Performance Management Data Repository install failure for readahead

Products

• CA Infrastructure Management
• CA Performance Management - Usage and Administration
• DX NetOps

Issue/Introduction

Installing a three node Performance Management (PM) Data Repository (DR) Vertica cluster.

Using the "Install the Database with Sudo User Passwordless SSH Configured" option.

The dr_validate.sh script shows all OK messages.

The dr_install.sh script fails and returns the following error message.

Prerequisites not fully met during local (OS) configuration for
verify-<Host_IP>.xml:
    FAIL (S0020): https://my.vertica.com/docs/9.1.x/HTML/index.htm#cshid=S0020
        Readahead size of sdb (/dev/sdb) is too low for typical systems: 256 <
        2048

We see the same error message from the two servers that will be nodes 2 and 3.

The server that will be node 1, the one these scripts are run on, doesn't show the same error.

Environment

All supported Performance Management releases

Cause

The dr_validate.sh script was only run once on the node meant to be node 1, the one where the dr_install.sh script will be run.

Resolution

The dr_validate.sh script, per the steps listed in the "Install the Database with Sudo User Passwordless SSH Configured" option, must be run once each on each node in the cluster being built.

Run the dr_validate.sh script on each node individually. Ensure the DbHostNames value in the drinstall.properties is the same as the host its being run on.