I've encountered a bug at several clients that upgraded to Oracle 12c - 184.108.40.206 - that requires the combination of several new adaptive features introduced with Oracle 12c.It needs an execution plan that makes use of adaptive join methods, plus at runtime the activation of the new "statistics feedback for joins" feature that was also introduced with Oracle 12c. Note that in 11.2 there was already the "cardinality feedback" feature that only applies to single table cardinality misestimates, but not to join cardinality misestimates.In case then the join method used at runtime is a Nested Loop join - not necessarily the join method preferred initially, so a runtime switch from Hash to Nested Loop join also reproduces the problem - the "statistics feedback for joins" feature generates a bad OPT_ESTIMATE hint for the join cardinality that always seems to be one, like the following: OPT_ESTIMATE(...
For the next couple of weeks I'll be picking up various random notes I've made during the sessions that I've attended at OOW. This particular topic was also a problem discussed recently at one of my clients, so it's certainly worth to be published here.
In one of the optimizer related sessions it was mentioned that for highly volatile data - for example often found in Global Temporary Tables (GTT) - it's recommended to use Dynamic Sampling rather than attempting to gather statistics. In particular for GTTs gathering statistics is problematic because the statistics are used globally and shared across all sessions. But GTTs could have a completely different data volume and distribution per session so sharing the statistics doesn't make sense in such scenarios.
So using Dynamic Sampling sounds like a reasonable advice and it probably is in many such cases.