In a clustered environment with multiple sites, one of the active nodes in the primary site (in general the one that comes first in the list and so the one which is likely the primary of the cluster) shows consistently 90% CPU usage. No specific LDAP update or other client operations are being carried out in the node.
One of the possible causes for this is several custom report generation processes running simultaneously. Such a situation may occur if one has not been able to complete before other kick in, thus preventing each other from finishing and resulting in high CPU usage.
One way to check if this is the case is to ssh into the appliance by using the ssh debug patch provided by Broadcom Support and check if several processes like the following are running
CA PAM, all releases
Kill all /var/www/htdocs/uag/web/opm/opm_cron.pl processes and situation should come back to normal