Upgrades and Reinstallation

When upgrading or reinstalling software used with Capacity Advisor, there are different considerations and procedures for the following, described in the next sections:

Updating software on the CMS
Updating agents on the managed nodes

Upgrading Software on the CMS

When VSE is removed and reinstalled, Capacity Advisor can miss collecting some data on the Central Management System (CMS). To avoid this, back up the VSE Management Software database before removing VSE and restore the database after it is reinstalled, using the following commands:

vseinitconfig —b (before removing VSE)
vseinitconfig —s (after reinstalling VSE)

For more information about using vseinitconfig and restoring the VSE Management Software database, see the appropriate VSE Management Software installation guide at http://docs.hp.com/en/vse.html.

Upgrading Agents on the Managed Nodes

When upgrading the Utilization Provider to version 1.6.0.0 on a managed node, the data collected on that node that day is lost. Only data after the upgrade is kept. For example, if the Utilization Provider upgrade occurs at 3:00 pm UTC on a managed node, the previous 15 hours of data are lost for that node. No data from prior days is lost. The data-gathering "day" starts at midnight UTC. If the upgrade of the Utilization Provider occurs at noon Central Daylight Time (UTC+5), 17 hours of utilization data are lost. To minimize data loss, upgrade the Utilization Provider as soon after UTC midnight as possible.

To minimize missing data, you may use the following procedure:

Procedure 1-1 To Upgrade an Agent on a Managed Node from the Command Line

Collect data from the managed node. For example,
$ capcollect system.to.be.upgraded

Export the utilization data for this day from the system to save it. For example,

$ capprofile -x -b 20070309 -e 20070310 system.to.be.upgraded >/var/tmp/system.profile

Upgrade the utilization provider software on the managed node shortly after UTC midnight; for example, on Monday.
Allow capcollect to collect the partial day from the new provider after the UTC day is over, that is, after UTC midnight, say; for example, on Tuesday.
IMPORTANT: This step is crucial, because if you re-import your exported data before the UTC day is over, capcollect will overwrite it when it collects for the day.

Import the saved profile data for the system. For example,

$ capprofile -i -o system.to.be.upgraded </var/tmp/system.profile

Using this procedure is practical only if the cost of losing a day's data is greater than the cost of using the procedure.

Upgrades and Reinstallation

Technical documentation

» Table of Contents

» Glossary

» Index

Upgrading Software on the CMS

Upgrading Agents on the Managed Nodes