Database activity stops drops suddenly.
Database OLTP hangs occur intermittently.
All the clients sessions stop responding.
PROMON detects a minimum of activity.
From a transactional standpoint, active transactions are hung.
Clients are blocked on specific record ids.
Some clients are also blocked on TXEC, TXES etc.
Focusing on the blocked clients, the locked buffers show which user / PID has the resource that the others are waiting on. In these cases, the users that have what the others appear to be blocking waiting for, are also hung on a latch possibly for a buffer.
Stack traces against the hung clients all show varying degrees of waiting (latSleep, utnap, nanaosleep, nsleep, latWaitOnSem, latUsrWait, bmblkwait. etc)
New client sessions can login to the database, but their application sessions immediately freeze.
