PGA too large
Hi,
We have a RAC 10g with two nodes. Sometimes both of nodes paging or use a lot of swap. When we get AWR and ADDM, we see that SGA+PGA>physical memory available in each node.
The database has sga_tareget=20G and pga_aggregate_target=6G, but PGA used in one node is 14G and in other node is 27G. The physical memory in echa node is 32G.
When this happens, we see 3 or 4 sessions in echa node executing the same SQL. The explain plan of this SQL:
--------------------------------------------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time | Inst |IN-OUT|
We have a RAC 10g with two nodes. Sometimes both of nodes paging or use a lot of swap. When we get AWR and ADDM, we see that SGA+PGA>physical memory available in each node.
The database has sga_tareget=20G and pga_aggregate_target=6G, but PGA used in one node is 14G and in other node is 27G. The physical memory in echa node is 32G.
When this happens, we see 3 or 4 sessions in echa node executing the same SQL. The explain plan of this SQL:
--------------------------------------------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time | Inst |IN-OUT|
0