2017-11-24 17:55 GMT+01:00 Tom Lane <t...@sss.pgh.pa.us>: > bricklen <brick...@gmail.com> writes: > > > If you are on a (non-production) *nix server you can use: > > sync && echo 3 > /proc/sys/vm/drop_caches > > You would also need to restart the postmaster, to get rid of whatever > is in Postgres' shared buffers. >
And restart postgres first, otherwise the shutdown checkpoint will repopulate some of the buffers you just dropped via drop_caches. On Fri, Nov 24, 2017 at 9:09 AM, hmidi slim <hmidi.sl...@gmail.com> wrote: > I execute the command sync && echo 3 > /proc/sys/vm/drop_caches but I > still got buffers hit added to that buffers read now.Why I got the buffers > hit? > Please don't top post in this mailling list. If the same buffer is accessed repeatedly in a query, then some of those accesses will be from the cache even if it were completely cold to start with. Cheers, Jeff
