You can do maintenance of a veritas cluster service group, in the following manner:
1) Run: /sbin/gabconfig -c -n1 on the active service group node.
2) Edit the file /etc/gabtab on the active service group node to and insert the following:
/sbin/gabconfig -c -n1
3) Freeze the service group by excuting, hagrp -freeze [service_group] command.
4) Manually stop the service group application. Veritas cluster will detect the failure but it won't perform a failover because the service group is frozen
5) Apply patch or configuration change to the application.
6) Manually start the service group application. Veritas cluster will detect the availability of service group in it's next monitoring cycle.
7) Unfreeze the service group: hagrp -unfreeze [service_group]. Even if veritas hasn't detected the availability of app after manual startup, it won't perform a failover as it waits for the application to come up on its own. Eventually it will detect that application is already live and will update its status.