Questions regarding database host memory management (host started swapping when there were roughly 8
Hi
Oracle 19.6
2 node RAC database with GRID
Oracle Linux 7.8
It is physical server (not a virtual machine).
Three databases running on that node. But only one is critical one, 2 others are quite low on load.
I need some help understanding memory management on these linux hosts.
My problem is that few days ago host started swapping at the time, when there were about 8GB of free memory left on that system.
I have identified that at the time of 9:17, there were executed a certain query, which had 500+ union operators and parsing this single sql statement itself took roughly 2-3gb of PGA memory, and looks like this did not fit on RAM, so it started swapping. Interesting thing is, there were close to 8GB of memory free at the time it started swapping, any idea why is that..?