How to trace memory resize ops

Oracle 11.2.0.4.171130 EE on Windows 64 Bit

 

 

We've been having problems with memory resize operations shrinking the buffer cache

below specified limits:

 

Specified:   16GB db_cache_size, 3GB shared_pool_size, 20GB sga_max_size.

Changes to:   approx. 3.5GB db_cache_size, 16GB shared_pool_size

 

This occurs about 2-3 times per month, usually after hours, severely decreases database

performance and ends with a bounce of the database.

 

 

According to MOS 1269139.1 the resizes are expected behavior.

 

 

Ist there some way to trace / find out what exactly causes this?

(SQL, background process, other details of the root cause)

 

My customer is not willing to set _MEMORY_IMM_MODE_WITHOUT_AUTOSGA=false

and risk ORA-04031. If that changes, could one see something with a trace of

ORA-04031?

 

 

 

Thanks and best regards

Ingrid Voigt