We want to run standard vacuum process on our production database which is over 100 GB and have millions of dead tuples.
Can anyone suggest what system parameters we need to keep in mind for setting cost-based vacuum settings? I mean like CPU/IO/Memory/Disk.
We cannot run vacuum full
as the database should be up and running continuously so we just want to attain most appropriate value without affecting the system much.
CLUSTER
andVACUUM FULL
take their locks with a reason.