Shared buffers seem too low to me, usually you go for 20 - 25% of RAM for a dedicated DB server. work_mem also seems low, but that depends on your actual queries. You should at least log the usage of temp files and maybe look into pgbadger to analyze your logs. Max_connections is way too high - espcially when you use a pooler. I get along using ca. 100 connections serving several hundreds of users using jdbc pooling. You might want to take a look at pgtune. Von meinem iPad gesendet > Am 09.06.2015 um 10:07 schrieb AL-Temimi, Muthana <muthana.al-temimi@xxxxxxxxxxxxxxxxxxxxx>: > > Hello Jan, > > The shared_buffers ist 1024MB of the postgresql database and the kernel.shmmax=2147483648 (2GB) of linux OS. > > And here is the some postgresql configurations: > > work_mem=4MB > max_connections=2800 > shared_buffers=1024MB > > and the configuration of pgpool > init_childern=350 > max_pool=4 > > Regards > Muthana > > -----Ursprüngliche Nachricht----- > Von: pgsql-admin-owner@xxxxxxxxxxxxxx [mailto:pgsql-admin-owner@xxxxxxxxxxxxxx] Im Auftrag von Jan Lentfer > Gesendet: Montag, 8. Juni 2015 16:23 > An: pgsql-admin@xxxxxxxxxxxxxx > Betreff: Re: cache Memory of server > > Am 2015-06-08 15:53, schrieb AL-Temimi, Muthana: >> See the free command: >> >> am 08.06.2015 um 15:13 Uhr: --active connection: 305 >> >> srvpgsql1:/opt/pgsql_data # free >> >> total used free shared buffers cached >> >> Mem: 12199684 8758400 3441284 1269784 231324 7139400 >> >> -/+ buffers/cache: 1387676 10812008 >> >> Swap: 6289404 0 6289404 >> >> >> ---------------------------------------------------------------------- >> >> >> am 08.06.2015 um 15:53 Uhr: --active connection: 278 >> >> srvpgsql1:/opt/pgsql_data # free >> >> total used free shared buffers cached >> >> Mem: 12199684 8686228 3513456 1269784 232164 7164288 >> >> -/+ buffers/cache: 1289776 10909908 >> >> Swap: 6289404 0 6289404 >> > > > That is basically what Scott said: you are watching the Kernel FS cache. It may only be a coincidence that it increased together with the postgres sessions. A high number here is usually somehting good, because a lot of your filesystems reads will be served from RAM. Looking at your numbers, I would say you are "all good" (except as Scott said, mabye try to reduce number of parallel sessions) - big fs cache and still free RAM. > What are your settings for shared buffers btw? > > Jan > > > -- > Sent via pgsql-admin mailing list (pgsql-admin@xxxxxxxxxxxxxx) To make changes to your subscription: > http://www.postgresql.org/mailpref/pgsql-admin -- Sent via pgsql-admin mailing list (pgsql-admin@xxxxxxxxxxxxxx) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-admin
