The Session Manager shows as out of sync. After refreshing the session manager status it stayed the same. After restarting the cluster it went out of sync again within a few minutes.
Analysis of system logs showed that the table going out of sync was g_u in uag. Engineering determined that the problem was caused by devices having too many tags assigned to the device, as many as 75 per device. The true limitation was the size of a MySQL query, which concatenated all the tagIDs. The problem was resolved by changing the configuration to not use so many tags.