When there are a reasonable number of nodes on a Solaris server started via svcadm, if it is unable to start up in 300 seconds it is automatically killed. A larger value is required for the start_timeout in /opt/opennms/contrib/smf-manifest.xml e.g.
<exec_method
type='method'
name='start'
exec='/opt/opennms/contrib/svc-opennms start'
timeout_seconds='600'>
<method_context>
<method_credential user='opennms' group='opennms' privileges='basic,net_icmpaccess,net_privaddr' />
</method_context>
</exec_method>

If the installed value is insufficient then the workround is:

svccfg -s opennms
setprop start/timeout_seconds = nnn
listprop
exit
svcadm refresh opennms