HP Serviceguard manual Types of Upgrade, How To Tell when the Cluster Re-formation Is Complete

CAUTION: From the time when the old cluster manager is shut down until the new cluster manager forms its first cluster, a node failure will cause the entire cluster to fail. HP strongly recommends that you use no Serviceguard commands other than cmviewcl (1m) until the new cluster manager successfully completes its first cluster re-formation. In the case of a Rolling Upgrade (page 358), this happens when the cmrunnode for the last node executes; in a Non-Rolling Upgrade (page 359), it happens during the initial cmruncl. See “How To Tell when the Cluster Re-formation Is Complete” for more information. For an upgrade to Serviceguard A.11.19, this recommendation is additional to the Limitations of Rolling Upgrades (page 360) and Limitations of Non-Rolling Upgrades (page 366), and supersedes any “Limitation” that is more lenient.

There may be further caveats that apply to specific upgrade paths; before you proceed, make sure you read the “Announcements”, “Compatibility”, and “Installing Serviceguard” sections of the latest version of the HP Serviceguard Version A.11.19 Release Notes at http://www.hp.com/go/ hpux-serviceguard-docs .

If you are using a Quorum Server, make sure you also read the latest version of the HP Serviceguard Quorum Server A.04.00 Release Notes before you proceed. These are also posted on posted at http://www.hp.com/go/hpux-serviceguard-docs, under HP Serviceguard Quorum Server Software.

How To Tell when the Cluster Re-formation Is Complete

You can tell when the cluster has re-formed under the new cluster manager by monitoring syslog on any node. syslog will record that Serviceguard has started and finished upgrading the node to the new cluster manager (Cluster Membership Protocol version 2), and then, once all nodes have been upgraded, you will see a message indicating that the new cluster has formed. Watch for three messages similar to the following (for clarity, intervening messages have been replaced with ellipses [...]):

Nov 14 13:52:46 bbq1 cmcld[20319]: Starting to upgrade this node to Cluster Membership Protocol version 2 [....]

Nov 14 13:52:46 bbq1 cmcld[20319]: Finished upgrading this node to Cluster Membership Protocol version 2 [....]

Nov 14 13:52:47 bbq1 cmcld[20319]: Membership: membership at 3 is FORMED (coordinator 1) includes: 1 3 2 4 excludes

When you see the Membership: message, and specifically the capitalized word FORMED in this message, the transition to the new cluster manager is complete and you can use all the Serviceguard commands.

Types of Upgrade

Rolling Upgrade

In a rolling upgrade, you upgrade the HP-UX operating system (if necessary) and the Serviceguard software one node at a time without bringing down your cluster. A rolling upgrade can also be done any time one system needs to be taken offline for hardware maintenance or patch installations.

This method is among the least disruptive, but your cluster must meet both general and release-specific requirements. See “Guidelines for Rolling Upgrade” (page 359).

Rolling Upgrade Using DRD

DRD stands for Dynamic Root Disk. Using a Dynamic Root Disk allows you to perform the update on a clone of the root disk, then halt the node and reboot it from the updated clone root disk.

A rolling upgrade using DRD is like a rolling upgrade, but is even less disruptive because each node is down for a shorter time. It is also very safe; if something goes wrong you can roll back to the original (pre-upgrade) state by rebooting from the original disk.

This method is the least disruptive, but you need to make sure your cluster is eligible; see“Restrictions for DRD Upgrades” (page 359).

You can obtain the DRD software free from http://www.hp.com/go/drd.

358 Software Upgrades