User`s guide

21
Upgrading Nodes in Clusters
a If you used the Parallels Infrastructure Manager application (renamed to Parallels Virtual
Automation in Parallels Virtuozzo Containers 4.7) for administering your cluster nodes, you
also need to upgrade this application. First of all, update Parallels Infrastructure Manager to
its latest version. When installing Parallels Virtuozzo Containers 4.7, specify the --pva-
agent option, for example:
# ./vzinstall-linux-x86_64.bin install --vzinstall-opts "--clustermode --pva-agent"
or
# ./install --clustermode --pva-agent
6 Once the installation is complete, restart the node. If you chose to upgrade Parallels Virtual
Automation, wait for it to get upgraded.
7 Make sure that the /vz directory does not contain any files or folders.
8 Uncomment the /vz entry to the /etc/fstab file.
9 Enable the startup of the cman and rgmanager services, and disable the startup of the vz,
pvapp, and pvaagentd services:
# chkconfig cman on
# chkconfig rgmanager on
# chkconfig vz off
# chkconfig pvapp off
# chkconfig pvaagentd off
For detailed information on installing Parallels Virtuozzo Containers 4.7, see the Parallels Virtuozzo
Containers 4.7 Installation Guide. For upgrade details, see Upgrading from SWsoft Virtuozzo 3.0
Service Pack 1 and Upgrading from Parallels Virtuozzo Containers 4.0 and 4.6 (p. 8).
Finishing the Upgrade
Once you upgrade all active and passive nodes, configure the cluster configuration file
(/etc/cluster/cluster.conf) as follows:
1 Add to the <resources> section these lines:
<script file="/etc/init.d/pvaagentd" name="pvascr"/>
<script file="/etc/init.d/pvapp" name="ppscr"/>
2 Add to the <service> section of each clustered service these lines:
<script ref="pvascr"/>
<script ref="ppscr"/>
3 Increase the value of the config_version parameter by 1.
4 Save the file, and run the ccs_tool utility to update the cluster configuration.
5 Restart all cluster nodes one by one.
Finally, you need to unfreeze the clustered services on all active nodes:
# clusvcadm -U clustered_service_name
If you used the -l option to lock the clustered service, specify the -u option after clusvcadm to
unlock the service. Unlock all services on all nodes where you locked them.