bin installer, for example: /opt/CA/IMDataRepository_vertica6

  2. Perform a manual backup of your Data Repository, type the following command:

    /opt/vertica/bin/vbr.py --task backup --config-file configuration_filename
    configuration_filename

    Indicates the directory path and filename of the configuration file you created previously. This file is located where you ran the backup utility (/opt/vertica/bin/vbr.py).

    For example:

    /opt/vertica/bin/vbr.py --task backup --config-file /home/vertica/vert-db-production.ini

    If you are prompted about the authenticity of the host, answer yes.

    Note: In a cluster installation, you only have to perform this step on one of the hosts that are participating in the cluster.

    Important! If you back up Data Repository and have not been regularly backing up Data Repository previously, it can take several hours or more to back up Data Repository

  3. As root, stop the Data Aggregator by running "/etc/init.d/dadaemon stop" from the data aggregator

  4. As the Linux database administrator user, log in to the database server you use for Data Repository as the database administrator user and stop Data Repository:

    1. Type the following command: /opt/vertica/bin/adminTools

    2. The Administration Tools dialog opens.

      1. Select (4) Stop Database.

      2. Press the Space bar next to the database name, select OK, and press Enter.

    3. You are prompted for the database password.

      1. Enter the database password and press Enter. The Data Repository database stops.

  5. Note: If Data Repository does not stop, select (2) Stop Vertica on Host from the (7) Advanced Tools Menu. In a cluster environment, select (2) Stop Vertica on Host for each host in the cluster. If Data Repository still does not stop, select (3) Kill Vertica Process on Host in the Advanced menu. In a cluster environment, you may need to select (3) Kill Vertica Process on Host on more than one host in the cluster.

  6. Determine which hosts Vertica is running on by doing the following steps:

    1. Select option 6 (Configuration Menu).

    2. Select option 3 (View Database).

    3. Select the database.

    4. Make a note of the hostnames. You will need these hostnames later in this procedure.

    5. Exit the adminTools utility after Data Repository stops.

  7. Log on to the computer where Data Repository is installed as the root user.

  8. To run the Data Repository installation package, type the following command as the root user:

    1. rpm -Uvh installation_directory/vertica-6.0.2-0.x86_64.RHEL5.rpm

      For example:

      rpm -Uvh /opt/CA/IMDataRepository_vertica6/vertica-6.0.2-0.x86_64.RHEL5.rpm

  9. To invoke the Data Repository installer, type the following command:

    1. /opt/vertica/sbin/install_vertica -s hostname[,hostname,hostname] -u dr_username -l home_directory -L /opt/CA/IMDataRepository_vertica6/vlicense.dat [-r installation_directory]

      -s hostname[,hostname,hostname]

      Indicates the name of the hosts where you want to install Data Repository. For a single-node installation, specify one hostname. For a cluster installation, specify a comma-separated list of three hostnames without whitespace before or after the commas. Each host in the cluster is set up for the installation.

      Note: Do not use 'localhost' in the case of a single-node installation. Specify the actual hostname of the computer.

      -u dr_username

      Indicates the username of the vertica Linux database administrator user account you created previously.

      -l home directory

      Indicates the Linux user home directory.

      [-r installation_directory]

      (Cluster installation only) Indicates the installation directory where the Data Repository installation package resides. You extracted the license file and the Data Repository installation package to this installation directory.

      For example:

      /opt/vertica/sbin/install_vertica -s drhost -u dradmin -l /opt/dradmin -L /tmp/vlicense.dat -r /opt/CA/IMDataRepository_vertica6/vertica-6.0.2-0.x86_64.RHEL5.rpm

  10. As the linux database administrator (for example, dradmin), Llog on to the computer where Data Repository is installed.

  11. Verify that you installed Data Repository correctly by doing the following steps:

    1. Type the following command: /opt/vertica/bin/adminTools

    2. The Administration Tools dialog opens.

    3. Verify that the top of the banner indicates that the database version is 6.0.2-0.

  12. (Single-Node and Cluster Installations) As the root user, Sstop the vertica_agent on each computer where you installed Data Repository. This agent is unnecessary and consumes CPU:

    1. Open a command prompt and type the following command: service vertica_agent stop

    2. Type the following command: chkconfig --del vertica_agent

  13. As the linux database administrator, launch adminTools again.

    1. /opt/vertica/bin/adminTools

    2. Restart Data Repository by selecting option 3 (Start Database) from the main menu of the Administration Tools dialog.

  14. As the root user, restart the Data Aggregator with the following command: "/etc/init.d/dadaemon start" from the data aggregator host.

Environment

Release:
Component: IMAGGR