Last night we had some delayed sessions with rule type "object delayed query". The no. of sessions was around 4 which is very small to provoke TASM throttles to delay those sessions. I wander what other reasons could be for the delay which lasted for more than 3 hours. As this can create huge problems in future.
Regarding the throttles and longer than expected delay, there are a number of things that could have happened to cause this behavior. Hard to say which one without an analysis but some ideas.
1) If there are workload management rules that move active queries to the same grouping / WD as the throttle, this can actually result in more queries running (active queries are not considered for delay due to concerns around locking) than the throttle allows for and then results in more queries needing to complete before the wait queue queries can run.
2) Could be queries with extensive skew executing at the same time
3) Could be due to a burst of similar queries
4) Check the performance of the active queries at that time, they could potentially have been blocked or in distress which made them run longer than expected that impacted those in the queue
5) Could be caused by any out of AMP worker task conditions around that time
What version are you in ? We have some bugs in 13.10 system which causes the queries to the delayed state with no reason :)